ArticlePDF Available

Improved Near‐Surface Continental Climate in IPSL‐CM6A‐LR by Combined Evolutions of Atmospheric and Land Surface Physics

Authors:

Abstract and Figures

This work is motivated by the identification of the land‐atmosphere interactions as one of the key sources of uncertainty in climate change simulations. It documents new developments in related processes, namely boundary layer/convection/clouds parameterizations and land surface parameterization in the Earth System Model of the Institut Pierre Simon Laplace (IPSL). Simulations forced by Sea Surface Temperature are produced with different combinations of atmospheric and land surface parameterizations. They are used to explore the sensitivity to the atmospheric physics and/or soil physics of - major biases in the near surface variables over continents, - the energy and moisture coupling established at the soil/atmosphere interface in not too wet (energy limited) and not too dry (moisture limited) soil moisture (SM) regions also known as transition or "hot‐spot" regions - the river runoff at the outlet of major rivers. The package implemented in the IPSL‐Climate Model (CM) for the phase 6 of the Coupled Models Intercomparison Project (CMIP6) allows to reduce several biases in the surface albedo, the snow cover and the continental surface air temperature in summer as well as in the temperature profile in the surface layer of the polar regions. The interactions between soil moisture and atmosphere in hot‐spot regions are in better agreement with the observations. Rainfall is also significantly improved in volume and seasonality in several major river basins leading to an overall improvement in river discharge. However, the lack of consideration of floodplains and human influences in the model, e.g. dams, irrigation, impacts the realism of simulated discharge.
Content may be subject to copyright.
Improved NearSurface Continental Climate in IPSL
CM6ALR by Combined Evolutions of Atmospheric
and Land Surface Physics
Frédérique Cheruy
1
, Agnès Ducharne
2
, Frédéric Hourdin
1
, Ionela Musat
1
,
Étienne Vignon
3
, Guillaume Gastineau
4
, Vladislav Bastrikov
5
, Nicolas Vuichard
5
,
Binta Diallo
1
, JeanLouis Dufresne
1
, Josene Ghattas
6
, JeanYves Grandpeix
1
,
Abderrahmane Idelkadi
1
, Lidia Mellul
1
, Fabienne Maignan
5
, Martin Ménégoz
7
,
Catherine Ottlé
5
, Philippe Peylin
5
, Jérôme Servonnat
5
, Fuxing Wang
1
, and
Yanfeng Zhao
1
1
Laboratoire de Météorologie Dynamique (LMD)/IPSL/Sorbonne Université/CNRS, UMR 8539, Paris, France,
2
Milieux
environnementaux, tranferts et interaction dans les hydrosystemes et les sols (Metis)/Sorbonne Université/IPSL/CNRS/
EPHE, Paris, France,
3
Environmental Remote Sensing Laboratory (LTE), École Polytechnique Fédérale de Lausanne
(EPFL), Lausanne, Switzerland,
4
Laboratoire dOcéanographie et du Climat: Expérimentation et Approches Numériques
(Locean)/Sorbonne Université/IPSL/CNRS, Paris, France,
5
Laboratoire des Sciences du Climat et de l'Environnement,
IPSL, unité mixte CEACNRSUVSQ, Gif sur Yvette, France,
6
Institut Pierre Simon Laplace (IPSL), CNRS, Paris, France,
7
Institut des Géosciences de l'Environnement (IGE), Univ. Grenoble Alpes, CNRS, 38000, Grenoble, France
Abstract This work is motivated by the identication of the landatmosphere interactions as one of the
key sources of uncertainty in climate change simulations. It documents new developments in related
processes, namely, boundary layer/convection/clouds parameterizations and land surface parameterization
in the Earth System Model of the Institut Pierre Simon Laplace (IPSL). Simulations forced by prescribed
oceanic conditions are produced with different combinations of atmospheric and land surface
parameterizations. They are used to explore the sensitivity to the atmospheric physics and/or soil physics of
major biases in the near surface variables over continents,
the energy and moisture coupling established at the soil/atmosphere interface in not too wet (energy
limited) and not too dry (moisture limited) soil moisture regions also known as transition or hotspot
regions,
the river runoff at the outlet of major rivers.
The package implemented in the IPSLClimate Model for the Phase 6 of the Coupled Models
Intercomparison Project (CMIP6) allows us to reduce several biases in the surface albedo, the snow cover,
and the continental surface air temperature in summer as well as in the temperature prole in the surface
layer of the polar regions. The interactions between soil moisture and atmosphere in hotspot regions are in
better agreement with the observations. Rainfall is also signicantly improved in volume and seasonality in
several major river basins leading to an overall improvement in river discharge. However, the lack of
consideration of oodplains and human inuences in the model, for example, dams and irrigation, impacts
the realism of simulated discharge.
Plain Language Summary Land surfaceatmosphere interactions play an essential role in the
climate system. They strongly modulate the regional climates and have impacts on the global scale for
instance through freshwater release into the oceans. Climate hazards (heat waves, droughts) and their
impacts on populations also strongly depend on interactions between land and atmosphere and on their
evolution with climate change. Climate models are precious tools to investigate how the Earth climate
behaves. The sixth phase of the Climate Model Intercomparison Project (CMIP6) provides important tools to
measure the progress and address the remaining open questions regarding the continental climate modeling.
The representation of the landatmosphere coupled system by the IPSLClimate Model involved in CMIP6
is thoroughly evaluated against observations and compared with simulations using the CMIP5 version.
Several biases concerning the temperature over land and over the ice sheets and with the snow cover are
signicantly reduced. Numerous improvements were made developping advanced parameterizations and
tuning of the radiation and of the turbulent mixing in the atmospheric model. The realism of the seasonal
©2020. The Authors.
This is an open access article under the
terms of the Creative Commons
Attribution License, which permits use,
distribution and reproduction in any
medium, provided the original work is
properly cited.
RESEARCH ARTICLE
10.1029/2019MS002005
Special Section:
The IPSL Climate Model Used
in CMIP6**
Key Points:
The representation of the
landatmosphere coupled system by
the IPSL model is thoroughly
evaluated
Improvements with respect to
previous versions are documented in
the context of the Coupled Model
Intercomparison Project, CMIP
Advanced parameterization of land
and atmospheric processes, tuning
of the radiation, and the turbulent
mixing yielded many improvements
Supporting Information:
Supporting Information S1
Correspondence to:
F. Cheruy,
frederique.cheruy@lmd.jussieu.fr
Citation:
Cheruy, F., Ducharne, A., Hourdin, F.,
Musat, I., Vignon, É., Gastineau, G.,
et al. (2020). Improved nearsurface
continental climate in IPSLCM6ALR
by combined evolutions of atmospheric
and land surface physics. Journal of
Advances in Modeling Earth Systems,
12, e2019MS002005. https://doi.org/
10.1029/2019MS002005
Received 30 DEC 2019
Accepted 17 JUL 2020
Accepted article online 30 JUL 2020
CHERUY ET AL. 1of33
cycle of hydrological variables such as the precipitation or the river discharge is also improved over many
regions. The new treatment of hydrology paves the way for future developments on water resource aspects in
the climate model.
1. Introduction
Earth's climate and its evolution are determined by interactions between the ocean, the atmosphere, ice
caps, and land surfaces under the external solar forcing and the atmospheric composition. For these reasons,
numerical models need to couple all these components of the system when they are used for running climate
projections to anticipate the impacts of climate change. In this general framework, the land
surfaceatmosphere interactions strongly modulate the regional climate (e.g., Seneviratne et al., 2010); they
particularly control climate hazards, and their consequences (Jaeger & Seneviratne, 2011; Miralles
et al., 2014) impact the freshwater discharge into the oceans and, in turn, the thermohaline circulation
(Peterson et al., 2002). They rely on complex overlap of multiple landatmosphere feedback processes and
depend on the representation of the interactions between the soil moisture and the boundary layer through
the partition of the available energy at the surface in sensible and latent heat, the impact on radiation (Betts
et al., 1996; Eltahir, 1998; Schär et al., 1999), the representation of the convection and its sensitivity to sub-
grid scale heterogeneities (Guillod et al., 2015; Taylor et al., 2011, 2012), the representation of soil moisture,
and the possible interplay with the atmospheric circulation (Boé, 2013; Hohenegger & Stevens, 2018). The
complexity and the variety of processes involved make the landatmosphere interactions one of the key
sources of uncertainty in climate change simulations at regional scale. As an example, analyses of CMIP5
models revealed considerable spread in the ability of models to reproduce observed correlation between pre-
cipitation and soil moisture in the tropics (Williams et al., 2012).
The various phases of the Climate Model Intercomparison Project (CMIP) give important milestones to mea-
sure the progress and the remaining open questions concerning the climate modeling and in particular the
parameterization of the land surfaceatmosphere interactions. Between Phases 5 and 6 of CMIP, signicant
efforts have been devoted to improving the atmospheric (Hourdin et al., 2020), the land surface, and hydro-
logical components of the Earth System Model of Institut Pierre Simon Laplace (IPSL) and to tuning the
Climate Model (CM). When the fully coupled model is used, compensating errors can hide the role played
by the subgrid scale processes that regulate a large part of the exchanges of energy, water, and matter
between the surface and the free atmosphere or constrain the related parameterizations to work in unrealis-
tic conditions (e.g., Diallo et al., 2017; Roehrig et al., 2013). On the contrary, working with individual com-
ponents impedes the activation of important couplings and feedbacks. Congurations with prescribed sea
surface temperature (SST) and sea ice concentration (SIC) allow us to overcome these difculties. These con-
gurations are referred to as Atmospheric Model Intercomparison Project (AMIP) congurations. Together
with an intermediate conguration, such as a nudged conguration in which the largescale dynamics (i.e.,
the zonal and meridional wind components) is nudged towards reanalysis (Cheruy et al., 2013; Coindreau
et al., 2007; Diallo et al., 2017), these AMIPlike congurations are used here to assess how realistic the con-
tinental surfaceatmosphere interactions simulated by the IPSLCM are and to help interpret the fully
coupled simulations produced with the atmospheric (LMDZ) and land surface (ORCHIDEE) components
of the IPSLCM.
The focus of the present analysis is put on the processes which control the energy and moisture exchange at
the surface. Main features of the near surface climate over continents in the historical simulations done with
the full IPSLCM are documented in a companion paper (Boucher et al., 2020), where the biogeochemical
aspects of the land surfaceatmosphere coupling are considered.
In section 2, the major changes of LMDZ, ORCHIDEE, and their interface are summarized, and the simula-
tions used for the analyses are described. The evolution of the main biases in nearsurface variables since
CMIP5 is analyzed in the third section, and sensitivity studies are used to identify the source of these biases.
The role of the parameterizations and of the adjustment or tuning (Hourdin et al., 2017) is discussed. In the
fourth section, the impact of the modied parameterizations on essential variables of the coupling (radia-
tion, evaporation, precipitation, and surface soil moisture) is discussed for hotspot regions (Koster et al., 2004)
such as Central North America and a region in the Sahel where the land surface coupling is strong but
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 2of33
largely modeldependent (Boé & Terray, 2008; Hohenegger et al., 2009). The fourth section also deals with
river discharge and its response to precipitation. It is a central target for a CM for several reasons: One of
them is that the freshwater discharge into the Arctic Ocean from the boreal rivers affects the global climate
system by impacting the thermohaline circulation (Peterson et al., 2002). It is also a valuable source of infor-
mation for utilization of global water resources and prevention of oods and drought which can both
increase the risk for populations in the context of climate change (Arnell & Gosling, 2013; Schewe
et al., 2014). For some basins, it is possible to compare the results with observations which provides an
assessment of the hydrological cycle over major watersheds. In the last section, the results are summarized,
and directions for further improvements are presented.
2. Model Physical Content and Setup of the Simulations
2.1. The Atmospheric Model
LMDZ is the atmospheric General Circulation Model (GCM) that has been developed for about 30 years at
the Laboratoire de Météorologie Dynamique (LMD). The versions of LMDZ used for Phases 5 and 6 (here-
after called 6A) of CMIP together with the conguration adopted are described in Hourdin et al. (2006)
and Hourdin et al. (2020). The main contribution of IPSL to CMIP5 was done with a package hereafter called
AP and referred to as IPSLCM5Ain the CMIP5 database. Version 6A is an improved version of the new
physicspackage, a preliminary version of which has been tested for CMIP5 (Hourdin et al., 2020) and is
referred to as IPSLCM5BLRin the CMIP5 database. The changes from the AP to the new physicsver-
sion are linked to a complete rethinking of the parameterizations of turbulence, convection, and clouds and
are described in Hourdin et al. (2013). The main model modications between the new physicsand 6A are
the revision of the eddy diffusion Yamada (1983) 1.5 order turbulent scheme already implemented in the
new physics, the introduction of a stochastic triggering designed to make the frequency of occurrence of
new convective systems within a mesh aware of the grid cell size (Rochetin, Couvreux, et al., 2014;
Rochetin, Grandpeix, et al., 2014), a modication of the thermal plume model for the representation of stra-
tocumulus clouds (Hourdin et al., 2019), the introduction of the latent heat release associated with water
freezing (not accounted for so far), and a new parameterization of non orographic gravity waves targeting
the representation of the quasibiennial oscillation (QBO). These changes were accompanied by a signicant
renement of the vertical grid, both for the QBO issue and for a better representation of boundary layer
clouds. The radiative codes in LMDZ are inherited from the ECMWF weather forecast model. In version
AP a wide bandspectral model was used both in the thermal infrared and in the shortwave (SW) spectrum
(Morcrette, 1991). In version 6A, the infrared part was replaced by the RRTM code (Mlawer et al., 1997),
based on a kcorrelated scheme with 16 spectral bands. For the SW radiation the number of spectral intervals
increased from 2 to 6 in order to better distinguish near infrared, visible, and ultraviolet radiation.
For the setting of the 6A version, particular attention was paid to the very stable boundary layers that occur
over the ice sheet plateaus, sea ice, and boreal lands. Such boundary layers can experience very weak and
intermittent turbulence even close to the ground surface, pushing the current stateoftheart subgrid mixing
parameterizations and underlying physical assumptions to their limits and even beyond. Together with the
renement of the vertical grid, the computation of the eddy diffusion in the Yamada (1983) scheme was
revised. Minimum threshold values of the mixing length and of the stability functions of the eddy diffusion
coefcient have been signicantly decreased (Table 2) to allow for a cutoff of turbulence at a few meters
above the surface in the very stable conditions encountered over the Antarctic Plateau and to obtain more
realistic sharp vertical gradients in very stable atmospheric boundary layers (ABLs) (Vignon et al., 2017).
Such threshold values are often set in operational numerical models to compensate for the nonrepresenta-
tion of subgrid mixing processes and to prevent excessive nearsurface cooling over land in winter (e.g.,
Sandu et al., 2013). The sensitivity of the continental temperature at seasonal and diurnal scale to the values
of the thresholds will be discussed in sections 3.2 and 3.4. Moreover, a new numerical treatment of the
Turbulent Kinetic Energy (TKE) equation in the new scheme prevents an articial cutoff of the turbulence
at standard time step values that could occur in previous model versions even at moderate stability (Hourdin
et al., 2020; Vignon, 2017).
The need to remove thresholds in the turbulence scheme to properly model the stable ABL over the
Antarctic Plateau also raises the need to parameterize more explicitly the additional sources of mixing in
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 3of33
other regions of the globe such as orographyinduced small scales gravitywave drag (Steeneveld et al., 2008)
or the drag induced by vertical obstacles penetrating the boundary layers such as trees (Masson &
Seity, 2009; Nepf, 1999). Due to the renement of the vertical grid of the model several layers can intersect
high vegetation. The loss of largescale kinetic energy due to these drags is converted into TKE. The evalua-
tion of the orographyinduced gravitywave drag is based on the scheme developed by Lott (1999), while the
drag due to high vegetation is set proportional to the vegetation fraction, which penetrates the boundary
layer. The two subgrid scale mixing processes generate TKE, which is accounted for in the prognostic equa-
tion (see Appendix A for details). The impact of these new developments on nearsurface atmospheric vari-
ables are illustrated in section 3.4.
2.2. The Land Surface Model
The land surface is described by the ORCHIDEE model v2.0. The ORCHIDEE model v2.0 computes primar-
ily the uxes of energy, water, and carbon that are exchanged between the different soil and plant reservoirs
and the exchange of these uxes with the atmosphere. In addition, it computes the stocks of water and car-
bon in the different soil and plant reservoirs and the energy stored in the different soil and snow layers.
Model state variables are prognostic, including the Leaf Area Index (LAI), as they are updated at each time
step after the calculation of the uxes between all reservoirs. The module computing dynamically the LAI,
the vegetation albedo, and the soil water stress function applied to transpiration is activated for all CMIP
experiments, except for the HighResMIP one (see section 2.4) that uses prescribed values. The vegetation
properties are dened by plant functional types (PFTs), and their fraction within each grid cell is globally
set from land cover maps that were derived specically for the CMIP6 simulations (Lurton et al., 2020).
These maps combine the historical maps from the land use harmonization database (LUH2v2h, Hurtt
et al., 2011) and the maps derived from satellite observations (Bontemps et al., 2015). See https://orchidas.
lsce.ipsl.fr/dev/lccci/, for more information. The water and energy budgets are computed at the same time-
step as the atmospheric physics (Hourdin et al., 2020) using classical soilvegetationatmosphere transfer
(SVAT) parameterizations. The most relevant modication since the version used for CMIP5 is related to
the soil hydrology, the snow scheme, and the background albedo. The 2layer conceptual parameterization
(hereafter referred to as Choi, (Ducoudré et al., 1993) used for CMIP5 is a double bucket model that has an
upper layer with a varying depth that can appear at the surface after a rainfall event to deal with shorttime
processes and disappears after dry spells (Manabe, 1969). Choirefers to the scheme that Choisnel devel-
oped and tested for cultivated area over France. Laval (1988) showed that this model improved the sensible
and latent heat ux computation on the original bucket model when introduced into the LMD Atmospheric
GCM. In the version used for CMIP6, the vertical water transport is described using Richard's equation (De
Rosnay et al., 2002; d'Orgeval et al., 2008) discretized with 11 layers. The layer thickness increases down-
wards and is doubled between each consecutive layer. The soil moisture column is active over 2 m; a free
drainage condition is imposed at the bottom of the reservoir. This scheme hereafter called ctrl, as it is
now the reference version for IPSLCM, is sometimes referred to as the 11layerORCHIDEE scheme.
