Content uploaded by Alessio Collalti
Author content
All content in this area was uploaded by Alessio Collalti on Jan 09, 2024
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tejr20
European Journal of Remote Sensing
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tejr20
Regional estimates of gross primary production
applying the Process-Based Model 3D-CMCC-FEM
vs. Remote-Sensing multiple datasets
D. Dalmonech, E. Vangi, M. Chiesi, G. Chirici, L. Fibbi, F. Giannetti, G. Marano,
C. Massari, A. Nolè, J. Xiao & A. Collalti
To cite this article: D. Dalmonech, E. Vangi, M. Chiesi, G. Chirici, L. Fibbi, F. Giannetti,
G. Marano, C. Massari, A. Nolè, J. Xiao & A. Collalti (2024) Regional estimates of gross
primary production applying the Process-Based Model 3D-CMCC-FEM vs. Remote-
Sensing multiple datasets, European Journal of Remote Sensing, 57:1, 2301657, DOI:
10.1080/22797254.2023.2301657
To link to this article: https://doi.org/10.1080/22797254.2023.2301657
© 2024 The Author(s). Published by Informa
UK Limited, trading as Taylor & Francis
Group.
View supplementary material
Published online: 09 Jan 2024.
Submit your article to this journal
View related articles
View Crossmark data
Regional estimates of gross primary production applying the Process-Based
Model 3D-CMCC-FEM vs. Remote-Sensing multiple datasets
D. Dalmonech
a,b
, E. Vangi
a
, M. Chiesi
c
, G. Chirici
d
, L. Fibbi
c
, F. Giannetti
d
, G. Marano
e
,
C. Massari
e
, A. Nolè
f
, J. Xiao
g
and A. Collalti
a,b
a
Forest Modelling Laboratory, Institute for Agriculture and Forestry Systems in the Mediterranean, National Research Council of Italy (CNR-
ISAFOM), Perugia, Italy;
b
National Biodiversity Future Center (NBFC), Palermo, Italy;
c
National Research Council of Italy, Institute of
BioEconomy,(CNR-IBE), Sesto Fiorentino, Italy;
d
GeoLab - Laboratory of Forest Geomatics, Dept. of Agriculture, Food, Environment and
Forestry, Università degli Studi di Firenze, Firenze, Italy;
e
Research Institute for Geo-Hydrological Protection, National Research Council (CNR-
IRPI), Perugia, Italy;
f
School of Agricultural, Forest, Food and Environmental Sciences University of Basilicata, Potenza, Italy;
g
Earth Systems
Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, USA
ABSTRACT
Process-based Forest Models (PBFMs) oer the possibility to capture important spatial and
temporal patterns of carbon uxes and stocks in forests. Yet, their predictive capacity should be
demonstrated not only at the stand-level but also in the context of broad spatial and temporal
heterogeneity. We apply a stand scale PBFM (3D-CMCC-FEM) in a spatially explicit manner at 1
km resolution in southern Italy. We developed a methodology to initialize the model that
includes information derived from the integration of Remote Sensing (RS) and the National
Forest Inventory (NFI) data and regional forest maps to characterize structural features of the
main forest species. Gross primary production (GPP) is simulated over 2005–2019 period and
the model predictive capability of the model in simulating GPP is evaluated both aggregated as
at species-level through multiple independent data sources based on dierent nature RS-
based products. We show that the model is able to reproduce most of the spatial (~2800 km
2
)
and temporal (32 years in total) patterns of the observed GPP at both seasonal, annual and
interannual time scales, even at the species-level. These promising results open the possibility
of conndently applying the 3D-CMCC-FEM to investigate the forests’ behaviour under climate
and environmental variability over large areas across highly variable ecological and bio-
geographical heterogeneity of the Mediterranean region.
ARTICLE HISTORY
Received 10 July 2023
Revised 14 December 2023
Accepted 31 December 2023
KEYWORDS
Process-based forest model;
wall-to-wall map; gross
primary production; national
forest inventory;
Mediterranean region
Introduction
Forest ecosystems absorb globally ~ 2 Gigatonnes of
Carbon (C) stocking the carbon in their biomass and
soil, thus acting as a net carbon sink. In Europe alone,
forest ecosystems, which cover about a 40%, currently
act as a net carbon sink for ~ 315 Megatonnes of CO
2
eq
and compensate for about 8% of EU-27’s total green-
house emissions (Verkerk et al., 2022). However,
adverse climate impacts such as heat waves and drought
(Allen et al., 2015; D’Andrea et al., 2020, 2021; Schuldt
et al., 2020) and increasing natural disturbance rates
(Grünig et al., 2023; Patacca et al., 2023) are all stressors
which have potentially significant effects on current and
future forest dynamics, jeopardizing the European for-
est ecosystems functioning and their carbon mitigation
potential under future climate change (De Marco et al.,
2022; Schuldt et al., 2020; Senf et al., 2020).