The potential of improvement of an early version of this scheme coupled with the AP and new physicsver-
sions of LMDZ has been tested in Cheruy et al. (2013) and Campoy et al. (2013). The soil thermodynamics
and in particular the soil thermal properties have been revised by Wang et al. (2016). They have a signicant
impact on the surface temperature and its high frequency variability in all regions except for the moist
regions (Cheruy et al., 2017). The vertical discretization for temperature is now identical to that adopted
for water, with a minimum soil depth increased to 10 m (and even 90 m when the soil freezing is accounted
for) so that the condition of zero ux at the bottom can be checked globally and annually. The soil properties
(hydraulic and thermal) depend on soil moisture and soil texture, with three possible classes (sandy loam,
loam, and clay loam). The dominant soil texture is assigned to each grid cell, based on the 1° soil texture
map of Zobler (1986). The soil heat capacity is parameterized as a function of the heat capacity of the dry soil
and the liquid water prole and when necessary the ice prole. The soil freezing is allowed and diagnosed in
each soil layer following a scheme proposed by Gouttevin et al. (2012), but the latent heat
release/consumption associated with water freezing/thawing is not accounted for. The freezing state of
the soil mainly impacts the computation of soil thermal and hydraulic properties, reducing for instance
the water inltration capacity at soil surface. Wang et al. (2013) replaced the snow scheme of Chalita and
Le Treut (1994) by a threelayer scheme of intermediate complexity largely inspired by that proposed by
Boone and Etchevers (2001). A routing module (Guimberteau et al., 2012; Polcher, 2003) transforms the
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 4of33
total runoff in each subbasin into river discharge through the ocean. This routing scheme relies on a cascade
of linear reservoirs along the river network (stream reservoirs), complemented in each grid cell by two local
reservoirs, to account for the delay between surface runoff and drainage, on the one hand, and overland and
groundwater ow to the stream reservoir, on the other hand. When using Choi, which does not separate
total runoff into surface runoff and drainage, an arbitrary partitioning is imposed, with 5% feeding the fast
reservoir and 95% feeding the slow reservoir (Guimberteau et al., 2014). In the multilayer version of
ORCHIDEE, evaporation from bare soil following a supply and demand pattern that is controlled by the
moisture present in the surface layers of the soil (the four soil layers of the model closest to the surface),
which evaporates at the potential rate if the soil moisture supply meets the demand.
The continental icecovered surfaces (ice sheets and glaciers) are not included in ORCHIDEE, but they are
treated in a specic module within LMDZ. Momentum and heat roughness heights as well as visible and
near infrared albedos are set to constant values representative of snow conditions over the Antarctic
Plateau (Vignon et al., 2018). The heat transfer in the snow and ice is parameterized as a conductive process
with a xed thermal inertia (Hourdin, 1992). The vertical grid is made of 11 vertical levels to represent the
efolding damping of thermal waves with typical periods from 1,800 seconds to 240 years. The value of the
snow thermal inertia was calibrated to obtain realistic surface temperature and diurnal cycle amplitudes
in Antarctica (Vignon et al., 2017).
Le Quéré et al. (2018) have recently used a version of ORCHIDEE (referred to as OrchideeTrunk), which is
simular to the version used for CMIP6 in an intercomparison project focussing on the carbon and water uxes
where 15 other land surface models (LSMs) were involved. The skill scores obtained by ORCHIDEE are
among the highest for most of the variables considered in this study and in particular for evapotranspiration,
LAI, and runoff (see table B2 in Le Quéré et al., 2018), which are directly involved in our study.
2.3. The Coupling with the Surface
In the surface layer the boundary layer model uses MoninObukhov theory and bulk formulations proposed
by Louis et al. (1982) to parameterize turbulent uxes. Several modications were made in the representa-
tion of the surface layer of LMDZ as well. First, and consistently with the changes done in the boundary layer
to allow strong decoupling in stable atmospheres, the socalled longtailstability functions from Louis
et al. (1982) that articially enhance the surface turbulent uxes in stable conditions were replaced by more
realistic shorttailfunctions from King et al. (2001). This was shown to signicantly improve the represen-
tation of surface temperature on the very at ice sheet of the Antarctic plateau.
A second important change is related to the computation of surface roughness height z
0
. At the surface itself,
heat and humidity transfer are dominated by molecular diffusion, which is less efcient than the momen-
tum transfer due to the pressure forces that are related to the geometry of the roughness elements of the sur-
face (Garratt & Hicks, 1973). For these reasons the roughness heights for the momentum are currently much
higher than that of heat or humidity. While a unique value was used in former versions for all the model
state variables, a different value is now used for horizontal momentum z0mand thermodynamical variables
z0hor tracers z0afor all individual type of subsurface (land, sea, seaice, continentalice). For each PFT
ORCHIDEE used a prescribed value for the roughness height for heat and moisture independent of the
development of the vegetation over continents. For a grid point composed of different types of vegetation,
an effective surface roughness is calculated based on the ux conservation over the grid point. This value
was also used for z0min LMDZ. Measurement campaigns often suggest that the roughness height for heat
should be onetenth of that for momentum for homogeneous surfaces and even less for heterogeneous sur-
faces (Malhi, 1996). Some studies propose that over vegetated areas the roughness height can be parameter-
ized as a function of the LAI. This is the case for the model proposed by Massman (1999) and tested by Su
et al. (2001), which has been implemented in ORCHIDEE v2.0. In forced mode, the dynamic roughness
heights computed for each PFT as a function of the LAI help reduce latent heat calculated in winter on tem-
perate sites, in good agreement with multiannual Fluxnet measurements (Figure S1, (https://uxnet.ux-
data.org/data/lathuiledataset/)). Still in forced mode, the dynamic roughness heights impact the river
discharge at the scale of individual watersheds with signicant improvements for the Danube and the
Mississippi watersheds (not shown). The impact of activating the dynamical roughness height in coupled
simulations is limited for the considered space and time scales (see section 3.4), but the option is activated
for all CMIP6 simulations.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 5of33
Another important change in ORCHIDEE is related to the individual albedo of the bare soil and of the 14
vegetated PFT, which is now optimized with respect to MODIS observations. This calculation of the indivi-
dual albedo leads to signicant improvements, especially over deserts and semiarid areas where the albedo
was signicantly underestimated. The improvement is illustrated by a comparison of the time series of the
albedo simulated with the CMIP5 and the CMIP6 versions of the model and measured at the sites in the
Sahel (Figure 1). The simulation is forced to follow the synoptic variability by relaxing the largescale circu-
lation toward meteorological analyses, which allows direct comparison of the time series of the observations
and the simulations (Cheruy et al., 2013; Coindreau et al., 2007).
The nearsurface (i.e., 2 m) temperature (CMIP6 variable tas, which is one of the most analyzed variables in
CMs especially in relation to climate change impacts) is diagnosed through a procedure based on the
MoninObukhov theory (Hess, 1995). The procedure involves both surface and rst modellevel variables.
In situations when turbulence is very weak and the atmospheric layer above the surface is dry but the
Figure 1. Time series of the surface albedo for year 2006 at Bamba, Agoufou, Wankama, and Nalohou stations, from top
to bottom panel, respectively. Local observations (blue) are compared with nudged simulations for CMIP5 (red) and
CMIP6 (black) physics. The simulation is forced to follow the synoptic variability by relaxing the large scale circulation
toward meteorological analyses, which allows direct comparison of the time series; observations were obtained in 2006,
the year of the AMMA Special Observing Period, and are available from the AMMACATCH database
(http://www.ammacatch.org).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 6of33
surface soil moisture is far from the residual moisture, the procedure occasionally fails, leading to
nonphysical values for one timestep. When the procedure fails screen level temperature can reach 450 K
and screen level specic humidity becomes negative (see Appendix B). Since the problem occurs rarely,
and when it does occur it is only during one timestep in the day (very exceptionally it can occur during
two or more timesteps in the day), it was undetected in the nal version used for the production of the
CMIP6. The problem affects the maximum daily nearsurface temperature, the minimum daily
nearsurface relative humidity, and marginally the daily averages of these quantities. It occurs
approximately 1,700 times (respectively, 2,200 times) in a simulated year of the AMIP (respectively,
PreIndustrial Control [piControl], Eyring et al., 2016) experiments, which is very rare compared to the
(365 × 144 × 142 × 96) times that the calculation is performed in one year of simulation. The CMIP6
experiments presented in this paper have not been rerun due to the time constraint imposed by the CMIP
exercise; however, an a posteriori correction method has been developed. All the CMIP6 data that have
been affected by this problem have been either unpublished or corrected a posteriori and republished.
The a posteriori correction method applied to the published data is
given in Appendix B with an estimation of the associated
uncertainties, which is of the order of several tenths of degrees for
the daily values. Due to the low value of the reconstruction errors
for the monthly mean values (30 times less than the reconstruction
errors of the daily values), it was decided not to make a correction
to the monthly values. The great advantage is that the monthly tas
values are absolutely consistent across all CMIP6 experiments,
regardless of whether the daily values have been corrected or not.
According to these investigations we are condent that all the
published values can be used safely for climate analysis.
2.4. Setup of the Simulations
To document the impact of the changes described in the previous sec-
tions, simulations forced by observed SST and SIC are produced by
combining nal versions of atmospheric physics (AP and 6A) and
of the soil hydrology (Choi and ctrl), namely, APChoi (corresponding
to the IPSLCM5A in the CMIP5 database), APctrl, 6AChoi, and
6Actrl (corresponding to the IPSLCM6A in the CMIP6 database)
(Table 1). A monthly mean climatology of SST and SIC calculated
over the years 19782008 is used for the simulations in order to mini-
mize the impact of the interannual variability in the evaluation. The
6Actrl experiment is also compared with the results of the AMIP
experiment for which a 20member ensemble has been produced
and is published in the CMIP6 database. The impact of the new devel-
opments is documented thanks to an additional set of sensitivity
experiments with the 6A physics, where the new features of the
LSM and the ABL are individually tested (Table 2). The horizontal
Table 1
Reference Simulations: APChoi Corresponds to the IPSLCM5A Conguration Used for CMIP5, and 6Actrl Corresponds to
the IPSLCM6A Conguration Used for CMIP6
Experiment Boundary conditions Atmosphere Vertical levels Land surface duration
APChoi Clim AP 39 Choi 20 years
APctrl Clim AP 39 Ctrl 20 years
6AChoi Clim 6A 79 Choi 20 years
6Actrl Clim 6A 79 Ctrl 20 years
AMIP Actual 6A 79 Ctrl 19781998
Note. The boundary conditons used (climatology [clim] or actual values of SST and SeaIce [actual]) are indicated in the
second column.
Table 2
Sensitivity Experiments for Processes Impacting the Interactions Between the
Land Surface and the Atmosphere
AMIPexperiment Description
NoOro Deactivating orography induced small scale
gravity wave drag
NoTree Deactivating vertical obstacle penetrating boundary
layer drag
NoSnowFreez Deactivating soil freezing and replacing the snow
scheme of
Chalita and Le Treut (1994) by Wang et al. (2013)
Noz0Su Deactivating dynamical roughness heights
(Su et al., 2001)
and using prescribed values with z0m=z0h
6Arsol Activating resistance to bare soil evaporation
6Aric No increased mixing in the stable PBL: ric = 0.20
(reference = 0.18) and lmximin = 0
6Aric83lmx Articially increased mixing in the stable PBL:
ric = 0.143, lmixmin = 1m (reference = 0m)
6AL+ Increased lift effect (SSO):gk
lift
= 1.0
6ALDecreased lift effect (SSO):gk
lift
= 0.0
6AD+ Increased block ow drag amplitude by
(SSO):gk
drag
= 1.2
6ADDecreased block ow drag amplitude by
(SSO):gk
drag
= 0.2
Note. The reference atmospheric physics is 6A with 79 vertical levels, and the
reference LSM is ctrl. The last four lines of the table refer to the sensitivity
experiments to the subgrid scale orography (SSO) schemes; the reference
values for 6A are gk
lift
= 0.1 and gk
drag
= 0.6, gk
lift
(respectively, gk
drag
)
correspond to C
d
(respectively, C
l
) in Lott (1999).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 7of33
grid is identical to the published CMIP6LR data LR stands for Low Resolution and corresponds to approx.
2.5
o
× 1.25
o
, namely, 144 × 142 grid points (Hourdin et al., 2020). We also document simulations performed
with a much ner grid of 50 km for the HighResMIP part of the CMIP6 exercise (Haarsma et al., 2016).
Comparison of these low and high resolution versions allows us to distinguish the part of the model bias
linked to the coarse resolution from that more fundamentally related to the model physical content.
Nudged simulations in which the largescale wind elds (zonal and meridional wind components) are
relaxed towards the ERAInterim reanalyzed winds (ERAI, Table 3) with a time constant of 3 hours are
also used and help assess a possible contribution from largescale circulation deciencies to the
continental bias. Based on previous experience, it is known that a time constant of several hours (312) is
short enough to constrain the large scale circulation and long enough for the physical parameterizations
to fully operate (for wind nudging at least). More details on this approach can be found in Diallo
et al. (2017). The rst 3 years of all experiments, corresponding to the spinup time of the hydrological
model, are disregarded in the analysis.
2.5. Reference Data Sets
The sets of global gridded data used as a reference to evaluate the sensitivity experiments are listed in Table
3. They consist of a siteobservations upscaled products for evaporation (Jung et al., 2011), satellitebased
land evaporation, and surface soil moisture derived through data assimilation processes in the Global
Land Evaporation Amsterdam Model (GLEAM) (Martens et al., 2017) and the ESACCI blended active
and passive microwave retrieval of surface soil moisture (Dorigo et al., 2017), CERESEBAF for surface
SW radiation (Kato et al., 2013), and the Global Precipitation Climatology Project (GPCP) monthly product
resulting from an integration of various satellite data sets and a gauge measurements analysis over land
(Adler et al., 2003) for the precipitation. The total column integrated water vapor is evaluated using the rea-
nalysis and extension of the NASA Water Vapor Project (NVAP) data set which comprises a combination of
radiosonde observations, Television and Infrared Operational Satellite (TIROS) Operational Vertical
Sounders (TOVS), and Special Sensor Microwave/Imager (SSM/I) data sets (Vonder Haar et al., 2012).
The river discharges are extracted from the Global Runoff Data Center (GRDC) database (Milliman &
Farnsworth, 2011). The Snow Cover Extent (SCE) is extracted from the output from the Interactive
Multisensor Snow and Ice Mapping System (IMS) at the National Ice Center (NIC) processed at Rutgers
University and included in the NOAA Climate Data Record (CDR) of Northern Hemisphere (NH) Snow
Cover Extent. For the minimum and maximum daily temperature, we used the Climatic Research Unit
(CRU) TimeSeries (TS) version 4.01 of highresolution gridded data (Harris et al., 2014).
The observations cover a period of at least 10 years compatible with the SST and SIC climatology used to
force the model. We suggest that not considering the exact same periods for the simulations and the
Table 3
Global Gridded Data Sets Used as Reference
Variable
(monthly means) Data set Date Reference
Radiation CERESEBAFL3BEd28 20012012 Kato et al. (2013)
Precipitation GPCP longtermmean 19792005 Adler et al. (2003)
Evapotranspiration In situ upscaled 19822011 Jung et al. (2011)
products (over land)
Evapotranspiration GLEAM 20012011 Martens et al. (2017)
Surface soil moisture ESACCI 20012011 Dorigo et al. (2017)
Surface soil moisture GLEAM 20012011 Martens et al. (2017)
Air temperature CRU 20012010 Harris et al. (2014)
(min/max daily)
Air temperature ERAI 19792014 Dee et al. (2011)
Horizontal winds ERAI 19792014 Dee et al. (2011)
Total precipitable NVAPM 19882009 Vonder Haar et al. (2012)
Water
River discharge GRDC 19812010 Milliman and Farnsworth (2011)
Snow cover NOAACDRSCE 20002009 Robinson et al. (2012)
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 8of33
observations only has a minor impact on the results given that the model internal variability is damped due
to the use of a climatological SST and SIC.
In the supplementary material, ERA5 data (https://cds.climate.copernicus.eu/cdsapp#!/home) are used as
a benchmark in addition to ERAInterim data to evaluate the bias in the airtemperature (supporting infor-
mation Figure S2).
3. Surface Energy Budget and Temperature
3.1. Impact of the Revision of the Eddy Diffusion Parameterization
The improvements resulting from the revision of the turbulent scheme between the AP version and the 6A
version of LMDZ are illustrated in Figure 2, showing the mean seasonal cycle of the air temperature for the
rst three atmospheric levels of LMDZ (version AP and 6A) together with the measurements recorded at six
levels on the 45 m height mast at Dome C (75.1S,123.3E), Antarctic Plateau. For version AP (Figure 2a) an
overall wintertime warm bias (up to 10 K) in the surface layer is noticeable. This version was also unable
to properly reproduce the dynamical behavior of the very stable Antarctic boundary layers (Vignon
et al., 2018), and further analysis of the vertical temperature prole in the rst few hundred meters above
Dome C revealed a signicant underestimation of the climatological temperature inversion (not shown).
For version 6A, both the nearsurface temperature and its vertical gradient are in good agreement with
observations from the surface up to the top of the mast (Figure 2). The dynamics of very stable boundary
layers is also much better simulated (Vignon et al., 2018) than in version AP. Figure 3 shows how version
AP and 6A perform in a singlecolumn conguration used to simulate the test case of DIurnal
landatmosphere Coupling Experiment (DICE) (Kansas, latitude 37.65°N, longitude 263.265°E) far from
the ice sheets regions. The simulations cover a period of 3 days and three nights and the last night which
is stable, and cloudfree is well suited to test the boundary layer scheme under stable condition. The hydro-
logical scheme is bypassed by prescribing the ratio βof evaporation to potential evaporation and the surface
thermal inertia to a value adjusted to the DICE case during the full run (AïtMesbah et al., 2015). For night
time the nearsurface temperature inversion is much stronger in 6A than for the AP run (Figure 3a). The sen-
sible heat ux is reduced with the 6A version and closer to the observations than the AP version (Figure 3b),
which produced a too strong vertical mixing.
3.2. Relative Impact of Atmospheric and Land Surface Components on the Biases of
NearSurface Variables
Most of the biases in evaporation, 2 m temperature, SW downward radiation at the surface, surface albedo,
precipitation, and total precipitable water can be analyzed by inspecting zonal mean variables over the con-
tinents (Figure 4). To further comment regional aspects, maps of mean annual, JJA and DJF bias in 2 m tem-
perature are depicted in Figure 5. The corresponding bias maps are displayed in the supplementary
Figure 2. Time series of the nearsurface monthly mean temperature at Dome C (75.1S,123.3E), Antarctic Plateau. Solid
lines show the APctrl simulation (a) and the 6Actrl simulation (b). Dashed and dotted lines refer to the 20112018
observational data set along a 45m meteorological mast (Genthon et al., 2013).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 9of33
information for the SW downward radiation at the surface (Figure S3), evaporation (Figure S4), surface
albedo (Figure S5), precipitation (Figure S6), and total precipitable water (Figure S7). The maps
corresponding to the rst member of the AMIP ensemble are also plotted, in order to conrm the
representativity of the 6Actrl experiment with respect to the AMIP published data. We veried that the
features discussed hereafter are shared by the other members of the AMIP ensemble. The statistical
signicance of the changes caused by the new land surface and atmospheric physics is assessed
geographically for each studied variable in Figures 6 (6Actrl6AChoi) and S8 (6ActrlAPctrl), with very
similar results to the differences maps for APctrlAPChoi and 6AChoiAPChoi (not shown).
A clear improvement of the CMIP6 reference conguration 6Actrl is related to the radiation budget. This
improvement is illustrated by the reduction of the bias in the downward SW radiation at the surface in
6Actrl (Figures 4d and 4j) and can be attributed to the improvement of the representation of the Cloud
Radiative Effect (CRE) coming from the modication in the parameterizations and to the improved tuning
of the model free parameters targeting the CRE (Hourdin et al., 2020). Consistently with the overall reduc-
tion in the SW radiation bias (Figure S3) and in the evaporation bias especially over the continental United
States (Figure S4) the strong warm bias over the midlatitudes in summer (Figure 5) that was shared by many
models participating in CMIP5 (Cheruy et al., 2014) is reduced in the 6Actrl conguration used for CMIP6.
Over the continental United States, AlYaari et al. (2019) also showed that the general agreement between
areas of strong warm bias and areas of strong precipitation and soil moisture decits is good. In the 6Actrl
conguration the precipitation decit is also signicantly reduced (Figure S9) and the surface soil moisture
is in better agreement with the observations (see section 4.1). In connection with the developments on the
vertical diffusion scheme, the warm bias that extended over a large part of the polar and boreal regions in
winter is reduced or even replaced by a cold bias over part of the Arctic continent and Ocean, Greenland,
and Antartica (Figure 5). The cold bias is probably overestimated over Greenland, the Artic Ocean, and
Antartica due to a warm bias diagnosed in ERAI (Jakobson et al., 2012; Reeves Eyre & Zeng, 2017;
Vignon et al., 2018). When using ERA5, as reference data set instead of ERAI, the bias over Greenland is
Figure 3. Temperature prole (a) and sensible heat ux (b) for the third night (stable) of the DICE case simulated with
the AP (red) and 6A (black) version of the atmospheric physics. The observed sensible heat ux is plotted together with
the simulations (blue).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 10 of 33
Figure 4. Zonal mean bias in DecemberJanuaryFebruary (DJF) and JuneJulyAugust (JJA) over continents for the
evaporation (a, g), the precipitation (b, h), the air temperature (c, i), the downward shortwave (SW) radiation (d, j),
the surface albedo (e, k), the total precipitable water (f, l) in 6Actrl (thick black curve), 6AChoi (dashed black curve),
APctrl (thick red curve), and APChoi (dashed red curve). For precipitation the blue curve corresponds to the absolute
value of the observations (yaxis on the right side). The references are described in Table 3.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 11 of 33
reduced in DJF (not shown). Over the Arctic continent Lindsay et al. (2014) report that ERAI has a bias of
less than 0.5 K compared to the observations.
As a result of the new snow scheme and of the optimization with respect to the MODIS observations, the
surface albedo in the ctrl model is improved in most regions in winter (Figures 4e and S5) and over deserts
(notably the Sahara) over the year. The new snow scheme improves the snow cover which was signicantly
underestimated with Choi (Figure 7). With the exception of the surface albedo and to a lesser extent the eva-
poration, the overall structure of the bias is only marginally sensitive to the land surface scheme whose
impact is mostly relevant at the regional scale (Figures 5, 6, S3, S4, S5, S6, and S7).
When considering the continents globally, Choi and ctrl both overestimate the evaporation (especially in
winter) regardless of the atmospheric model with which it is coupled. This overestimation is slightly less
Figure 5. Mean multiannual bias in 2 m temperature (tas) for DecemberJanuaryFebruary (rst column) and for JuneJulyAugust (middle column ) and for the
full year (last column) in the APChoi (rst row), APctrl (second row), 6AChoi (third row), 6Actrl (fourth row), and AMIP congurations (fth row). The reference
is given by ERAI reanalysis averaged for the 19792014 period.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 12 of 33
Figure 6. Signicance of the multiannual differences between congurations 6Actrl and 6Achoi, with gray coloring where the difference is not statistically
signicant based on Student's ttest (with a pvalue < 0.05). The rst ve maps show the yearly mean differences for the air temperature, evaporation,
precipitation, downward SW radiation at the surface, surface albedo, and the bottom right map shows the winter (DJF) difference in surface albedo. The displayed
means and standard deviations are calculated over the whole globe including non signicant points.
Figure 7. Mean annual bias in the snow fraction for the APChoi (left) and 6Actrl (right) congurations. The reference is taken from the NOAA Climate Data
Record Snow Cover Extent diagnosed from the Interactive Multisensor Snow and Ice Mapping System at the National Ice Center.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 13 of 33
for ctrl (Table 4), albeit this result is modulated at regional scale
(Figure 6). Investigating the minimum and maximum daily temperature
shows a widespread warm bias of daily minimum temperature over the
midlatitude (Figure 8). This bias is present over the whole year for the
AP physical package used for CMIP5, and only in JJA for the 6A package
used for CMIP6, it is very marginally sensitive to the land surface scheme.
This is consistent with the reduction of the turbulent mixing in the PBL
for the stable boundary layers obtained with the 6A atmospheric physics
and with the results of Wei et al. (2017), which suggested that a bias in
the simulated PBL mixing could very likely contribute to the temperature
bias common to most of the models that participated to CMIP5 with AMIP
experiments. The moist atmospheric bias over the midlatitudes in JJA
(Figure 4l) could contribute to the warm bias of daily minimum tempera-
ture by minimizing the nocturnal radiative cooling, but further investiga-
tion is needed to explain this bias, which is shared by other models
participating in CMIP6.
3.3. Atmospheric Process Sensitivity to the LSM Choice
The above analysis has shown that, for most variables and skills consid-
ered, the changes due to the atmospheric physics are larger and more
broadly signicant than the ones dues to the land surface physics, as con-
rmed by the comparison between Figures S8 and 6, respectively. We
detail here the sensitivity of the 6A atmospheric physics to the LSM choice (Figure 6). The differences
between 6AChoi and 6Actrl are statistically signicant (at the 5% level) over most of the continents for all
variables but precipitation. The seasonal differences (not show but for winter surface albedo) are very con-
sistent with the yearly differences. Ctrl induces a signicant cooling over an extended region going from
Siberia to China (up to 3 K locally in absolute value). This cooling is accompanied with an increase in eva-
potranspiration, some local reductions of the downward SW radiation, but also a widespread decrease of sur-
face albedo (mostly driven by the summer season thus by vegetation), which is probably overruled by the
large increase in albedo in winter. In contrast, large land areas of the southern hemisphere exhibit a signi-
cant warming from Choi to ctrl, along with an evaporation decrease, a decrease in surface albedo, and a
downward SW radiation increase. Two exceptions can be isolated to the cooling/warming response to eva-
poration increase/decrease. The rst one is the Sahara, where air temperature is reduced with ctrl, despite
signicant reduction of evaporation and increase of incoming SW radiation: The reason is the substantial
albedo increase in this area, like in most sparsely vegetated zones. The second exception comprises the
humid equatorial areas (intertropical convergence zone), where surface air temperature decreases without
any signicant evaporation change: There, the main driver seems to be the reduction of incoming surface
radiation, likely related to precipitation increases, although these changes are rarely signicant, and mostly
in JJA. Precipitation is also signicantly impacted by the choice of the LSM over monsoon regions, like
Western Africa in JJA and Southern Amazonia in DJF, where ctrl tends to reduce evaporation and precipita-
tion. The few spots over tropical oceans where the change in precipitation and evaporation are signicant
are probably due to slight modications of the circulation in response for instance to the temperature
changes. However, the amplitude of the changes is very low with respect to the typical oceanic values in
those regions. When considering the continents globally, Choi and ctrl both overestimate the evaporation
(especially in winter). This overestimation is slightly less for ctrl (Table 4), albeit this result is modulated
at regional scale (Figure S4).
3.4. Sensitivity Experiments
In order to further interpret the above results we use the sensitivity experiments described in Table 2.
Sensitivity simulations to the strength of the decoupling in stable condition were performed by changing the
values of the minimal mixing length (lmixmin) and of the critical Richardson number (ric) above which the
stability functions of the turbulent diffusion coefcient reach their lowerbound value (see Figure 2 in
Vignon et al., 2017 for details). Those two thresholds enhance the mixing and prevent the turbulence cutoff
in very stable conditions (Table 2, 6Aric, 6Aric83lmx). Figure 9 shows the impact of the sensitivity
Table 4
Mean Continental Biases in DJF and JJA for 2 m Temperature,
Evaporation, Surface Downward Radiation, Albedo, and for the Reference
Simulations (APChoi, APctrl, 6AChoi, 6Actrl)
2 m T Evaporation Precipitation
sfc SW
(down) albedo
Experiment (K) (mm/day) (mm/day) (W/m2) (×100)
DJF
APChoi 0.19 0.33 0.41 16.4 0.83
APctrl 0.39 0.30 0.47 15.2 0.72
6AChoi 1.40 0.36 0.57 1.07 0.33
6Actrl 1.47 0.34 0.58 0.63 0.29
Obs 279.4 0.99 1.82 155.4 17.2
JJA
APChoi 0.31 0.18 0.03 36.5 1.73
APctrl 0.05 0.11 0.03 35.8 0.19
6AChoi 0.15 0.36 0.42 14.2 0.89
6Actrl 0.37 0.28 0.37 16.27 0.01
Obs 293.8 1.75 2.06 227.5 18.
Note. APChoi corresponds to the IPSLCM5A conguration used for
CMIP5, and 6Actrl corresponds to the IPSLCM6A conguration used
for CMIP6. The last line corresponds to the mean value over continents
calculated with the observation described in Table 3.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 14 of 33
experiments on the zonal means in JJA and DJF. North of 60°N, the nearsurface temperature is highly
sensitive to these thresholds. Allowing less decoupling (Table 2, 6Ari83lmix) signicantly reduces the cold
bias over continental areas in winter, but it deteriorates the vertical temperature gradient over the
Antartic Plateau shown in Figure 2. A further increase of the decoupling with respect to the conguration
adopted for CMIP6 (Table 2, 6Aric) leads to a reduction of the wintertime minimal temperature but does
not impact the nighttime bias in summer for the 6A version (not shown).
As expected, the orographyinduced TKE production (Table 2, NoOro, also see Appendix A) tends to warm
the midlatitude and boreallatitude in winter which partially counterbalances the effect of the reduced ver-
tical diffusion for the stable boundary layers (not shown). The impact of deactivating the drag induced by the
vegetation penetrating the boundary layer (Table 2, NoTree) is negligible for the nearsurface temperature
(not shown).
A sensitivity simulation focusing on the evaporation for the ctrl model was also designed to target the bare
soil evaporation, which can reach the potential rate when the moisture in the rst four layers of the soil is
higher than the residual moisture (Table 2, 6Arsol). It is likely that the potential rate of evaporation leads
to an overestimation of evaporation when patches of soil begin to dry out in the grid cell. To overcome this
defect a resistance to bare soil evaporation can be added to the aerodynamic resistance. This approach has
been implemented in ORCHIDEE using the formulation proposed by Sellers et al. (1986). The activation
of this option reduces the evaporation (Figures 9a and 9g). However, amplifying the SW radiation bias at
the surface over the midlatitude north (Figures 9d and 9j) and reducing the evaporative cooling results in
Figure 8. Zonal mean bias over continents in the seasonal means of minimum and maximum daily temperature for the
reference congurations (thick black curve: 6Actrl, dashed black curve: 6AChoi, thick red curve: APctrl, dashed red
curve: APChoi).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 15 of 33
Figure 9. Zonal mean bias in DJF and JJA over the continents for the evaporation (a, g), the precipitation in percent and
the absolute value of the observation (b, h), the air temperature (c, i), the downward SW radiation at the surface(d, j),
the surface albedo (k, e), the total precipitable water (f, l) in the reference experiment 6Actrl (thick black curve), in
sensitivity experiments for the turbulent mixing 6Aric (thick blue curve), in 6Aric83lmx (blue thin curve), and in
sensitivity experiment with a resistance to bare soil evaporation activated (thick pink curve).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 16 of 33
a strong warm bias (Figure 9c and 9i). The impact of deactivating the dynamical roughness height (Table 2,
noz0Su) is detected at the regional scale on the evaporation and the temperature, but it is quite limited for
the considered space and time scales (not shown).
3.5. Specic Regional Changes
Several biases mostly rely on regional features and are discussed in this section.
Sahara: The cold bias between 15°N and 30°N in all simulations and for all seasons (Figures 4c and 4i) is
mostly the signature of a cold bias over the Sahara (Figure 5). It may be due to aerosol specication or to
the failure to consider emissivity lower than the unit in the radiative transfer calculations. The cold bias is
less pronounced with the AP physics because the strong positive bias in the downwelling SW radiation
and the underestimation of the surface albedo compensate the excessive surface cooling due to the over-
estimated value of the surface emissivity.
Tibetan Plateau and Highmountain Asia: There is an overall cold bias in winter between 30°N and 40°N
(Figures 4i and 5), which is particularly strong over the Tibetan plateau and Highmountain Asia where it
is associated with a surface albedo bias (Figure S5) and an overestimation of the snow fraction (Figure 7).
The albedo bias is not present in ORCHIDEE standalone simulations (not shown) and is lower when the
old snow scheme is activated. This is consistent with a difculty of the landatmosphere model to melt
snow leading to a too high albedo inducing a positive feedback on the temperature because of a decit
in net SW radiation at the surface. The weaker bias produced by the old snow scheme is consistent with
the underestimation of the snow albedo, which was already documented by Wang et al. (2013). These
results conrm that land surface atmosphere feedbacks play a signicant role in this region. The tempera-
ture and the albedo biases are weaker in the HighResMIP simulations (not shown) and in nudged simula-
tions (Figures S10 and S11). The weakening of the bias obtained with the increase in resolution or with the
wind nudging conrms that the regional circulation is an important component of the Highmountain
Asia climate. A high resolution allows us to represent more realistic contrasts of the snow cover between
the lowlands and the high mountains. It is also a way to better simulate the role of the orographic barrier
played by the Highmountain Asia that stops the northward transport of moisture originating from the
Indian subcontinent. This barrier explains the dryness of the Tibetan plateau (Krishnan et al., 2019;
Ménégoz et al., 2014; Sabin et al., 2013), where an excess of moisture ux is simulated at coarse resolution,
inducing a positive bias of snow cover that is enhanced by surface feedback. In the same way, by correct-
ing the regional circulation, the nudging can reduce the positive bias of snow cover which impacts the sur-
face albedo.
Central Asian lowlands: The more realistic representation of the snow albedo and the increased decou-
pling for stable boundary layers help obtain more realistic nearsurface temperatures but does not elim-
inate the strong warm bias present in winter on the Central Asian lowlands in CMIP5 (Figure 5, DJF).
The temperature bias is further reduced when the largescale circulation is relaxed toward meteorological
analysis. The nudging reduces also the total precipitable water (not shown) that is greatly overestimated
in this region. These results suggest that the largescale dynamics contribute to the bias by a too strong
moisture advection, the latter limiting radiative cooling. A residual negative bias in surface albedo
(Figure S5 DJF) can also contribute to the warm bias. In summer, the warm bias is also present, but it
is mainly associated with an excess of SW radiation at the surface.
Eastern Siberia: Regardless of the model version, a strong warm bias persists in the extreme northeast of
Siberia, north of the Sea of Okhotsk, and north of the Bering Sea. The bias is not present in the nudgedby
wind simulations (Figure S9), and it is less marked when the new snow scheme and the soil freezing are
activated (NoSnowFreez experiment in Table 2, not shown). The bias is also reduced when the decoupling
is increased. This suggests that both largescale circulation and local processes and their interactions play
a signicant role in this region.
Southern Great Plains: While substantial biases are reduced with respect to the APChoi conguration of
the model used for CMIP5, a warm bias remains over the Southern Great Plains. The Clouds Above the
United States and Errors at the Surface (CAUSES) experiment (Morcrette et al., 2018) in which IPSL par-
ticipated highlighted a strong decit of deepcloud events (reduced in the CMIP6 version with respect to
the version that participated in the intercomparison; Kwinten Van Weverberg, personal communication).
Concerning the precipitation, Van Weverberg et al. (2018, their Figure 13) show that none of the models
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 17 of 33
that participated in CAUSES are able to correctly represent the diurnal cycle of the precipitation evaluated
with the Atmospheric Radiation Measurement Best Estimate (Xie et al., 2010). The nudging does not
allow to reduce the bias. In this region, rainfall comes from two different convective regimes. The rst
regime is associated with a local triggering of convection induced by daytime heating, and the second
regime corresponds to the propagating systems over the Great Plains, initiated in the lee of the Rockies
(Klein et al., 2006). The precipitation associated with the rst regime is fairly well represented by the
LMDZ model with a maximum delayed in the afternoon, which is a robust improvement of the convective
scheme (Hourdin et al., 2020; Rio et al., 2009). The nighttime maximum due to propagative systems is
absent in most models of CMIP5/CMIP6 and in particular in the simulations of LMDZ, which has no
parameterization for this type of propagative system.
Amazonia and Central Africa: The strong warm bias present in the simulation with the AP atmospheric
physics does not exist in the simulation with the 6A physics (Figure 5) as a result of the strong reduction of
the downward SW radiation at the surface discussed in section 3.2.
3.6. Tuning of the Global Model and NearSurface Temperature Over Land
Signicant efforts have been made to improve the physical content of the parameterizations. Yet, they
remain an idealized and approximate representation of processes. As a consequence adjustment and tuning
are unavoidable when all the atmospheric, land surface, and oceanic components are coupled (Hourdin
et al., 2017).
A tuning of subgrid scale orography (SSO) was performed to better represent the atmospheric heat transport
toward the Arctic Ocean, which is a key region for seaice formation and melting. The SSO schemes are
applied to represent the blocking effect of orography at low levels and the breaking of gravity waves (see
Lott, 1998 and more details in Appendix A). The sensitivity experiments (Table 2) reveal that the SSO tuning
has an impact on the nearsurface temperature mostly during the cold season, from November to March (see
Figure 10), which is consistent with the established impacts of orography onto the large scale atmospheric
circulation (Holton, 2004). Increasing the blocking effect of orography through the drag scheme cools
Eurasia and warms western North America. This is consistent with a large blocking effect over the
Rockies when increasing the drag, inducing anomalous southerly warm advection upstream, and northerly
cold advection downstream (Holton, 2004). The sensitivity to the lift that modies the ow direction shows
different effects, with warm anomalies upstream of the Rockies and Himalayas and cold anomalies down-
stream of the Rockies and Himalayas, but with a larger amplitude, different location, and with a
zonalwavenumber 2 structure. The lift effect results from applying a force perpendicular to the local ow
over orographic barriers. It causes larger meridional ow anomalies than the drag, which explains the stron-
ger impact in terms of surface air temperature. The tuning of the version 6A was mainly done by increasing
the drag and slightly reducing the lift parameter so that the tuning may have contributed to enhance the cold
bias over Siberia, while reducing it over North America. However, the temperature anomalies explained by
the new tuning remain small when compared to the bias itself.