Nevertheless, ground data scarcity and short-term
monitoring efforts still represent major challenges in
studying the effects of climate change on forest
dynamics in Mediterranean areas because are
characterized by a large ecosystem heterogeneity and
a biogeographically diverse structure shaped by
human activity (Gauquelin et al., 2018; Médail et al.,
2019; Peñuelas et al., 2017). As the Mediterranean
region are known as a climate change “hotspot”
(Dubrovský et al., 2014; Noce et al., 2016), experien-
cing already increasing frequency in extreme events
such heat waves and droughts (Vogel et al., 2021), it is
thus crucial in these areas to provide large-scale forest
monitoring and eventually predict the future state of
forest ecosystems. Recent efforts have addressed the
shortage of ancillary or ground data by integrating
National Forest Inventories (NFI) data and high-reso-
lution remote-sensing (RS) data. These initiatives pro-
duced comprehensive wall-to-wall maps of various
forest variables, such as growing stocks volumes or
biomass (Chirici et al., 2020; Giannetti et al., 2022;
Nord-Larsen & Schumacher, 2012; Vangi et al., 2023;
Waser et al., 2017). These maps represent meaningful
data for and carbon cycle assessment (e.g. Vangi et al.,
2023). In parallel, process-based forest models (PBFMs)
CONTACT D. Dalmonech daniela.dalmonech@cnr.it Forest Modelling Laboratory, Institute for Agriculture and Forestry Systems in the
Mediterranean, National Research Council of Italy (CNR-ISAFOM), Perugia 06128, Italy
Supplemental data for this article can be accessed online at https://doi.org/10.1080/22797254.2023.2301657.
EUROPEAN JOURNAL OF REMOTE SENSING
2024, VOL. 57, NO. 1, 2301657
https://doi.org/10.1080/22797254.2023.2301657
© 2024 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which
permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been
published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
are analytical tools developed and tested over a wide
range of applications, because of their capability in simu-
lating forest ecosystem even on long-term dynamics
(Bugmann & Seidl, 2022; Vacchiano et al., 2012), carbon
fluxes exchange and stocks (Chiesi et al., 2010;
Dalmonech et al., 2022; Mahnken et al., 2022; Reyer,
2015; Reyer et al., 2014) under external environmental
variability by accounting for population dynamics and
inner physiological processes mechanistically
(Maréchaux et al., 2021; Pretzsch et al., 2008;
Vacchiano et al., 2012). On the other side, given the
large amount of requested data for their initialization
and parameterization, such models are mostly run at a
very local scale, i.e. site level (one hectare or a bit more),
where high-quality/measured ancillary data and meteor-
ological data are available (Collalti et al., 2016; Suárez-
Muñoz et al., 2023). Yet, initializing stand scale PBFMs
from actual, measured forest state variables (or close to
the observed states), rather than from equilibrium con-
ditions, is the desirable option to implicitly take into
account the climate and management history of the
site, and to more realistically simulate the response of
forests and their resilience, even in the context of climate
change, and natural and anthropogenic disturbances (e.
g. Kannenberg et al., 2020; Pretzsch et al., 2008; Zampieri
et al., 2021).
Despite their undoubted utility, only recently,
stand-scale PBFMs were applied on a regular grid
(Minunno et al., 2019; Sanchez-Ruiz et al., 2018)
with the purpose of estimating aggregated, country-
level, carbon stocks and wood products. Yet, in south-
ern Europe and in the Mediterranean the ability of a
PBFMs to simulate the GPP at large scale is crucial,
but largely overlooked, because photosynthesis
respond sensitively to both meteorological and climate
variability and spatial heterogeneity at the daily to
decadal scales (e.g. Fernández-Martínez et al., 2023;
Mahnken et al., 2022), therefore GPP can be, and has
been, considered a good proxy of the ecosystem phy-
siological functionality especially in Mediterranean
forests (Chen et al., 2023; Collalti et al., 2018). The
objective of this study is, thus, the application and the
testing of the biochemical, biophysical, process-based
forest model 3D-CMCC-FEM (Collalti et al., 2016,
2018, 2014), on a regular grid at 1 km spatial resolu-
tion in the Mediterranean area, initializing the model
through the use of spatial information derived by
integrating data of different nature: NFI data, RS-
based wall-to-wall map, and regional forest maps to
characterize structural features of the main forest spe-
cies. The final aim is to simulate GPP at regional scale.
As a case study, the model was tested over one of the
southernmost regions in Italy, the Basilicata Region,
which, as most regions in the Mediterranean basin,
spans over a multitude of ecological, morphological
and soil-type gradients and climate conditions. The
capability of the model to simulate the GPP in terms of
mean annual, seasonal and interannual variability is
evaluated by comparing model results against a port-
folio of different independent RS-based GPP-esti-
mates. The 3D-CMCC-FEM GPP vs. RS-based GPP
data agreement is presented, and sources of uncertain-
ties and challenges of applying a PBFMs with the
presented modeling strategy are also discussed.
Materials and methods
Study area
The Basilicata region has a spatial extent of about 10,000
km
2
and is located in southern Italy (Figure 1(a)). It is
characterized by typical Mediterranean climate condi-
tions with hot and dry summers and wet and mild
winters. The region has been chosen to test the model
Figure 1. a) study area in the Italian peninsula and elevation map of the Region Basilicata. The red line indicates the administrative
limits of the region, b) distribution of the dominant forest class at 1 × 1 km spatial resolution.
2D. DALMONECH ET AL.
in a complex biogeographic area with pronounced envir-
onmental gradients. The territory is characterized by
about 47% of mountain areas represented by the
Apennines Mountains, followed by hilly areas, about
45%, and then plain. Average annual precipitation varies
between ~ 500 and ~ 2000 millimeters (mm) per year,
mirroring the orographic complexity of the region and
the proximities to the sea, with a west-to-east gradient
from humid to dry sub-humid areas.
According to the last NFI (INFC 2015), forest vege-
tation and other wooded lands occupy 392,412 hec-
tares (ha), about 39% of the region. Deciduous species
cover 54% of the forest area and are represented
mainly by oaks spp (Q. cerris L., Q. ilex L.), which
dominated the hilly areas between 400 and 1200 m
above sea level (a.s.l.), and European beech (Fagus
sylvatica L.) the main species above 1000 m a.s.l.
Coniferous species are less abundant and are repre-
sented mainly by pines spp (P. halepensis Mill., and P.
nigra J.F. Arnold), often used for reforestation pur-
poses (Figure 1(b)).
The study area was tessellated into a 1 km spatial
resolution regional grid whose pixel area represents
the best compromise between the forcing variables
and the operative model resolution, for a total of ~
10,073 pixels. The regional grid served as a spatial
reference grid for resampling the input data needed
for the model initialization to 1 km spatial resolution.
The process-based model 3D-CMCC-FEM
The 3D-CMCC-FEM (“Three Dimensional – Couple
Model Carbon Cycle – Forest Ecosystem Module”) is an
ecophysiological, biogeochemical, biophysical pro-
cess-based model which simulates the dynamic of
carbon, water and nitrogen and the allocation through
a cohort-structured forest stands (Collalti et al., 2016,
2018, 2014, 2017, 2020, 2022; Dalmonech et al., 2022;
Marconi et al., 2017; Testolin et al., 2023), providing
detailed output from daily to annual time scale of
carbon fluxes and stocks. The model simulates forest
growth and structural development at varying envir-
onmental conditions and different climate, atmo-
spheric CO
2
concentrations and forest management
scenarios (Collalti et al., 2018; Dalmonech et al., 2022;
Testolin et al., 2023).
The daily gross photosynthesis is simulated
through the Farquhar–von Caemmerer–Berry bio-
chemical model (Farquhar et al., 1980), modified for
sun and shaded leaves (De Pury & Farquhar, 1997),
and acclimated for temperature (Kattge & Knorr,
2007). The carbon and nitrogen allocation schemes
are described extensively in Collalti et al. (2016,
2019, 2020) and Merganičová et al. (2019). Tree
removal can occur via management (Testolin et al.,
2023) or by natural mortality (Collalti et al., 2018).
Self-thinning, age-related, carbon starvation and
background mortality represent the different types of
mortalities simulated by the model (Collalti et al.,
2016). Soil hydrology is simulated by means of a
one-soil layer bucket model with free drainage. The
plant water availability in the model is thus modulated
by the soil depth (i.e. rooting depth until which the
water is uptake to sustain leaf transpiration), because
of the zero-dimensional soil model. Soil water stress
operates on canopy exchange processes via stomatal
and biochemical pathways (e.g. photosynthesis). An
in-depth description of the model’s underlying char-
acteristics, effects of climate change and model para-
meter sensitivity and uncertainty, as well as model
limitations, is reported in Collalti et al. (2019).