An essential aspect of the tuning is to ensure that the radiative budget at the top of the atmosphere is in equi-
librium and that the latitudinal distribution of each component of the radiative budget is as close as possible
to the observations. A particular care was given to the tuning of free model parameters impacting the top of
the atmosphere (TOA) radiation budget (Hourdin et al., 2017). Interestingly, none of the sensitivity studies
described above strongly impacted the TOA radiative budget. This indicates that specic tuning targeting the
land surface processes can be done independently to some extent. Such an approach has not been adopted
for the 6A version of the IPSLCM, but it could improve the performances of the model and reduce some bias
in future versions of the model (Li et al., 2019).
4. Improvement of the Realism of the Hydrological Cycle in the Coupled
Continental SurfaceAtmosphere System
The impact of the more physical hydrological scheme (ctrl) used for CMIP6 (section 2.2) and the impact of
the more realistic convective precipitation documented in Hourdin et al. (2020) on the hydrological cycle are
addressed in this section in two specic ways: the analysis of moisture and energy coupling at the surface at
regional spatialscale and monthly timescale and the analysis of the seasonal cycle of precipitation and river
discharge at the scale of individual watersheds.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 18 of 33
4.1. Soil MoistureEvaporationRadiationPrecipitation Coupling
The impact of the modied parameterizations on surface soil moisture, net SW radiation at the surface, eva-
poration, and precipitation is documented at regional scale in order to ensure homogeneous climate condi-
tions to prevail. We focus on two hotspot regions (Koster et al., 2004) where the soil moistureatmosphere
coupling is strong: the Central North America (CNA) region as dened in the Special Report on
Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation (SREX)
(Seneviratne et al., 2012) and a box in the Sahel (10, 30°E, 020°N). A third region corresponding to
Western Europe (WE) where the coupling is weaker is also considered. The grid points corresponding to
the WE box are selected according to the KoeppenGeiger climate classication system (Kottek et al., 2006)
(Region 21 in Figure S12).
For these regions, combined distributions of soil surface moisture, evaporation, net radiation at the surface,
and precipitation for the four reference congurations (6Actrl, APctrl, APChoi, 6AChoi) and for different
sets of observations (Table 3) are constructed based on monthly values for a 10year long period in which
all observations are available (20012010).
First considering the distribution for the simulated surface soil moisture itself, the regional histograms of the
surface soil moisture show that the Choi land surface hydrological scheme produces a very narrow distribu-
tion with unrealistic null value of the surface soil moisture for lower boundary and low maximum values of
the surface soil moisture (not shown). These characteristics of the surface soil moisture with the Choi
scheme can be explained by the crude representation of the hydrology by this scheme for which the
Figure 10. Top panel: 2 m temperature (CMIP6 variable tas) anomalies (in K) induced by the drag and lift parameters
over (left) Eurasia (EUR, 2080°N, 10°W180°E) and (right) North America (NAM, 2080°N, 18010°W). The blue line
indicates the difference between 6AL+ and 6AL. The red line indicates the difference between 6AD+ and 6AD.
Bottom panels: surface air temperature anomalies (in K) induced by the (left) drag and (right) lift parameters in the
Northern Hemisphere. Colors are shown only for signicant areas ( pvalue of Student ttest lower than 10%).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 19 of 33
surface layer exists only intermittently. When considering the GLEAM and ESACCI soil moisture products,
the width of the distribution is signicantly smaller with the ESACCI product than with the GLEAM
product (not shown). The fact that GLEAM takes the upper 010 cm into account while ESACCI
correlates better with soil moisture up to 5 cm depth (Dorigo et al., 2017) while the surface soil moisture
in the simulations corresponds to the moisture in the top 10 cm of soil might contribute to the differences.
In addition, GLEAM and ESACCI soil moisture products are considered as observations, but they are
highly dependent on the underlying models used to produce them and therefore suffer limits identied by
Koster et al. (2009) that call for great caution regarding the reliability of the absolute values retrieved. For
these reasons we prefer using the standardized Soil Moisture Index (SMI) dened in R. D. Koster
et al. (2009; see their Equation 1), and the soilmoisture information at monthly time scale is mostly used
to discriminate between very dry, moderately dry, moderately moist, and very moist soils in the
corresponding regional distributions for evaporation, net SW radiation at the surface, and precipitation.
For soil moisture in the Sahel region (Figure 11) the summer observations feature a Ushaped distribution in
which dry and saturated states prevail. This Ushaped distribution is reproduced by both schemes (Choi and
Figure 11. Regional histograms computed from monthly values of the individual grid points corresponding to the Sahel
box (10:30°E, 0:20°N) in JJA. The histograms are constructed for a 10year long period in which all observations are
available (20012010). Each row is dedicated to a particular variable: surface standardized soil moisture (rst row),
net SW radiation at the surface (second row), evaporation (third row), and precipitation (fourth row). The rst four
columns correspond to the reference experiments, and the last two columns correspond to the different sets of
observations indicated above the corresponding histograms. The colors depict the PDF from the minimum to rst
quartile (dark red) from rst quartile to the median (pale orange), from median to third quartile (cyan line), and from the
third quartile to the maximum (blue line). For soil moisture, the yaxis is cut at .25 (representing 25% of the quartile)
for the sake of readability but the driest quartile peaks at 0.8 (corresponding to 80% of the quartile) for APctrl and the
moister quartile peaks at .8 for APChoi and APctrl. For evaporation theyaxis is cut at .14 (corresponding to 14% of a
quartile), but 55% (APChoi) and 90% (APctrl) of the evaporation associated with the rst quartile is less than 0.1 mm/day.
For the precipitation, the yaxis is cut at .12, but 70% , 85%, 15 %, and 40 % of the precipitation associated with the
driest soil moisture quartile are less then 0.1 mm/day for APChoi, APctrl, 6AChoi, and 6Actrl and 20% and 10 %
for GLEAM and ESACCI.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 20 of 33
ctrl) with strong differences: Choi favors the moistest contents much more than the observations while ctrl
leads to a trimodal distribution. This feature has been observed for several other regions with different
climate and is the signature of using one dominant soil texture among three possible ones in each grid
cell of the region, while in reality, many different soil textures coexist and lead to a mixed behavior.
However, the Ushape is also present indicating that the scheme tends to favor dry or saturated situations
for each texture as well. For all regions, the highest value of the net SW radiation is overestimated by as
much as 20 Wm
2
. This holds for both AP and 6A versions of the model and for each soil moisture
quartile. Various hypothesis can be formulated: This bias can either rely on a difculty in processing
CERES observations to retrieve the net radiation at the surface or rely on LMDZ. In this case, a problem
with the radiative transfer code or a lack of simulated clouds or an underestimation of their radiative
impact can be invoked. For the Sahel, when the surface is moist, the 6Actrl conguration tends to
underestimate the occurrence of situations with an elevated evaporation rate and overestimate the
occurrence of situations with low values of the net SW radiation (second and third columns in Figure 11).
This feature can be interpreted as a too frequent occurrence of radiationlimited evaporative regimes with
Figure 12. Regional histograms computed from monthly values of the individual grid points corresponding to the SREX CNA region (Seneviratne et al., 2012) in
JJA. The histograms are constructed for a 10year long period in which all observations are available (20012010). Each row is dedicated to a particular variable:
surface standardized soil moisture (mrsos, rst row), net SW radiation at the surface (second row), evaporation (third row), and precipitation (fourth row).
The rst four columns correspond to the four reference experiments and the last two columns to the different sets of observations indicated above the
corresponding histograms. The colors depict the PDF from the minimum to rst quartile (dark red) from rst quartile to the median (pale orange), from
median to third quartile (cyan line), and from the third quartile to the maximum (blue line).
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 21 of 33
respect to the soil moisturelimited evaporative regimes in this region. For CNA and for AP physics radiation
is either insensitive to soil moisture (APChoi) or the low radiation is surprisingly associated with the driest
soil quartile (APctrl) while, for 6A physics, low radiation associated with extended cloud cover is rather asso-
ciated with the wettest quartile, which is consistent with the CERES product. Over both CNA and Sahel and
for the driest surface soil moisture quartile, AP tends to favor little or no rainfall at the monthly time scale,
probably oversimulating dry events. This feature is much weaker with 6A physics and in better agreement
with the observations (bottom row in Figures 11 and 12). When Choi hydrology is activated with 6A physics,
dry soils tend to have a sustained rate of evaporation, while the 11layer hydrology also allows low evapora-
tion rates consistent with the observations. When AP physics is activated, dry soils tend to be associated with
too weak evaporation rates, this feature being more pronounced with 11layer hydrology. Additional infor-
mation concerning the evolution of the performances of the atmospheric model in the Sahel with the AP and
6A atmospheric physics are given in Diallo et al. (2017). In the hotspot regions, the 6Actrl conguration used
for CMIP6 is the closest to observations due to both improved atmospheric physics and representation of soil
hydrology. Low precipitation rates (at monthly timescale) associated with dry soil are also overestimated in
Western Europe. In this region where the soil moistureatmosphere coupling is expected not to be dominant,
the simulated net SW radiation, the simulated evaporation, and the simulated precipitation appear to be
more sensitive to the soil moisture than the observed ones (Figure S13).
4.2. Seasonal Cycle of Precipitation and River Discharge
Figure 13 shows the seasonal cycle of precipitation observed and simulated by the four sensitivity experi-
ments described above for 14 major watersheds together with the seasonal cycle of the river discharge
observed and simulated at 14 stations on the rivers of the same major basins. For four out of the ve boreal
basins (Yukon, McKenzie, Ienisei, and Lena) the precipitation is often overestimated in all congurations.
For some basins including Mississippi, Congo, and Amazonia, the seasonal cycle of simulated precipitation
is signicantly improved in volume or in phase in the conguration used for CMIP6 (6Actrl). For instance, in
Tocantins basin in the Cerrado, the duration of the dry season is now reduced in agreement with the obser-
vations. This improvement can be attributed to changes in the parameterizations of the atmospheric physics.
The impact of the LSM is limited, except over some midlatitude basins such as the Danube where the volume
of precipitation is controlled by atmospheric physics and continental hydrology and is overestimated with
the 6Actrl conguration used for CMIP6. With the 6Actrl conguration, simulated river discharges are also
improved for the Mississippi, Amazonia, and Congo, owing to improved precipitation volume. The seasonal
timing of river ow is different from that of rainfall because of the time needed for water to circulate in soils
and along river systems after it has reached the ground. This timing is usually correct, with errors resulting
from those of the simulated precipitation (e.g., intensity and location of rainfall events inside the water-
sheds), simulated land surface processes (e.g., snowmelt dynamics, permafrost, and transit times in the soil),
and the fact that residence times of the routing reservoirs only depend on the type of reservoir (stream, over-
land, and groundwater) and the grid cell slope, while other regional factors can be important. In particular,
the absence of oodplains in all the simulations largely explains the overestimation of river discharge in the
Niger (d'Orgeval et al., 2008) and Congo and may contribute to the early peak ows of the Amazon
(Guimberteau et al., 2012). The parameterizations of the land surface processes have a major effect in the ve
Arctic rivers, with a higher ow and earlier maximum when ground freezing is activated. This effect
improves the simulated discharge in the two basins with the largest fraction of permafrost (Yenisei and
Lena, in eastern Siberia). In the other three basins (Ob, Yukon, and McKenzie), the extent of frozen soils
may be overestimated, and the overestimation of the river discharge by 6Actrl can also be related to the lack
of dams and oodplains in the model (Gouttevin et al., 2012), with a potential feedback on permafrost extent,
since a stronger cooling is required to freeze a wet soil than a dry soil. The Bramahputra (India) discharge
shows improved volume and seasonality with the 6Actrl conguration, while the maximum of the precipita-
tion is underestimated. For this particular river that originates from the Angsi glacier located in Tibet, the
change in atmospheric physics improves the timing while the maximum discharge is improved (reduced)
with the activation of the soil freezing. This nonintuitive impact of soil freezing is caused by an atmospheric
feedback, with less precipitation in the watershed if the freezing is activated. Yet, the positive bias of all
simulated discharges might rather be related to massive irrigation in this basin (Guimberteau et al., 2012),
which is not taken into account in these simulations.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 22 of 33
Figure 13. Multiannual mean seasonal cycle of the precipitation (upper panel) and of river discharge (lower panel)
observed and simulated for 14 major river basins and for the four reference experiments: 6Actrl, Choi6A, APctrl, and
ChoiAP and for the NoSnowFreezexperiment described in Table 2. The observations refer to the GPCP product for
precipitation and to the Global Runoff Data Center (GRDC) database for the river discharges (Milliman &
Farnsworth, 2011). The gray shaded areas indicate the interannual variability of the observed precipitation in the basin
area (upper panel) and the interannual variability of the river discharge at the measurement stations.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 23 of 33
5. Concluding Discussion
The quality of the coupled atmosphereland continental surface system implemented in the IPSLCM for
CMIP6 is evaluated, and the relative role of atmospheric and land surface processes in controlling the cou-
pling at the surface is analyzed and quantied. The following conclusions are reached:
The improvement of the radiative balance and in particular the surface downward SW radiation makes it
possible to reduce several temperature biases, some of which were shared by many models that partici-
pated in the CMIP5 exercise (e.g., summer bias in midlatitudes Stouffer et al., 2017). This conrms the
essential role of the radiation and its interactions with clouds for continental climates.
The temperature in the surface layer of the polar regions is signicantly improved thanks to the rened
turbulent diffusion scheme for stable situations and to the new longwave radiative scheme in LMDZ ver-
sion 6A (Vignon et al., 2018). The boreal regions respond with a slightly excessive reduction of the daily
minimum temperature while in CMIP5, several models including LMDZ shared a warm bias (Wei
et al., 2017). A more detailed consideration of the turbulent mixing linked to the subgrid orography or
high vegetation may help to partially compensate for this cooling, but further tests and evaluation are
necessary.
With the exception of the surface albedo, the snow cover, and to a lesser extent the evaporation, the over-
all structure of the nearsurface biases is only marginally sensitive to the land surface scheme whose
impact is mostly relevant at the regional scale. However, for a given description of the atmospheric phy-
sics, the differences induced by the change in the LSM are statistically signicant (at the 5% level) over
most of the continents for all variables examined but precipitation.
The multilayer hydrology gives a representation of the surface soil moisture in better agreement with
available observations than the Choi scheme, and the representation of evaporation in regions of strong
coupling of the continental surface with the atmosphere is signicantly improved.
The snow scheme of intermediate complexity implemented in ORCHIDEE leads to a better description of
the snow cover on the continents. Mountainous regions and in particular the Tibetan Plateau and
Highmountain Asia remain challenging because radiative feedbacks and an imperfect description of
the circulation in these regions at regional scale induce a strong cold bias. Further renements of the snow
scheme over complex terrains and of the atmospheric circulation are required to reduce these biases.
The calculation of the fraction of frozen water in the soil implemented in the multilayer hydrology com-
bined with the improved realism in volume and seasonality of the precipitation simulated with the 6A ver-
sion of LMDZ has improved the seasonal cycle of rivers discharge in several major river basins.
Further developments based on the current version of the coupled atmosphereland continental surface sys-
tem are also identied:
The attempt to take into account sources of turbulent mixing such as orographyinduced small scales
gravitywave drag (Steeneveld et al., 2008) or the drag induced by vertical obstacles penetrating the
boundary layers such as trees needs to be further rened.
The benet of using the dynamical roughness lengths as proposed by Massman (1999) and tested locally
by Su et al. (2001) over homogeneously vegetated surfaces (shrub, cotton, grass) has still to be thoroughly
evaluated in the context of the imperfect heterogeneous landatmosphere coupling. In this context, the
bulk formulae for ux calculation use a unique value of the roughness length, aggregated over possibly
highly heterogeneous subgrid surfaces, and a potentially wide range of contrasting subgrid surfaces sees
the same boundary layer properties. The development of more robust parameterizations for ux calcula-
tion over heterogeneous surfaces could benet in the future from highresolution simulations such as
Large Eddy Simulations.
For CMIP6, even though it would have reduced an overall overestimation of the evaporation, we consid-
ered it preferable not to activate the evaporation resistance of the bare soil in its current state to avoid rein-
forcing a warm bias in summer that would affect the quality of the simulations. Further work is needed to
better calibrate the intensity of the evaporation resistance, which also impacts the ratio of transpiration to
total evapotranspiration, shown to exert a key inuence on biophysical feedback strength in both present
and future climates (Zeng et al., 2017). Owing to the number of intricated parametrizations in a CM, such
work cannot be done in isolation, and our results show that particular attention must be paid to the uncer-
tainties of cloud parameterizations and cloudradiation interactions.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 24 of 33
The multilayer hydrology enables to consider new developments for the CM. One of them is the introduc-
tion of realistic groundwater description, which may alleviate some biases by means of enhanced evapo-
transpiration owing to capillary rise from the water table (Campoy et al., 2013; Wang et al., 2018). The
inclusion of irrigation in the simulations could also help reducing persistent biases (Puma &
Cook, 2010), especially in places where it is fed by groundwater abstraction at nonrenewable rates, like
in India or the U.S. Great Plains (AlYaari et al., 2019; Famiglietti, 2014).
In the version of ORCHIDEE used for CMIP6, the soil freezing is diagnosed in each soil layer, but the
latent heat release/consumption associated with water freezing/thawing is not accounted for. This is,
together with the better description of soil organic matter decomposition (Guimberteau et al., 2018), a pre-
liminary step to account for the biogeochemical implications and positive feedback to global warming due
to permafrost disappearance.
Since the CMIP6 version, a description of the nitrogen cycle and its coupling to the carbon cycle has been
implemented in ORCHIDEE (Vuichard et al., 2019). The impact of soil nitrogen availability (and more
generally of soil nutriments) is crucial for plant growth but also for the energy and water cycle. Very
recently we also included an ensemble of developments to improve the representation of forest dynamic
and forest management with the inclusion of (i) a new canopy radiative transfer scheme (two streams
model), (ii) a new carbon allocation scheme based on observed allometric relationships, and (iii) age
and diameter classes and management practices (from natural to coppices). These developments
described in Naudts et al. (2015) have a direct impact on the surface climate, changing the albedo of forest,
the roughness length (varying with tree height dynamic), the latent and sensible heat uxes, and the over-
all surface temperature (see an application over Europe in Naudts et al., 2016).
Interestingly, none of the sensitivity tests to the surface processes described in this paper signicantly
impacted the TOA radiative budget, an essential target of the tuning of global CMs. This indicates that
there is latitude for independent tuning for TOA radiation and for the land surface processes. Such an
approach has not been adopted for the 6Actrl version of the IPSLCM, but it could improve the perfor-
mance of the model and reduce some bias in future versions of the model (Li et al., 2019). The tuning
of the free parameters is now recognized as necessary step in model development (Hourdin et al., 2017)
that should not rule out the improvement of the physical content of parameterizations.
Appendix A: GravityWave and HighVegetation Drag Induced TKE
LMDZ deals with two effects of the subgrid orography on the atmospheric ow:
1. the orographic blocking effect (called drag),
2. the orographic effect on the wind direction (called lift).
The drag and lift effects are described in Lott (1999). These two effects were modied during the tuning pro-
cess (Gastineau et al., 2020). The drag and lift parameterizations (Lott & Miller, 1997) encompass two pro-
cesses: (i) the blockingof the ow leading to a ow separation at the relief anks and (ii) the
orographic gravitywave drag. The latter accounts for the drag due to wave breaking in the middle atmo-
sphere as well as for the drag induced by lowlevel dissipation and breaking of trapped lee waves
(Lott, 1998). The drag effect is calculated applying a local force opposed to the local ow, and it is used in
all CMs (Sandu et al., 2019). The lift effect is less widely used and involves a force perpendicular to the
local ow.
For the setup of the sixth version of the model, the effect of the drag exerted by vegetation protruding into the
rst model layers has also been parameterized in LMDZ following Nepf (1999) and Masson & Seity (2009).
Orographic gravitywave breaking and dissipation (e.g., Epifanio & Qian, 2008; Sun et al., 2015) as well as
owcanopy interactions (Finnigan, 2000) have been shown to be common paths to turbulence generation.
More generally, every drag exerted on an air ow is associated to a loss of large scale kinetic energy and to an
energy cascade from large scale kinetic energy to small scale turbulence (TKE) and ultimately to dissipation
by molecular viscosity and conversion into enthalpy (Stull (1990), section 5.3).
In LMDZ, any drag parameterization dgcalculates a wind tendency [du/dt|
dg
,dv/dt|
dg
] for all vertical
levels in each atmospheric column. This tendency can be expressed as the vertical divergence of a momen-
tum stress ðρũ~
wdg;ρe
v~
wdgÞ;namely,
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 25 of 33
u
tdg
¼
1
ρρũ~
wdg
z;(A1)
v
tdg
¼
1
ρρe
v~
wdg
z;(A2)
where ρis the air density and uand vare the zonal and meridional components of the wind vector, respec-
tively. The loss of kinetic energy kin an atmospheric layer associated to the parameterized drag dgthus
reads
k
tdg
¼δzu
ρũ~
wdg
zþvρe
v~
wdg
z
"#
¼δzρũ~
wdg
u
zþρũ~
wdg
v
z

|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ΓKðzÞ
δz
uρũ~
wdg
zþ
vρe
v~
wdg
z
"#
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ΨðzÞ
;
(A3)
where δzis the depth of the considered atmospheric layer. We will see hereafter that Γ
K
is an exchange
term between large scale kinetic energy and TKE while Ψcorresponds to the vertical divergence of power
associated to the parametrized stress. Once integrated over a whole atmospheric column, as
0Ψdz =0,
Equation A3 reads
tKjdg ¼Z
0
zuρũ~
wdz þZ
0
zvρe
v~
wdz;(A4)
where
K¼Z
0
kdz ¼Z
0
ρðu2þv2Þ
2dz (A5)
(Boville & Bretherton, 2003).
To guarantee energy conservation in LMDZ version 6A, tKjdg was initially calculated for each drag parame-
terization and then converted into enthalpy in each atmospheric column. To account for a more realistic
mixing in the boundarylayer and to preclude articial thermal decouplings over the continents, the loss
of energy associated to the highvegetation and orographic gravitywave drag was then transferred to subgrid
TKE before being converted into enthalpy, thereby enhancing the mixing in the boundarylayer. Practically,
this is done as follows.
The parameterization of the vertical turbulent mixing in LMDZ version 6A is based on a local diffusion
scheme combined with a massux scheme for convective boundary layers, the socalled thermal plume
model(Hourdin et al., 2002; Rio et al., 2010). The local diffusion scheme is a 1.5 order closure Kgradient
scheme developed by Yamada (1983) in which the diffusion coefcients depend on the TKE calculated with
a prognostic equation
TKE
t¼1
ρ
zðρKe
TKE
zÞ
|fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl}
Turbulent diffusion
uwu
z
vwv
z
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
Shear production
þg
θv
wθ
v
|fflfflfflfflffl{zfflfflfflfflffl}
Buoyancy term
TKE3=2
cl
|fflfflfflfflfflffl{zfflfflfflfflfflffl}
Dissipation
;
(A6)
where cis a real constant, θ
v
the virtual potential temperature, uwand vwthe components of the tur-
bulent momentum ux, g
θv
wθ
vthe buoyancy ux, and K
e
a turbulent diffusion coefcient.
The conversion of large scale energy into TKE due to the orographic gravitywave drag and highvegetation
drag can therefore be taken into account by including the Γ
K
terms associated to those parameterizations as
additional shear productionterms into Equation A6. For the vegetation, the drag coefcient is propor-
tional to the fraction of protruding vegetation in the gridbox. More details and sensitivity tests can be found
in Vignon (2017). One might also want to add the TKE tendency due to the owblocking component of the
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 26 of 33
subgrid orographic drag scheme. However, the underlying physical mechanism responsible for the energy
cascade associated to ow blocking is not a priori obvious. This aspect deserves further investigation.
Appendix B: A Posteriori Correction of the ScreenLevel Variables
B1 Diagnostics at the Screen Level
The calculation of the screenlevel variables, tas (2 m temperature), huss (2 m specic humidity), and uas
and vas (eastward and northward surface wind) is done iteratively following Hess (1995). It is based on
the MoninObukhov similarity theory for the surface layer and the bulk formulation of the turbulent ux
proposed by Louis et al. (1982). The 2 m relative humidity, hurs, is then diagnosed from huss and the satu-
rated specic humidity at temperature tas.
The wind, the temperature, and the specic humidity proles in the surface layer follow equations:
κu
u
¼lnðz
z0m
ÞΨMðz
LÞ
ðΘΘsÞ
Θ
¼1
κðlnðz
z0h
ÞΨHðz
LÞþΨHðzoh
LÞÞ
ðqqsurf Þ
q
¼1
κðlnðz
z0h
ÞΨHðz
LÞþΨHðzoh
LÞÞ;
8
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
:
(B1)
with κthe empirical von Karman constant, Lthe MoninObukhov length, and Ψthe stability functions for
the stability parameter ζ=z
L.u
is the friction velocity, Θ
the temperature scale, and q
the humidity
Figure B1. Cumulated histogram of the reconstruction errors for daily 2 m t. The yaxis is logarithmic. The red curve
corresponds to the difference between the daily mean obtained with the original run and with the instantaneous
values bounded with the surface and rst atmospheric level temperature (ON experiment). The blue curve corresponds to
the a posteriori correction.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 27 of 33
scale. An empirical formulation for the stability functions is given by Dyer (1974). According to the
MoninObukhov theory, L, u
,Θ
, and q
are evaluated at the surface and are independent of zin the con-
stant ux layer. A rst guess of the screen variables is estimated owing to Equation B1. Then the Louis
bulk formulation and the scale variables are used to calculate an updated value of the screen level vari-
ables. In situations where the turbulence is vanishing and the atmosphere above the surface is dry but
the surface soil moisture is signicantly above the residual value, a wrong diagnostic of q
surf
in
ORCHIDEE led to inconsistencies in the stability diagnostics between the rstguess evaluation and the
use of the Louis formulation. In such conditions, the calculation can produce unrealistic (overestimated)
values of tas up to 450 K together with negative values of relative humidity. Luckily, apart from a few
exceptional events, this occurs only one time a day at most. Thanks to that it was possible to a posteriori
correct the screen level values for simulations for which the minimum daily relative humidity was
archived. In the vast majority of cases, these failures occur in stable conditions. In such conditions
Ψ
H
=
5z
Land one shows easily that Θis a monotonous function of z, which implies that Θis comprised
between Θ
s
, the surface temperature, and Θ
1
, the temperature at the rst atmospheric level of the model.
A simulation where the screen level temperature is bounded at each timestep with the surface and the air
temperature at the rst atmospheric level will then be used to validate the a posteriori correction.
B2. A Posteriori Correction for the ScreenLevel Variables
The a posteriori reconstruction algorithm is described hereafter. The general idea of the algorithm is to
replace the erroneous values (daily maximum air temperature, tasmax or surface daily minimum relative
humidity, hursmin) by an interpolation between the previous and the following day without failure. The rar-
ity of the failure of the screen variable calculation makes this approach feasible. The erroneous values (fail-
ure) are detected by looking for negative values of hursmin.
Step 1: We detect possible failure by identifying all the grid points and days (of index k) for which the esti-
mated nearsurface humidity is negative.
Step 2: We correct the daily mean temperature by correcting the maximum in the daily mean using infor-
mation from the last and next day without failure as follows.For the derivation, we denote by Tthe 2 m
temperature, tas.The daily average value will be noted T=ΣN
1Ti=N, where Nis the number of timesteps
iwithin a day and the maximum T
max
. We introduce the daily maximum anomaly D=Tmax T.We
apply the interpolation in time between the last (l) and next (n) day without failure to D, leading to the
corrected value D=(1a)D
l
+aD
n
with a=(kl)/(nl).Then we compute the corrected daily aver-
aged temperature as Tnoticing that
TT¼ðT
max TmaxÞ=N¼ðTTþD
DÞ=N;(B2)
so that
T¼TþðD
DÞ=ðN1Þ:(B3)
Step 3: We correct the maximum temperature from the corrected daily mean temperature Tand inter-
polated daily anomaly Das
T
max ¼TþD;(B4)
which can be written as well using Equation B3 as
T
max ¼TþðTTÞþD¼TþND
D
N1:(B5)
For the daily values of tas, this approach leads to replacing a potential error of about 1
96 × 150 K (for a max-
imum error of 150 K on the instantaneous value of the temperature, 96 being the number of timesteps in one
day for LMDZ version 6) by an uncertainty of at most 1
96 of the daily maximum anomaly (that is about 10K
96 ),
that is, more than 10 times less. For tasmax the reconstruction procedure avoids creating extremes based on
erroneous (and irrealistic) screen variable values.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 28 of 33
A similar threestep approach is applied to correcting the relative humidity but taking as an information
from the last and previous day without failure, the ratio R¼RH min
RH , so that the procedure reads
Step 1: Same as for T.
Step 2: We apply the interpolation in time between the last (l) and next (n) day without failure to R, lead-
ing to the corrected valueR
=(1a)R
l
+aR
n
with a=(kl)/(nl). Then we compute the corrected
daily averaged relative humidity as RH noticing that
RHRH ¼RH
min RHmin

=N¼RRHRHmin
N;(B6)
and the corrected values are calculated with the following equations.
RH¼
RH
RHmin
N

1
R
N

:(B7)
Step 3: We correct the minimum relative humidity from the corrected daily mean relative humidity RH
and interpolated ratio R
as
RH
min ¼RRH:(B8)
B3. Evaluation of the Uncertainty Relying on the A Posteriori Correction
The a posteriori reconstruction (hereafter called OFF) is evaluated against the results of the nearsurface
temperature diagnosed online in the model (hereafter called ON) and bounded at each timestep with the
surface and the air temperature at the rst atmospheric level. In the ON experiment the bounding is applied
only for diagnostic purpose and does not affect the behaviour of the model.
For each grid point and each day of the 36 years of an AMIP experiment, the reconstruction error is evalu-
ated with the difference between the OFF and the ON experiments.
Figure B1 shows the cumulated histogram of reconstruction errors with the a posteriori method and the ON
bounding method for the daily mean and maximum daily temperature. For the majority of grid points and
days, the OFF and ON methods give similar results. The reconstruction error lies within the range (0.2 K,
0.4 K). These small differences between the two methods for daily values show that the nearsurface tem-
perature is not fundamentally modied by the OFF correction compared to what would be obtained with
an online correction. The reconstruction errors for the monthly mean nearsurface temperature would have
been negligible compared to the daily errors, being 30 times smaller than daily errors.
Data Availability Statement
The version of LMDZ and ORCHIDEE used for the production of CMIP6 will be made available at the fol-
lowing address (http://www.lmd.jussieu.fr/lmdz/pub). In the ORCHIDEE community, the model is
referred as Orchidee Trunk,which is the ofcial version developed at IPSL. The version used for the spe-
cic simulations runs for this paper is the svnrelease 3427 in the LMDZ6/branches/IPSLCM6.0.15 and the
svnrelease 5626 in the tags/ORCHIDEE_2_0/ORCHIDEE_OL branche. Simulations data used in the pre-
sent paper will be made available with a DOI if the paper is accepted for publication.
References
Adler, R., Huffman, G., Chang, A., Ferraro, R., Xie, P., Janowiak, B., et al. (2003). The Version 2 Global Precipitation Climatology Project
(GPCP) monthly precipitation analysis (1979Present). Journal of Hydrometeorology,4, 11471167.
AïtMesbah, S., Dufresne, J. L., Cheruy, F., & Hourdin, F. (2015). The role of thermal inertia in the representation of mean and diurnal
range of surface temperature in semiarid and arid regions. Geophysical Research Letters,42, 75727580. https://doi.org/10.1002/
2015GL065553
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 29 of 33
Acknowledgments
This work was supported by the
DEPHY2 project funded by the French
national program LEFE/INSU. The
CMIP6 project at IPSL used the HPC
resources of TGCC under the
allocations 2016A0030107732, 2017
R0040110492, and 2018R0040110492
(project gencmip6) provided by GENCI.
One of the authors (Y. Zhao) beneted
from the French state aid managed by
the ANR under the Investissements
d'avenir'' program with the reference
ANR11IDEX0004 17EURE0006.
This study beneted from the ESPRI
(Ensemble de Services Pour la
Recherche l'IPSL) computing and data
centre (https://mesocentre.ipsl.fr),
which is supported by CNRS, Sorbonne
Universite, Ecole Polytechnique, and
CNES and through national and
international grants. Analysis
presented in 4.14.1 beneted from the
support of the CMUG and ESACCI
programs (http://www.esacmugcci.
org). We also thank two anonymous
reviewers for their helpful comments
on the original manuscript.
AlYaari, A., Ducharne, A., Cheruy, F., Crow, W. T., & Wigneron, J. P. (2019). Satellitebased soil moisture provides missing link between
summertime precipitation and surface temperature biases in CMIP5 simulations over conterminous United States. Scientic Reports,9,
1657. https://doi.org/10.1038/s41598018383095
Arnell, N. W., & Gosling, S. N. (2013). The impacts of climate change on river ow regimes at the global scale. Journal of Hydrology,486,
351364. https://doi.org/10.1016/j.jhydrol.2013.02.010
Betts, A. K., Ball, J. H., Beljaars, A. C. M., Miller, M. J., & Viterbo, P. A. (1996). The land surfaceatmosphere interaction: A review based on
observational and global modeling perspectives. Journal of Geophysical Research,101(D3), 72097225. https://doi.org/10.1029/
95JD02135
Boé, J. (2013). Modulation of soil moisture/precipitation interactions over France by large scale circulation. Climate Dynamics,40, 875 892.
https://doi.org/10.1007/s0038201213806
Boé, J., & Terray, L. (2008). Uncertainties in summer evapotranspiration changes over Europe and implications for regional climate
change. Geophysical Research Letters,35, L05702. https://doi.org/10.1029/2007GL032417
Bontemps, S., Boettcher, M., Brockmann, C., Kirches, G., Lamarche, C., Radoux, J., et al. (2015). Multiyear global land cover mapping at
300 M and characterization for climate modelling: Achievements of the land cover component of the ESA climate change initiative. In
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences (pp. 323328). https://doi.org/10.5194/
isprsarchives-XL-7-W3-323-2015
Boone, A., & Etchevers, P. (2001). An intercomparison of three snow schemes of varying complexity coupled to the same land surface
model: Localscale evaluation at an Alpine site. Journal of Hydrometeorology,2(4), 374394. https://doi.org/10.1175/1525
7541(2001)002<0374:AIOTSS>2.0.CO;2
Boucher, O., Servonnat, J., Albright, A. L., Aumont, O., Balkanski, Y., Bastrikov, V., et al. (2020). Presentation and evaluation of the IPSL
CM6ALR climate model. Journal of Advances in Modeling Earth Systems,12, e2019MS002010. https://doi.org/10.1029/2019MS002010
Boville, B. A., & Bretherton, C. S. (2003). Heating and kinetic energy dissipation in the NCAR community atmosphere model. Journal of
Climate,16, 38773887. https://doi.org/10.1175/15200442
Campoy, A., Ducharne, A., Cheruy, F., Hourdin, F., Polcher, J., & Dupont, J. C. (2013). Response of land surface uxes and precipitation to
different soil bottom hydrological conditions in a general circulation model. Journal of Geophysical Research: Atmospheres,118,
10,72510,739. https://doi.org/10.1002/jgrd.50627
Chalita, S., & Le Treut, H. (1994). The albedo of temperate and boreal forest and the Northern Hemisphere climate: A sensitivity experi-
ment using the LMD GCM. Climate Dynamics,10, 231240. https://doi.org/10.1007/BF00208990
Cheruy, F., Campoy, A., Dupont, J. C., Ducharne, A., Hourdin, F., Haeffelin, M., et al. (2013). Combined inuence of atmospheric physics
and soil hydrology on thesimulated meteorology at the SIRTA atmospheric observatory. Climate Dynamics,40, 22512269. https://doi.
org/10.1007/s003820121469y
Cheruy, F., Dufresne, J. L., Ait Mesbah, S., Grandpeix, J. Y., & Wang, F. (2017). Role of soil thermal inertia in surface temperature and soil
moisturetemperature feedback. Journal of Advances in Modeling Earth Systems,9,29062919. https://doi.org/10.1002/2017MS001036
Cheruy, F., Dufresne, J. L., Hourdin, F., & Ducharne, A. (2014). Role of clouds and landatmosphere coupling in midlatitude continental
summer warm biases and climate change amplication in CMIP5 simulations. Geophysical Research Letters,41, 64936500. https://doi.
org/10.1002/2014GL061145
Coindreau, O., Hourdin, F., Haeffelin, M., Mathieu, A., & Rio, C. (2007). Assessment of physical parametrizations using a global climate
model with strtchable grid and nudging. Monthly Weather Review,135, 14741490. https://doi.org/10.1175/MWR3338.1
d'Orgeval, T., Polcher, J., & de Rosnay, P. (2008). Sensitivity of the West African hydrological cycle in ORCHIDEE to inltration processes.
Hydrology and Earth System Sciences,12, 13871401.
De Rosnay, P., Polcher, J., Bruen, M., & Laval, K. (2002). Impact of a physically based soil water ow and soilplant interaction represen-
tation for modeling largescale land surface processes. Journal of Geophysical Research,107(D11), 4118. https://doi.org/10.1029/
2001JD000634
Dee, D., Uppala, S., Simmons, A., Berrisford, P., Poli, P., Kobayashi, S., et al. (2011). The ERAInterim reanalysis: Conguration and per-
formance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society,137(656), 553597.
Diallo, F. B., Hourdin, F., Rio, C., Traore, A. K., Mellul, L., Guichard, F., & Kergoat, L. (2017). The surface energy budget computed at the
gridscale of a climate model challenged by station data in West Africa. Journal of Advances in Modeling Earth Systems,9, 27102738.
https://doi.org/10.1002/2017MS001081
Dorigo, W., Wagner, W., Albergel, C., Albrecht, F., Balsamo, G., Brocca, L., et al. (2017). ESA CCI Soil Moisture for improved Earth system
understanding: Stateofthe art and future directions. Remote Sensing of Environment.
Ducoudré, N., Laval, K., & Perrier, A. (1993). SECHIBA, a new set of parameterizations of the hydrologic exchanges at the landatmosphere
interface within the LMD atmospheric general circulation model. Journal of Climate,6, 248273.
Dyer, A. J. (1974). A review of uxprole relationships. BoundaryLayer Meteorology,7, 363372. https://doi.org/10.1007/BF00240838
Eltahir, E. A. B. (1998). A soil moisture rainfall feedback mechanism: 1. Theory and observations. Water Resour ces Research,34(4), 765776.
https://doi.org/10.1029/97WR03499
Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer, R. J., & Taylor, K. E. (2016). Overview of the Coupled Model
Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geoscientic Model Development,9(5), 19371958.
https://doi.org/10.5194/gmd919372016
Famiglietti, J. S. (2014). The global groundwater crisis. Nature Climate Change,4(11), 945948.
Finnigan, J. (2000). Turbulence in plant canopies. Annual Review of Fluid Mechanics,32(1), 519571. https://doi.org/10.1146/annurev.
uid.32.1.519
Garratt, J. R., & Hicks, B. B. (1973). Momentum, heat and water vapour transfer to and from natural and articial surfaces. Quarterly
Journal of the Royal Meteorological Society,99(422), 680687. https://doi.org/10.1002/qj.49709942209
Gastineau, G., Lott, F., Mignot, J., & Hourdin, F. (2020). Alleviation of an arctic sea ice bias in a coupled model through modications in the
subgridscale orographic parameterization. Journal of Advances in Modeling Earth Systems,12, e2020MS002111. https://doi.org/10.1029/
2020MS002111
Genthon, C., Six, D., Gallée, H., Grigioni, P., & Pellegrini, A. (2013). Two years of atmospheric boundary layer obser vations on a 45m tower
at Dome C on the Antarctic Plateau. Journal of Geophysical Research: Atmospheres,118, 32183232. https://doi.org/10.1002/jgrd.50128
Gouttevin, I., Krinner, G., Ciais, P., Polcher, J., & Legout, C. (2012). Multiscale validation of a new soil freezing scheme for a landsurface
model with physicallybased hydrology. The Cryosphere,6(2), 407430. https://doi.org/10.5194/tc64072012
Guillod, B., Orlowsky, B., Miralles, D., Teuling, A. J., & Seneviratne, S. I. (2015). Reconciling spatial and temporal soil moisture effects on
afternoon rainfall. Nature Communication,6, 6443. https://doi.org/10.1038/ncomms7443
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 30 of 33
Guimberteau, M., Drapeau, G., Ronchail, J., Sultan, B., Polcher, J., Martinez, J. M., et al. (2012). Discharge simulation in the subbasins of
the Amazon using ORCHIDEE forced by new datasets. Hydrology and Earth System Sciences,16(3), 911935. https://doi.org/10.5194/
hess169112012
Guimberteau, M., Ducharne, A., Ciais, P., Boisier, J. P., Peng, S., De Weirdt, M., & Verbeeck, H. (2014). Testing conceptual and physically
based soil hydrology schemes against observations for the Amazon Basin. Geoscientic Model Development,7(3), 11151136. https://doi.
org/10.5194/gmd711152014
Guimberteau, M., Laval, K., Perrier, A., & Polcher, J. (2012). Global effect of irrigiation and its impact on the onset of the indian summer
monsoon. Climate Dynamics,39, 1329. https://doi.org/10.1007/s0038201112525
Guimberteau, M., Zhu, D., Maignan, F., Huang, Y., Yue, C., DantecNédélec, S., et al. (2018). ORCHIDEEMICT (v8. 4.1), a land surface
model for the high latitudes: model description and validation. Geoscientic Model Development,11(1), 121163. https://doi.org/10.5194/
gmd111212018
Haarsma, R. J., Roberts, M. J., Vidale, P. L., Senior, C. A., Bellucci, A., Bao, Q., et al. (2016). High Resolution Model Intercomparison Project
(HighResMIP v1.0) for CMIP6. Geoscientic Model Development,9(11), 41854208. https://doi.org/10.5194/gmd941852016
Harris, I., Jones, P., Osborn, T., & Lister, D. (2014). Updated highresolution grids of monthly climatic observations the CRU TS3.10
Dataset. International Journal of Climatology,34(3), 623642. https://doi.org/10.1002/joc.3711
Hess, R. (1995). On computing screen temperatures, humidities and anemometerheight winds in largescale models. Australian
Meteorological Magazine,44, 139145.
Hohenegger, C., Brockhaus, P., Bretherton, C., & Schär, C. (2009). The soil moisture precipitation feedback in simulations with explicit and
parameterized convection. Journal of Climate,22(19), 50035020. https://doi.org/10.1175/2009JCLI2604.1
Hohenegger, C., & Stevens, B. (2018). The role of the permanent wilting point in controlling the spatial distribution of precipitation.
Proceedings of the National Academy of Sciences,11,56925697. https://doi.org/10.1073/pnas.1718842115
Holton, J. R. (2004). An Introduction to Dynamic Meteorology. Academic Press.
Hourdin, F. (1992). Etude et simulation numérique de la circulation générale des atmosphéres planétaires (in french) (PhD thesis). Paris,
France: Laboratoire de Météorologie Dynamique.
Hourdin, F., Couvreux, F., & Menut, L. (2002). Parameterization of the dry convective boundary layer based on a mass ux representation
of thermals. Journal of the Atmospheric Sciences,59, 11051123.
Hourdin, F., Grandpeix, J. Y., Rio, C., Bony, S., Jam, A., Cheruy, F., et al. (2013). LMDZ5B: The atmospheric component of the IPSL climate
model with revisited parameterizations for clouds and convection. Climate Dynamics,40(9), 21932222. https://doi.org/10.1007/s00382
0121343y
Hourdin, F., Jam, C., Rio, Couvreux, Sandu, I., Lefebvre, M. P., Brient, F., & Idelkadi, A. (2019). Unied parameterization of convective
boundary layer transport and clouds with the thermal plume model. Journal of Advances in Modeling Earth Systems,11, 29102933.
https://doi.org/10.1029/2019MS001666
Hourdin, F., Mauritsen, T., Gettelman, A., Golaz, J. C., Balaji, V., Duan, Q., et al. (2017). The art and science of climate model tuning.
Bulletin of the American Meteorological Society,98(3), 589602. https://doi.org/10.1175/BAMSD1500135.1
Hourdin, F., Musat, I., Bony, S., Braconnot, P., Codron, F., Dufresne, J. L., et al. (2006). The LMDZ4 general circulation model: Climate
performance and sensitivity to parametrized physics with emphasis on tropical convection. Climate Dynamics,27, 787813. https://doi.
org/10.1007/s0038200601580
Hourdin, F., Rio, C., Grandpeix, J., Madeleine, J., Cheruy, F., Rochetin, N., et al. (2020). LMDZ6A: The atmospheric component of the IPSL
climate model with improved and better tuned physics. Journal of Advances in Modeling Earth Systems,12, e2019MS001892. https://doi.
org/10.1029/2019MS001892
Hurtt, G. C., Chini, L. P., Frolking, S., Betts, R. A., Feddema, J., Fischer, G., et al. (2011). Harmonization of landuse scenarios for the period
15002100: 600 years of global gridded annual landuse transitions, wood harvest, and resulting secondary lands. Climatic Change,
109(107). https://doi.org/10.1007/s1058401101532
Jaeger, E., & Seneviratne, S. (2011). Impact of soil moisture atmosphere coupling on European climate extremes and trends in a regional
climate model. Climate Dynamics,36,19191939.
Jakobson, E., Vihma, T., Palo, T., Jakobson, L., Keernik, H., & Jaagu s, J. (2012). Validation of atmospheric reanalyses over the central Arctic
Ocean. Geophysical Research Letters,39, L10802. https://doi.org/10.1029/2012GL051591
Jung, M., Reichstein, M., Margolis, H. A., Cescatti, A., Richardson, A. D., Arain, M. A., et al. (2011). Global patterns of landatmosphere
uxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations.
J. Geophys. Res.,116, G00J07. https://doi.org/10.1029/2010JG001566
Kato, S., Loeb, N. G., Rose, F. G., Doelling, D. R., Rutan, D. A., Caldwell, T. E., et al. (2013). Surface irradiances consistent with
CERESderived topofatmosphere shortwave and longwave irradiances. Journal of Climate,26, 27192740. https://doi.org/10.1175/
JCLID1200436.1
King, J. C., Connolley, W. M., & Derbyshire, S. H. (2001). Sensitivity of modelled Antarctic climate to surface and boundary layer ux
parametrizations. Quarterly Journal of the Royal Meteorological Society,127(573), 779794. https://doi.org/10.1002/qj.49712757304
Klein, S. A., Jiang, X., Boyle, J., Malyshev, S., & Xie, S. (2006). Diagnosis of the summertime warm and dry bias over the U.S. Southern Great
Plains in the GFDL climate model using a weather forecasting approach. Geophysical Research Letters,33, L18805. https://doi.org/
10.1029/2006GL027567
Koster, R., Dirmeyer, P., Guo, Z., Bonan, G., Chan, E., Cox, P., et al. (2004). Regions of strong coupling between soil moisture and preci-
pitation. Science,305(5687), 1138.
Koster, R. D., Guo, Z., Yang, R., Dirmeyer, P. A., Mitchell, K., & Puma, M. J. (2009). On the nature of soil moisture in land surface models.
Journal of Climate,22(16), 43224335. https://doi.org/10.1175/2009JCLI2832.1
Kottek, M., Grieser, J., Beck, C., Rudolf, B., & Rubel, F. (2006). World Map of the KöppenGeiger climate classication updated.
Meteorologische Zeitschrift,15, 259263. https://doi.org/10.1127/09412948/2006/0130
Krishnan, R., Sabin, T. P., Madhura, R. K., Vellore, R. K., Mujumdar, M., Sanjay, J., et al. (2019). Nonmonsoonal precipitation response
over the Western Himalayas to climate change. Climate Dynamics,52(7), 40914109. https://doi.org/10.1007/s0038201843572
Laval, K. (1988). Experience with surface processes at LMD. Workshop on Parametrization of Fluxes over Land Surface. Shineld Park,
Reading: ECMWF. https://www.ecmwf.int/node/10662
Le Quéré, C., Andrew, R. M., Friedlingstein, P., Sitch, S., Hauck, J., Pongratz, J., et al. (2018). Global carbon budget 2018. Earth System
Science Data,10(4), 21412194.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 31 of 33
Li, S., Rupp, D. E., Hawkins, L., Mote, P. W., McNeall, D., Sparrow, S. N., et al. (2019). Reducing climate model biases by exploring para-
meter space with large ensembles of climate model simulations and statistical emulation. Geoscientic Model Development,12(7),
30173043. https://doi.org/10.5194/gmd1230172019
Lindsay, R., Wensnahan, M., Schweiger, A., & Zhang, J. (2014). Evaluation of seven different atmospheric reanalysis products in the Arctic.
Journal of Climate,27(7), 25882606. https://doi.org/10.1175/JCLID1300014.1
Lott, F. (1998). Linear mountain drag and averaged pseudomomentum ux proles in the presence of trapped lee waves. Tellus,50A,
1225.
Lott, F. (1999). Alleviation of stationary biases in a GCM through a mountain drag parametrization scheme and a simple representation of
mountain lift forces. Monthly Weather Review,127, 788800.
Lott, F., & Miller, M. J. (1997). A new subgridscale orographic drag parametrization: Its formulation and testing. Quarterly Journal of the
Royal Meteorological Society,123, 101127.
Louis, J. F., Tiedtke, M., & Geleyn, J. F. (1982). A short history of the PBL parameterization at ECMWF, Workshop on planetary boundary
layer parameterization, 2527 november 1981 (pp. 5979). Shineld Park, Reading: ECMWF.
Lurton, T., Balkanski, Y., Bastrikov, V., Bekki, S., Bopp, L., Braconnot, P., et al. (2020). Implementation of the CMIP6 Forcing Data in the
IPSLCM6ALR Model. Journal of Advances in Modeling Earth Systems,12, e2019MS001940. https://doi.org/10.1029/2019MS001940
Malhi, Y. (1996). The behaviour of the roughness length for temperature over heterogeneous surfaces. Quarterly Journal of the Royal
Meteorological Society,533, 10951125. https://doi.org/10.1002/qj.49712253305
Manabe, S. (1969). Climate and the ocean circulation I. The atmospheric circulation and the hydrology of the Earth's surface. Monthly
Weather Review,97(11), 739774.
Martens, B., Miralles, D. G., Lievens, H., van der Schalie, R., de Jeu, R. A. M., FernándezPrieto, D., et al. (2017). GLEAM v3: satellitebased
land evaporation and rootzone soil moisture. Geoscientic Model Development,10(5), 19031925. https://doi.org/10.5194/gmd101903
2017
Massman, W. (1999). A model study of kBH1 for vegetated surfaces using localized neareldLagrangian theory. Journal of Hydrology,
223(1), 2743. https://doi.org/10.1016/S00221694(99)001043
Masson, V., & Seity, Y. (2009). Including atmospheric layers in vegetation and urban ofine surface schemes. Journal of Applied
Meteorology and Climatology,48(7), 13771397. https://doi.org/10.1175/2009JAMC1866.1
Ménégoz, M., Krinner, G., Balkanski, Y., Boucher, O., Cozic, A., Lim, S., et al. (2014). Snow cover sensitivity to black carbon deposition in
the Himalayas: From atmospheric and ice core measurements to regional climate simulations. Atmospheric Chemistry and Physics ,14(8),
42374249. https://doi.org/10.5194/acp1442372014
Milliman, J. D., & Farnsworth, K. L. (2011). River discharge to the coastal ocean: A global synthe sis: Cambridge University Press. https://doi.
org/10.1017/CBO9780511781247
Miralles, D. G., Teuling, A. J., van Heerwaarden, C. C., & ViláGuerau de Arellano, J. (2014). Megaheatwave temperatures due to com-
bined soil desiccation and atmospheric heat accumulation. Nature Geoscience,7, 756761. https://doi.org/10.1038/ngeo2141
Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., & Clough, S. A. (1997). Radiative transfer for inhomogeneous atmospheres:
RRTM, a validated correlatedk model for the longwave. Journal of Geophysical Research,102(D14), 16,66316,682. https://doi.org/
10.1029/97JD00237
Morcrette, J. J. (1991). Radiation and cloud radiative properties in the European Centre for Medium Range Weather Forecasts forecasting
system. Journal of Geophysical Research,96(D5), 91219132. https://doi.org/10.1029/89JD01597
Morcrette, C. J., Van Weverberg, K., Ma, H. Y., Ahlgrimm, M., Bazile, E., Berg, L. K., et al. (2018). Introduction to CAUSES: Description of
weather and climate models and their nearsurface temperature errors in 5day hindcasts near the Southern Great Plains. Journal of
Geophysical Research: Atmospheres,123, 26552683. https://doi.org/10.1002/2017JD027199
Naudts, K., Chen, Y., McGrath, M. J., Ryder, J., Valade, A., Otto, J., & Luyssaert, S. (2016). Europe's forest management did not mitigate
climate warming. Science,351(6273), 597600.
Naudts, K., Ryder, J., McGrath, M., Otto, J., Chen, Y., Valade, A., et al. (2015). A vertically discretised canopy description for ORCHIDEE
(SVN r2290) and the modications to the energy, water and carbon uxes. Geoscientic Model Development,8, 20352065.
Nepf, H. M. (1999). Drag, turbulence, and diffusion in ow through emergent vegetation. Water Resources Research ,35(2), 479489. https://
doi.org/10.1029/1998WR900069
Peterson, B. J., Holmes, R. M., McClelland, J. W., Vörösmarty, C. J., Lammers, R. B., Shiklomanov, A. I., et al. (2002). Increasing river
discharge to the Arctic Ocean. Science,298(5601), 21712173. https://doi.org/10.1126/science.1077445
Polcher, J. (2003). Les processus de surface à l'échelle globale et leur interaction avec l'atmosphère (in french), Habilitation à diriger des
recherches thesis, Université Pierre et MarieCurie (Paris, France).
Puma, M., & Cook, B. (2010). Effects of irrigation on global climate during the 20th century. Journal of Geophysical Research,115, D16120.
https://doi.org/10.1029/2010JD014122
Reeves Eyre, J. E. J., & Zeng, X. (2017). Evaluation of Greenland near surface air temperature datasets. The Cryosphere,11(4), 15911605.
https://doi.org/10.5194/tc1115912017
Rio, C., Hourdin, F., Couvreux, F., & Jam, A. (2010). Resolved versus parametrized boundarylayer plumes. Part II: continuous formula-
tions of mixing rates for massux schemes. BoundaryLayer Meteorology,135, 469483. https://doi.org/10.1007/s105460109478
Rio, C., Hourdin, F., Grandpeix, J.Y., & Lafore, J.P. (2009). Shifting the diurnal cycle of parametrized deep convection over land.
Geophysical Research Letters,36, L07809. https://doi.org/10.1029/2008GL036779
Robinson, D. A., Estilow, T. W., & Program, N. C. (2012). NOAA Climate Data Record (CDR) of Northern Hemisphere (NH) Snow Cover
Extent (SCE), Version 1, 20190703T18:47:12Z. https://doi.org/10.7289/V5N014G9
Rochetin, N., Couvreux, F., Grandpeix, J. Y., & Rio, C. (2014). Deep convection triggering by boundary layer thermals. Part I: LES analysis
and stochastic triggering formulation. Journal of the Atmospheric Sciences,71(2), 496514. https://doi.org/10.1175/JASD120336.1
Rochetin, N., Grandpeix, J. Y., Rio, C., & Couvreux, F. (2014). Deep Convection Triggering by Boundary Layer Thermals. Part II: Stochastic
Triggering Parameterization for the LMDZ GCM. Journal of the Atmospheric Sciences,71(2), 515538. https://doi.org/10.1175 /JASD12
0337.1
Roehrig, R., Bouniol, D., Guichard, F., Hourdin, F., & Redelsperger, J. L. (2013). The present and future of the west african monsoon: A
processoriented assessment of CMIP5 simulations along the AMMA Transect. Journal of Climate,26(17), 64716505. https://doi.org/
10.1175/JCLID1200505.1
Sabin, T. P., Krishnan, R., Ghattas, J., Denvil, S., Dufresne, J. L., Hourdin, F., & Pascal, T. (2013). High resolution simulation of the South
Asian monsoon using a variable resolution global climate model. Climate Dynamics,41(1), 17 3194. https://doi.org/10.1007/s00382012
16588
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 32 of 33
Sandu, I., Beljaars, A., Bechtold, P., Mauritsen, T., & Balsamo, G. (2013). Why is it so difcult to represent stably stratied conditions in
numerical weather prediction (NWP) models? Journal of Advances in Modeling Earth Systems,5, 117133. https://doi.org/10.1002/
jame.20013
Sandu, I., van Niekerk, A., Shepherd, T. G., Vosper, S. B., Zadra, A., Bacmeister, J., et al. (2019). Impacts of orography on largescale
atmospheric circulation. Npj Climate and Atmospheric Science,2,18. https://doi.org/10.1038/s4161201900659
Schär, C., Lüthi, D., Beyerle, U., & Heise, E. (1999). The soilprecipitation feedback: A process study with a regional climate model. Journal
of Climate,12, 722741.
Schewe, J., Heinke, J., Gerten, D., Haddeland, I., Arnell, N. W., Clark, D. B., et al. (2014). Multimodel assessment of water scarcity under
climate change. Proceedings of the National Academy of Sciences,111(9), 32453250. https://doi.org/10.1073/pnas.1222460110
Sellers, P. J., Mintz, Y., Sud, Y. C., & Dalcher, A. (1986). A simple biosphere model (SIB) for use within general circulation models. Journal
of the Atmospheric Sciences,43(6), 505531. https://doi.org/10.1175/15200469(1986)043<0505:ASBMFU>2.0.CO;2
Seneviratne, S. I., Corti, T., Davin, E. L., Hirschi, M., Jaeger, E. B., Lehner, I., et al. (2010). Investigating soil moisture climate interactions in
a changing climate: A review. EarthScience Reviews,99(34), 125161. https://doi.org/10.1016/j.earscirev.2010.02.004
Seneviratne, S. I., Nicolls, N., Easterling, D., Goodess, C. M., Kanae, S., Kossin, J., et al. (2012). Changes in climate extremes and their
impacts on the naturalphysical environment. In C. B. Field et al. (Eds.), Managing the risks of extreme events and disasters to advance
climate change adaptation,A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change (IPCC)
(pp. 109230). Cambridge, UK and New York, NY: Cambridge University Press.
Steeneveld, G. J., Holtslag, A. A. M., Nappo, C. J., van de Wiel, B. J. H., & Mahrt, L. (2008). Exploring the possible role of smallscale terrain
drag on stable boundary layers over land. Journal of Applied Meteorology and Climatology,47(10), 25182530. https://doi.org/10.1175/
2008JAMC1816.1
Stouffer, R. J., Eyring, V., Meehl, G. A., Bony, S., Senior, C., Ste vens, B., & Taylor, K. E. (2017). CMIP5 scientic gaps and recommendations
for CMIP6. Bulletin of the American Meteorological Society,98(1), 95105. https://doi.org/10.1175/BAMSD1500013.1
Stull, R. B. (1990). An Introduction to Boundary Layer Meteorology: Kluver, Boston.
Su, Z., Schmugge, T., Kustas, W. P., & Massman, W. J. (2001). An evaluation of two models for estimation of the roughness height for heat
transfer between the land surface and the atmosphere. Journal of Applied Meteorology,40(11), 19331951. https://doi.org/10.1175/1520
0450(2001)040<1933:AEOTMF>2.0.CO;2
Taylor, C. M., de Jeu, R. A. M., Guichard, F., Harris, P. P., & Dorigo, W. A. (2012). Afternoon rain more likely over drier soils. Nature,4,
423426. https://doi.org/10.1038/nature11377
Taylor, C. M., Gounou, A., Guichard, F., Harris, P. P., Ellis, R. J., Couvreux, F., & de Kauwe, M.(2011). Frequency of Sahelian storm
initiation enhanced over mesoscale soilmoisture patterns. Nature Geoscience,4,430433. https://doi.org/10.1038/ngeo1173
Van Weverberg, K., Morcrette, C. J., Petch, J., Klein, S. A., Ma, H. Y., Zhang, C., et al. (2018). CAUSES: Attribution of surface radiation
biases in NWP and climate models near the U.S. Southern Great Plains. Journal of Geophysical Research: Atmospheres,123, 36123644.
https://doi.org/10.1002/2017JD027188
Vignon, E. (2017). The extreme atmospheric boundary layer over the Antarctic Plateau and its representation in climate models (PhD
thesis). Grenoble, France: Université GrenobleAlpes.
Vignon, E., Hourdin, F., Genthon, C., Gallée, H., Bazile, E., Lefebvre, M. P., & Madeleine, J.B. (2017). Antarctic boundary layer parame-
trization in a general circulation model: 1D simulations facing summer observations at Dome C. Journal of Geophysical Research:
Atmospheres,122, 68186843. https://doi.org/10.1002/2017JD026802
Vignon, E., Hourdin, F., Genthon, C., Gallée, H., Bazile, E., Lefebvre, M. P., et al. (2017). Parametrization of the boundary layer over the
Antarctic Plateau in a general circulation model: 1D simulations against summertime observations at Dome C. Journal of Geophysical
Research: Atmospheres,122, 68186843. https://doi.org/10.1002/2017JD026802
Vignon, E., Hourdin, F., Genthon, C., Van de Wiel, B. J. H., Gallée, H., Madeleine, J. B., & Beaumet, J. (2018). Modeling the dynamics of the
atmospheric boundary layer over the antarctic plateau with a general circulation model. Journal of Advances in Model Earth Syste ms,10,
98125. https://doi.org/10.1002/2017MS001184
Vignon, E., van de Wiel, B. J. H., van Hooijdonk, I. G. S., Genthon, C., van der Linden, S. J. A., van Hooft, J. A., et al. (2017). Stable boundary
layer regimes at dome c, antarctica: Observation and analysis. Quarterly Journal of the Royal Meteorological Society,143(704), 12411253.
https://doi.org/10.1002/qj.2998
Vonder Haar, T. H., Bytheway, J. L., & Forsythe, J. M. (2012). Weather and climate analyses using improved global water vapor observa-
tions. Geophysical Research Letters,39, L15802. https://doi.org/10.1029/2012GL052094
Vuichard, N., Messina, P., Luyssaert, S., Guenet, B., Zaehle, S., Ghattas, J., et al. (2019). Accounting for carbon and nitrogen interactions in
the global terrestrial ecosystem model ORCHIDEE (trunk version, rev 4999): Multiscale evaluation of gross primary production.
Geoscientic Model Development,12(11), 47514779.
Wang, F., Cheruy, F., & Dufresne, J. L. (2016). The improvement of soil thermodynamics and its effects on land surface meteorology in the
IPSL climate model. Geoscientic Model Development,9(1), 363381. https://doi.org/10.5194/gmd93632016
Wang, F., Ducharne, A., Cheruy, F., Lo, M. H., & Grandpeix, J. Y. (2018). Impact of a shallow groundwater table on the global water cycle in
the IPSL landatmosphere coupled model. Climate Dynamics,50, 35053522. https://doi.org/10.1007/s0038201738209
Wang, T., Ottlé, C., Boone, A., Ciais, P., Brun, E., Morin, S., et al. (2013). Evaluation of an improved intermediate comp lexity snow scheme
in the ORCHIDEE land surface model. Journal of Geophysical Research: Atmospheres,118, 60646079. https://doi.org/10.1002/
jgrd.50395
Wei, N., Zhou, L., & Dai, Y. (2017). Evaluation of simulated climatological diurnal temperature range in CMIP5 models from the per-
spective of planetary boundary layer turbulent mixing. Climate Dynamics,49,122. https://doi.org/10.1007/s0038201633230
Williams, C. J. R., Allan, R. P., & Kniveton, D. R. (2012). Diagnosing atmosphereland feedbacks in CMIP5 climate models. Environmental
Research Letters,7(4), 044003. https://doi.org/10.1088/17489326/7/4/044003
Xie, S., McCoy, R. B., Klein, S. A., Cederwall, R. T., Wiscombe, W. J., Clothiaux, E. E., et al. (2010). Clouds and more: ARM climate
modeling best estimate data: A new data product forclimate studies. Bulletin of the American Meteorological Society,91,1320.
Yamada, T. (1983). Simulations of nocturnal drainage ows by a q
2
lTurbulence Closure Model. Journal of the Atmospheric Sciences,40,
91106.
Zeng, Z., Piao, S., Li, L. Z., Zhou, L., Ciais, P., Wang, T., et al. (2017). Climate mitigation from vegetation biophysical feedbacks during the
past three decades. Nature Climate Change,7(6), 432436.
Zobler, L. (1986). A World Soil File for Global Climate Modeling: National Aeronautics and Space Administration, Goddard Space Flight
Center, Institute for Space Studies.
10.1029/2019MS002005
Journal of Advances in Modeling Earth Systems
CHERUY ET AL. 33 of 33
... while the other includes a novel subgrid hillslope flow parameterization (Section 2.1.2). As described in Section 2.2, we use LMDZOR, the coupled land-atmosphere component of the IPSL-CM6 climate model from the Institut Pierre Simon Laplace (Boucher et al., 2020;Cheruy et al., 2020). First, we focus the analysis on the historical period to explore the effects of hillslope flow on yearly and seasonal average values in terms of both sensitivity and realism against observations (Section 3.1). ...
... ) is a process-based model that describes the fluxes of mass, momentum, and heat between the surface and the atmosphere (Krinner et al., 2005). The version used here as a reference (called ORCHIDEE-REF in the following) corresponds to version 2.0, included in the IPSL-CM6 climate model for CMIP6 simulations and described in Cheruy et al. (2020), Boucher et al. (2020), and Tafasca et al. (2020). Here, we summarize the main characteristics of the model. ...
Article
Full-text available
We analyzed the influence of hillslope flow on projections of climate change by comparing two transient climate simulations with the IPSL climate model between 1980 and 2100. Hillslope flow induces a reorganization and increment of soil moisture (+10%), which increases evapotranspiration (+4%) and precipitation (+1%) and decreases total runoff (−3%) and air temperature (−0.1 °C) on an annual average over land for 1980–2010 when compared to simulation not representing hillslope flow. These changes in land/atmosphere fluxes are not homogenous and depend on regional climate and surface conditions. Hillslope flow also influences climate change projections. On average over land, it amplifies the positive trend of soil moisture (+23%), evapotranspiration (+50%), and precipitation (+7%) and slightly attenuates global warming (−1%), especially for daily maximum air temperature. The role of hillslope flow in supporting surface/atmosphere fluxes is more evident at a regional scale. Where precipitation is projected to decrease, hillslope flow is shown to attenuate the related declines in evapotranspiration, precipitation, and total runoff, regardless of aridity conditions and mean air temperature. Where precipitation is projected to increase, hillslope flow amplifies evapotranspiration enhancement but attenuates the increase in precipitation and total runoff. Warming is generally attenuated, especially in semiarid and cold areas, and humid and warm/temperate regions, but the signal is weak. These results demonstrate the role of hillslope flow in enhancing water and energy fluxes between the surface and the atmosphere. They also suggest that including hillslope flow in climate models would weaken the projected intensification of hydrological extreme events.
... albedo). This darkening has already been observed over the 2 Methods and Data 2.1 ORCHIDEE land surface model 55 The ORCHIDEE (ORganizing Carbon and Hydrology in Dynamic Ecosystems) land surface model is the terrestrial component of the IPSL Earth system model (ESM) used in climate projections (Boucher et al., 2020;Cheruy et al., 2020). Either run off-line (i.e., driven by prescribed meteorological forcing) or coupled with an atmospheric model (i.e., as part of the ESM), ORCHIDEE describes the exchanges of energy, water, and carbon between the atmosphere and the continental biosphere. ...
Preprint
Full-text available
Greenland ice sheet mass loss continues to accelerate as global temperatures increase. The surface albedo of the ice sheet determines the amount of absorbed solar energy, which is a key factor in driving surface snow and ice melting. Satellite retrieved albedo allows us to compare and optimise modelled albedo over the entirety of the ice sheet. We optimise the parameters of the albedo scheme in the ORCHIDEE land surface model for three random years taken over the 2000–2017 period and validate over the remaining years. In particular, we want to improve the albedo at the edges of the ice sheet since they correspond to ablation areas and show the greatest variations in runoff and surface mass balance. By giving a larger weight to points at the ice sheet's edge, we improve the model-data fit by reducing the RMSD by over 25 % for the whole ice sheet for the summer months. This improvement is consistent for all years, even those not used in the calibration step. We conclude by showing which additional model outputs are impacted by changes to the albedo parameters encouraging future work using multiple data streams for optimisation.
Article
Full-text available
The Budyko framework consists of a curvilinear relationship between the evaporative ratio (i.e., actual evaporation over precipitation) and the aridity index (i.e., potential evaporation over precipitation) and defines evaporation's water and energy limits. A basin's movement within the Budyko space illustrates its hydroclimatic change and helps identify the main drivers of change. On the one hand, long‐term aridity changes drive evaporative ratio changes, moving basins along their Budyko curves. On the other hand, historical human development can cause river basins to deviate from their curves. The question is if basins will deviate or follow their Budyko curves under the future effects of global warming and related human developments. To answer this, we quantify the movement in the Budyko space of 405 river basins from 1901–1950 to 2051–2100 based on the outputs of seven models from the Coupled Model Intercomparison Project ‐ Phase 6 (CMIP6). We account for the implications of using different potential evaporation models and study low‐ and high‐emissions scenarios. We find considerable differences of movement in Budyko space regarding direction and intensity when using the two estimates of potential evaporation. However, regardless of the potential evaporation estimate and the scenario used, most river basins will not follow their reference Budyko curves (>72%). Furthermore, the number of basins not following their curves increases under high greenhouse gas emissions and fossil‐fueled development SP585 and across dry and wet basin groups. We elaborate on the possible explanations for a large number of basins not following their Budyko curves.
Article
Full-text available
The modulation of the winter impacts of Arctic sea-ice loss by the Pacific Decadal Oscillation (PDO) is investigated in the IPSL-CM6A-LR ocean–atmosphere general circulation model. Ensembles of simulations are performed with constrained sea-ice concentration following the Polar Amplification Model Intercomparison Project (PAMIP) and initial conditions sampling warm and cold phases of the PDO. Using a general linear model, we estimate the simulated winter impact of sea-ice loss, PDO and their combined effects. On the one hand, a negative North Atlantic Oscillation (NAO)-like pattern appears in response to sea-ice loss together with a modest deepening of the Aleutian Low. On the other hand, a warm PDO phase induces a large positive Pacific–North America pattern, as well as a small negative Arctic Oscillation pattern. Both sea-ice loss and warm PDO responses are associated with a weakening of the poleward flank of the eddy-driven jet, an intensification of the subtropical jet and a weakening of the stratospheric polar vortex. These effects are partly additive; the warm PDO phase therefore enhances the response to sea-ice loss, while the cold PDO phase reduces it. However, the effects of PDO and sea-ice loss are also partly non-additive, with the interaction between both signals being slightly destructive. This results in small damping of the PDO teleconnections under sea-ice loss conditions, especially in the stratosphere. The sea-ice loss responses are compared to those obtained with the same model in atmosphere-only simulations, where sea-ice loss does not significantly alter the stratospheric polar vortex.
Article
Full-text available
The Antarctic plateau, characterized by cold and dry weather conditions with very little precipitation, is mostly covered by snow at the surface. This paper describes an intercomparison of snow models, of varying complexity, used for numerical weather prediction or academic research. The results of offline numerical simulations, carried out during 15 d in 2009, on a single site on the Antarctic plateau, show that the simplest models are able to reproduce the surface temperature as well as the most complex models provided that their surface parameters are well chosen. Furthermore , it is shown that the diversity of the surface parameters of the models strongly impacts the numerical simulations, in particular the temporal variability of the surface temperature and the components of the surface energy balance. The models tend to overestimate the surface temperature by 2-5 K at night and underestimate it by 2 K during the day. The observed and simulated turbulent latent heat fluxes are small, of the order of a few W m −2 , with a tendency to underestimate , while the sensible heat fluxes are in general too intense at night as well as during the day. The surface temperature errors are consistent with too large a magnitude of sensible heat fluxes during the day and night. Finally, it is shown that the most complex multilayer models are able to reproduce well the propagation of the daily diurnal wave, and that the snow temperature profiles in the snowpack are very close to the measurements carried out on site.
Article
Full-text available
The freshwater 1-D FLake lake model was coupled to the ORCHIDEE land surface model to simulate lake energy balance at the global scale. A multi-tile approach has been chosen to allow the modeling of various types of lakes within the ORCHIDEE grid cell. Thus, three different lake tiles have been defined according to lake depth which is the most influential parameter of FLake, but other properties could be considered in the future. Several depth parameterization strategies have been compared, differing by the way to aggregate the depth of the subgrid lakes, i.e., arithmetical, geometrical, harmonical mean and median. Five atmospheric reanalysis datasets available at 0.5∘ or 0.25∘ resolution have been used to force the model and assess model systematic errors. Simulations have been performed, evaluated and intercompared against observations of lake water surface temperatures provided by the GloboLakes database over about 1000 lakes and ice phenology derived from the Global Lake and River Ice Phenology database. The results highlighted the large impact of the atmospheric forcing on the lake energy budget simulations and the improvements brought by the highest resolution products (ERA5 and E2OFD). The median of the root square mean errors (RMSEs) calculated at global scale ranges between 3.2 and 2.7 ∘C among the forcings, CRUJRA and ERA5 leading respectively to the worst and best results. The depth parameterization strategy appeared to be less influential, with RMSE differences less than 0.1 ∘C for the four aggregation scenarios tested. The simulation of ice phenology presented systematic errors whatever the forcing and the depth parameterization used. Large systematic errors were highlighted such as negative biases on the onset and positive biases on the offset. Freezing onset was shown to be the less sensitive to atmospheric forcing with the median of the errors ranging between 10 and 14 d. Larger errors up to 25 d were observed on the simulation of the end of the freezing period. Such errors, already highlighted in previous works, could be explained by scale effects and deficiencies in the modeling of snow–ice processes not accounting for partial ice cover. Various pathways are drawn to improve the model results, including the use of remote sensing data to better constrain the lake radiative parameters (albedo and extinction coefficient) as well as the lake depth thanks to the recent and forthcoming high-resolution satellite missions.
Article
Full-text available
Carbonyl sulfide (COS) is an atmospheric trace gas of interest for C cycle research because COS uptake by continental vegetation is strongly related to terrestrial gross primary productivity (GPP), the largest and most uncertain flux in atmospheric CO2 budgets. However, to use atmospheric COS as an additional tracer of GPP, an accurate quantification of COS exchange by soils is also needed. At present, the atmospheric COS budget is unbalanced globally, with total COS flux estimates from oxic and anoxic soils that vary between −409 and −89 GgS yr−1. This uncertainty hampers the use of atmospheric COS concentrations to constrain GPP estimates through atmospheric transport inversions. In this study we implemented a mechanistic soil COS model in the ORCHIDEE (Organising Carbon and Hydrology In Dynamic Ecosystems) land surface model to simulate COS fluxes in oxic and anoxic soils. Evaluation of the model against flux measurements at seven sites yields a mean root mean square deviation of 1.6 pmol m−2 s−1, instead of 2 pmol m−2 s−1 when using a previous empirical approach that links soil COS uptake to soil heterotrophic respiration. However, soil COS model evaluation is still limited by the scarcity of observation sites and long-term measurement periods, with all sites located in a latitudinal band between 39 and 62∘ N and no observations during wintertime in this study. The new model predicts that, globally and over the 2009–2016 period, oxic soils act as a net uptake of −126 GgS yr−1 and anoxic soils are a source of +96 GgS yr−1, leading to a global net soil sink of only −30 GgS yr−1, i.e. much smaller than previous estimates. The small magnitude of the soil fluxes suggests that the error in the COS budget is dominated by the much larger fluxes from plants, oceans, and industrial activities. The predicted spatial distribution of soil COS fluxes, with large emissions from oxic (up to 68.2 pmol COS m−2 s−1) and anoxic (up to 36.8 pmol COS m−2 s−1) soils in the tropics, especially in India and in the Sahel region, marginally improves the latitudinal gradient of atmospheric COS concentrations, after transport by the LMDZ (Laboratoire de Météorologie Dynamique) atmospheric transport model. The impact of different soil COS flux representations on the latitudinal gradient of the atmospheric COS concentrations is strongest in the Northern Hemisphere. We also implemented spatiotemporal variations in near-ground atmospheric COS concentrations in the modelling of biospheric COS fluxes, which helped reduce the imbalance of the atmospheric COS budget by lowering soil COS uptake by 10 % and plant COS uptake by 8 % globally (with a revised mean vegetation budget of −576 GgS yr−1 over 2009–2016). Sensitivity analyses highlighted the different parameters to which each soil COS flux model is the most responsive, selected in a parameter optimization framework. Having both vegetation and soil COS fluxes modelled within ORCHIDEE opens the way for using observed ecosystem COS fluxes and larger-scale atmospheric COS mixing ratios to improve the simulated GPP, through data assimilation techniques.
Article
Full-text available
The Last Interglacial period (129–116 ka) is characterised by a strong orbital forcing which leads to a different seasonal and latitudinal distribution of insolation compared to the pre-industrial period. In particular, these changes amplify the seasonality of the insolation in the high latitudes of the Northern Hemisphere. Here, we investigate the Arctic climate response to this forcing by comparing the CMIP6 lig127k and piControl simulations performed with the IPSL-CM6A-LR (the global climate model developed at Institut Pierre-Simon Laplace) model. Using an energy budget framework, we analyse the interactions between the atmosphere, ocean, sea ice and continents. In summer, the insolation anomaly reaches its maximum and causes a rise in near-surface air temperature of 3.1 ∘C over the Arctic region. This warming is primarily due to a strong positive anomaly of surface downwelling shortwave radiation over continental surfaces, followed by large heat transfer from the continents to the atmosphere. The surface layers of the Arctic Ocean also receive more energy but in smaller quantity than the continents due to a cloud negative feedback. Furthermore, while heat exchange from the continental surfaces towards the atmosphere is strengthened, the ocean absorbs and stores the heat excess due to a decline in sea ice cover. However, the maximum near-surface air temperature anomaly does not peak in summer like insolation but occurs in autumn with a temperature increase of 4.2 ∘C relative to the pre-industrial period. This strong warming is driven by a positive anomaly of longwave radiation over the Arctic Ocean enhanced by a positive cloud feedback. It is also favoured by the summer and autumn Arctic sea ice retreat (-1.9×106 and -3.4×106 km2, respectively), which exposes the warm oceanic surface and thus allows oceanic heat storage and release of water vapour in summer. This study highlights the crucial role of sea ice cover variations, Arctic Ocean, as well as changes in polar cloud optical properties on the Last Interglacial Arctic warming.
Preprint
Full-text available
The modulation of the winter impacts of Arctic sea ice loss by the Pacific Decadal Oscillation (PDO) is investigated in the IPSL-CM6A-LR ocean-atmosphere general circulation model. Ensembles of simulations are performed with constrained sea ice concentration corresponding to pre-industrial, present-day and future states, and initial conditions sampling warm and cold phases of the PDO. Using a general linear model, we estimate the simulated winter impact of sea ice loss, PDO and their combined effects. In response to sea ice loss, the Arctic lower troposphere warms and a negative North-Atlantic oscillation like pattern appears together with a weak deepening of the Aleutian Low. The two patterns are associated with a weakening of the poleward flank of the eddy-driven jet, while in the stratospheric the polar vortex weakens. Besides, a warm PDO phase induces a large positive Pacific North America pattern, as well as a small negative Arctic oscillation pattern associated with a weakening of the stratospheric polar vortex. However, the effects of PDO and Arctic sea ice loss are not additive. The Arctic sea ice teleconnections in both troposphere and stratosphere are reduced by the PDO, most importantly in the stratosphere. The results are discussed and compared to those obtained with the same model in atmosphere-only simulations, where sea ice loss does not significantly alter the stratospheric polar vortex.
Article
Full-text available
This paper provides initial results from a multi-model ensemble analysis based on the volc-pinatubo-full experiment performed within the Model Intercomparison Project on the climatic response to Volcanic forcing (VolMIP) as part of the sixth phase of the Coupled Model Intercomparison Project (CMIP6). The volc-pinatubo-full experiment is based on an ensemble of volcanic forcing-only climate simulations with the same volcanic aerosol dataset across the participating models (the 1991–1993 Pinatubo period from the CMIP6-GloSSAC dataset). The simulations are conducted within an idealized experimental design where initial states are sampled consistently across models from the CMIP6-piControl simulation providing unperturbed preindustrial background conditions. The multi-model ensemble includes output from an initial set of six participating Earth system models (CanESM5, GISS-E2.1-G, IPSL-CM6A-LR, MIROC-E2SL, MPI-ESM1.2-LR and UKESM1). The results show overall good agreement between the different models on the global and hemispheric scales concerning the surface climate responses, thus demonstrating the overall effectiveness of VolMIP's experimental design. However, small yet significant inter-model discrepancies are found in radiative fluxes, especially in the tropics, that preliminary analyses link with minor differences in forcing implementation; model physics, notably aerosol–radiation interactions; the simulation and sampling of El Niño–Southern Oscillation (ENSO); and, possibly, the simulation of climate feedbacks operating in the tropics. We discuss the volc-pinatubo-full protocol and highlight the advantages of volcanic forcing experiments defined within a carefully designed protocol with respect to emerging modelling approaches based on large ensemble transient simulations. We identify how the VolMIP strategy could be improved in future phases of the initiative to ensure a cleaner sampling protocol with greater focus on the evolving state of ENSO in the pre-eruption period.
Article
Full-text available
Abstract In climate models, the subgrid‐scale orography (SSO) parameterization imposes a blocked flow drag at low levels that is opposed to the local flow. In IPSL‐CM6A‐LR, an SSO lift force is also applied perpendicular to the local flow to account for the effect of locally blocked air in narrow valleys. Using IPSL‐CM6A‐LR sensitivity experiments, it is found that the tuning of both effects strongly impacts the atmospheric circulation. Increasing the blocking and reducing the lift lead to an equatorward shift of the Northern Hemisphere subtropical jet and a reduction of the midlatitude eddy‐driven jet speed. It also improves the simulated synoptic variability, with a reduced storm‐track intensity and increased blocking frequency over Greenland and Scandinavia. Additionally, it cools the polar lower troposphere in boreal winter. Transformed Eulerian Mean diagnostics also show that the low‐level eddy‐driven subsidence over the polar region is reduced consistent with the simulated cooling. The changes are amplified in coupled experiments when compared to atmosphere‐only experiments, as the low‐troposphere polar cooling is further amplified by the temperature and albedo feedbacks resulting from the Arctic sea ice growth. In IPSL‐CM6A‐LR, this corrects the warm winter bias and the lack of sea ice that were present over the Arctic before adjusting the SSO parameters. Our results, therefore, suggest that the adjustment of SSO parameterization alleviates the Arctic sea ice bias in this case. However, the atmospheric changes induced by the parametrized SSO also impact the ocean, with an equatorward shift of the Northern Hemisphere oceanic gyres and a weaker Atlantic meridional overturning circulation.
Article
Full-text available
Abstract This study presents the global climate model IPSL‐CM6A‐LR developed at Institut Pierre‐Simon Laplace (IPSL) to study natural climate variability and climate response to natural and anthropogenic forcings as part of the sixth phase of the Coupled Model Intercomparison Project (CMIP6). This article describes the different model components, their coupling, and the simulated climate in comparison to previous model versions. We focus here on the representation of the physical climate along with the main characteristics of the global carbon cycle. The model's climatology, as assessed from a range of metrics (related in particular to radiation, temperature, precipitation, and wind), is strongly improved in comparison to previous model versions. Although they are reduced, a number of known biases and shortcomings (e.g., double Intertropical Convergence Zone [ITCZ], frequency of midlatitude wintertime blockings, and El Niño–Southern Oscillation [ENSO] dynamics) persist. The equilibrium climate sensitivity and transient climate response have both increased from the previous climate model IPSL‐CM5A‐LR used in CMIP5. A large ensemble of more than 30 members for the historical period (1850–2018) and a smaller ensemble for a range of emissions scenarios (until 2100 and 2300) are also presented and discussed.
Article
Full-text available
Plain Language Summary Climate Model Intercomparison Project Phase 6 is an international project to compare the results from climate model simulations performed according to a common protocol. Such simulations require boundary conditions (called “climate forcings”), which are fed to the models in order to represent, for example, long‐lived greenhouse gases, ozone, atmospheric aerosols, or land surface properties. The same forcing data sets are used by the different modeling groups who carry out the Climate Model Intercomparison Project Phase 6 simulations; however, their implementation may differ as it depends on the model structure. This article gives details of how these forcing data were implemented in the IPSL‐CM6A‐LR model. Some of the forcing data are common to all types all simulations, whereas others depend on the runs considered. Radiative forcings, as estimated in the model, are presented for some of the forcing mechanisms.
Article
Full-text available
Abstract This study presents the version of the LMDZ global atmospheric model used as the atmospheric component of the Institut Pierre Simon Laplace coupled model (IPSL‐CM6A‐LR) to contribute to the 6th phase of the international Coupled Model Intercomparison Project (CMIP6). This LMDZ6A version includes original convective parameterizations that define the LMDZ “New Physics”: a mass flux parameterization of the organized structures of the convective boundary layer, the “thermal plume model,” and a parameterization of the cold pools created by reevaporation of convective rainfall. The vertical velocity associated with thermal plumes and gust fronts of cold pools are used to control the triggering and intensity of deep convection. Because of several shortcomings, the early version 5B of this New Physics was worse than the previous “Standard Physics” version 5A regarding several classical climate metrics. To overcome these deficiencies, version 6A includes new developments: a stochastic triggering of deep convection, a modification of the thermal plume model that allows the representation of stratocumulus and cumulus clouds in a unified framework, an improved parameterization of very stable boundary layers, and the modification of the gravity waves scheme targeting the quasi‐biennal oscillation in the stratosphere. These improvements to the physical content and a more well‐defined tuning strategy led to major improvements in the LMDZ6A version model climatology. Beyond the presentation of this particular model version and documentation of its climatology, the present paper underlines possible methodological pathways toward model improvement that can be shared across modeling groups.
Article
Full-text available
Nitrogen is an essential element controlling ecosystem carbon (C) productivity and its response to climate change and atmospheric [CO2] increase. This study presents the evaluation – focussing on gross primary production (GPP) – of a new version of the ORCHIDEE model that gathers the representation of the nitrogen cycle and of its interactions with the carbon cycle from the OCN model and the most recent developments from the ORCHIDEE trunk version. We quantify the model skills at 78 FLUXNET sites by simulating the observed mean seasonal cycle, daily mean flux variations, and annual mean average GPP flux for grasslands and forests. Accounting for carbon–nitrogen interactions does not substantially change the main skills of ORCHIDEE, except for the site-to-site annual mean GPP variations, for which the version with carbon–nitrogen interactions is in better agreement with observations. However, the simulated GPP response to idealised [CO2] enrichment simulations is highly sensitive to whether or not carbon–nitrogen interactions are accounted for. Doubling of the atmospheric [CO2] induces an increase in the GPP, but the site-averaged GPP response to a CO2 increase projected by the model version with carbon–nitrogen interactions is half of the increase projected by the version without carbon–nitrogen interactions. This model's differentiated response has important consequences for the transpiration rate, which is on average 50 mm yr−1 lower with the version with carbon–nitrogen interactions. Simulated annual GPP for northern, tropical and southern latitudes shows good agreement with the observation-based MTE-GPP (model tree ensemble gross primary production) product for present-day conditions. An attribution experiment making use of this new version of ORCHIDEE for the time period 1860–2016 suggests that global GPP has increased by 50 %, the main driver being the enrichment of land in reactive nitrogen (through deposition and fertilisation), followed by the [CO2] increase. Based on our factorial experiment and sensitivity analysis, we conclude that if carbon–nitrogen interactions are accounted for, the functional responses of ORCHIDEE r4999 better agree with the current understanding of photosynthesis than when the carbon–nitrogen interactions are not accounted for and that carbon–nitrogen interactions are essential in understanding global terrestrial ecosystem productivity.
Article
Full-text available
Abstract The representation of stratocumulus clouds, and of the stratocumulus to cumulus transitions which are ubiquitous features of marine boundary layer clouds, remains a challenge for climate models. We show how a mass flux representation of boundary layer convective structures combined with an eddy diffusivity scheme, the “thermal plume model,” first developed to represent cumulus clouds, can also adequately simulate stratocumulus and the stratocumulus to cumulus transition in a climate model. To achieve this, the detrainment formulation, in which detrainment increases for increasing negative buoyancy, has to be slightly modified: the buoyancy of a thermal plume parcel of air is computed by comparing the virtual potential temperature θv,th of the parcel with that of the surrounding environment θv,env at a given distance above instead of at the same level. This is consistent with the picture of detrained air parcels that experience some overshoot and reach a final destination at a level lower than the one at which they effectively leave the cloud or organized convective plume. The impacts of this modification are documented both for selected cases of stratocumulus, in comparison with large‐eddy simulations, and in full 3‐D climate simulations, in comparison with satellite observations of cloud cover. The modified scheme provides a uniform treatment of the dry convective boundary layer, of cumulus clouds, of stratocumulus, and of the transition from stratocumulus to cumulus. It is included in the most recent version of the LMDZ atmospheric general circulation model.
Article
Full-text available
Understanding the unfolding challenges of climate change relies on climate models, many of which have large summer warm and dry biases over Northern Hemisphere continental midlatitudes. This work, with the example of the model used in the updated version of the weather@home distributed climate model framework, shows the potential for improving climate model simulations through a multiphased parameter refinement approach, particularly over the northwestern United States (NWUS). Each phase consists of (1) creating a perturbed parameter ensemble with the coupled global–regional atmospheric model, (2) building statistical emulators that estimate climate metrics as functions of parameter values, (3) and using the emulators to further refine the parameter space. The refinement process includes sensitivity analyses to identify the most influential parameters for various model output metrics; results are then used to cull parameters with little influence. Three phases of this iterative process are carried out before the results are considered to be satisfactory; that is, a handful of parameter sets are identified that meet acceptable bias reduction criteria. Results not only indicate that 74 % of the NWUS regional warm biases can be reduced by refining global atmospheric parameters that control convection and hydrometeor transport, as well as land surface parameters that affect plant photosynthesis, transpiration, and evaporation, but also suggest that this iterative approach to perturbed parameters has an important role to play in the evolution of physical parameterizations.
Article
Full-text available
Some of the largest and most persistent circulation errors in global numerical weather prediction and climate models are attributable to the inadequate representation of the impacts of orography on the atmospheric flow. Existing parametrization approaches attempting to account for unresolved orographic processes, such as turbulent form drag, low-level flow blocking or mountain waves, have been successful to some extent. They capture the basic impacts of the unresolved orography on atmospheric circulation in a qualitatively correct way and have led to significant progress in both numerical weather prediction and climate modelling. These approaches, however, have apparent limitations and inadequacies due to poor observational evidence, insufficient fundamental knowledge and an ambiguous separation between resolved and unresolved orographic scales and between different orographic processes. Numerical weather prediction and climate modelling has advanced to a stage where these inadequacies have become critical and hamper progress by limiting predictive skill on a wide range of spatial and temporal scales. More physically based approaches are needed to quantify the relative importance of apparently disparate orographic processes and to account for their combined effects in a rational and accurate way in numerical models. We argue that, thanks to recent advances, significant progress can be made by combining theoretical approaches with observations, inverse modelling techniques and high-resolution and idealized numerical simulations.
Article
Full-text available
Past studies have shown that climate simulations have substantial warm and dry biases during the summer in the conterminous United States (CONUS), particularly in the central Great Plains (CGP). These biases have critical implications for the interpretation of climate change projections, but the complex overlap of multiple land-atmosphere feedback processes make them difficult to explain (and therefore correct). Even though surface soil moisture (SM) is often cited as a key control variable in these processes, there are still knowledge gaps about its specific role. Here, we use recently developed remotely sensed SM products to analyse the link between spatial patterns of summertime SM, precipitation and air temperature biases over CONUS in 20 different CMIP5 simulations. We identify three main types of bias combinations: (i) a dry/warm bias over the CGP region, with a significant inter-model correlation between SM and air temperature biases (R = −0.65), (ii) a wet/cold bias in NW CONUS, and (iii) a dry/cold bias in SW CONUS. Combined with irrigation patterns, these results suggest that land-atmosphere feedbacks over the CGP are not only local but have a regional dimension, and demonstrate the added-value of large-scale SM observations for resolving the full feed-back loop between precipitation and temperature.
Article
Full-text available
Accurate assessment of anthropogenic carbon dioxide (<span classCombining double low line"inline-formula">CO2 ) emissions and their redistribution among the atmosphere, ocean, and terrestrial biosphere - the "global carbon budget" - is important to better understand the global carbon cycle, support the development of climate policies, and project future climate change. Here we describe data sets and methodology to quantify the five major components of the global carbon budget and their uncertainties. Fossil <span classCombining double low line"inline-formula">CO2 emissions (<span classCombining double low line"inline-formula"> E FF ) are based on energy statistics and cement production data, while emissions from land use and land-use change (<span classCombining double low line"inline-formula"> E LUC ), mainly deforestation, are based on land use and land-use change data and bookkeeping models. Atmospheric <span classCombining double low line"inline-formula">CO2 concentration is measured directly and its growth rate (<span classCombining double low line"inline-formula"> G ATM ) is computed from the annual changes in concentration. The ocean <span classCombining double low line"inline-formula">CO2 sink (<span classCombining double low line"inline-formula"> S OCEAN ) and terrestrial <span classCombining double low line"inline-formula">CO2 sink (<span classCombining double low line"inline-formula"> S LAND ) are estimated with global process models constrained by observations. The resulting carbon budget imbalance (<span classCombining double low line"inline-formula"> B IM ), the difference between the estimated total emissions and the estimated changes in the atmosphere, ocean, and terrestrial biosphere, is a measure of imperfect data and understanding of the contemporary carbon cycle. All uncertainties are reported as <span classCombining double low line"inline-formula