Forest data source
The model requires the description of the forest struc-
tural characteristics: i.e. diameter at breast height
(DBH), tree height (H), stand density (number of
trees per cell) and age class, in order to be initialized
and to run the simulations. In this work, to initialize
the model for each grid cell of the matrix, we used data
from the second NFI for 2005 (www.inventariofores
tale.org). The NFI is based on a three-phase, systema-
tic, unaligned sampling design with 1 km grid cells. In
the first phase, 301,300 points were extracted and
classified using aerial orthophotos into forest/non-
forest categories. In the second phase, a field survey
was carried out in a sub-sample of the first-phase
points falling in the forest category, to collect qualita-
tive information such as forest type, management, and
property. Finally, in the third phase, for a sub-sample
of 6782 points extracted from the second-phase
points, a dendrometric survey was carried out for
circular plots of a 13 m radius. All tree stems with
DBH of at least 2.5 cm were callipered, and for a
subsample, height was measured.
Field-survey data from the NFI were used to pro-
duce a “wall-to-wall” map of the forest basal area at a
23 m spatial resolution. This map consists of random
forests predictions of basal area per hectare for all 23
m spatial resolution forest pixels. The random forests
model was trained using NFI plot-level data and
Landsat and other RS-based datasets as predictors,
including climate information such as minimum,
mean, and maximum temperature, and daily precipi-
tations from the E-OBS dataset, the land-only gridded
daily observational dataset for Europe (see Section
2.4). The statistical model fitting and tuning steps
were carried out using the “randomForest” package
in the statistical software R 4.0.5 (Liaw & Wiener,
2002). More information about the procedure can be
found in Chirici et al. (2020), Vangi et al. (2021) and
Giannetti et al. (2022). The pixel-level estimations of
the basal area range between 5 and 43 m
2
ha
−1
with a
mean value of 12 m
2
ha
−1
that is in line with the range
EUROPEAN JOURNAL OF REMOTE SENSING 3
reported in the context of NFI for the field plots data
estimation.
The basal area data were then resampled to the
regional grid at 1 km resolution. The resampled amp
was then masked according to the regional forest map
by Constantini et al., (2006). This map provided the
forest type according to Barbati et al. (2007) and the
development stage of the forests. Estimation of the
forest structural data used to initialize the model in
each regional grid cell is described in section 2.5.1.
Forcing and soil data
The 3D-CMCC-FEM was forced with daily maximum
(Tmax,°C) and minimum (Tmin,°C) air temperatures,
precipitation (P, mm day
−1
), downward short-wave
radiation at the surface (SW MJ m
−2
day
−1
) and relative
humidity (RH, %). Meteorological data for the period
2005–2019 were retrieved from the E-OBS v.23.1e
gridded dataset (Cornes et al., 2018), which is provided
at 0.1° decimal degree resolution. The E-OBS dataset
has already been used in environmental impact studies
(e.g. Rita et al., 2020), climate scenarios bias correction
(Dosio & Paruolo, 2011; Rojas et al., 2011) and bench-
mark activities (Herrera et al., 2019; Lorenz et al., 2019;
Massari et al., 2020; Moreno & Hasenauer, 2016).
All the physical variables were bilinearly interpolated
to the regional grid at 1 km resolution, and the tem-
perature data were corrected for the topographic effect
by applying a lapse rate correction based on elevation
differences between the E-OBS reference elevation and
the finest and most accurate DEM (Digital Elevation
Model) currently available in Italy, obtained in the
framework of the TINITALY project. The
TINITALYDEM is a national DEM of Italy at 10 m
resolution (Tarquini et al., 2009), and for the lapse
rate correction it was resampled at 1 km resolution of
the regional grid. The lapse rate estimates of −5°C km
−1
for Tmax and −3°C km
−1
for Tmin were derived from
termo-pluviometric ground station measurements over
an elevation transect in the Basilicata region.
The model was forced by global annual atmo-
spheric CO
2
concentrations from Dlugokencky and
Tans (https://www.esrl.noaa.gov/gmd/ccgg/trends/),
covering in total the years 2005–2019.
The model requires information on soil depth as
well as soil texture for each grid cell. As a proxy of soil
depth, we used the estimated depth available to root
from the European Soil Database Derived Data pro-
duct ESBD v2 (Hiederer, 2013) provided at 1 km reso-
lution. The lower depth value of each class of the map
was attributed, and a maximum of 1 m for the rooting
depth was set. This boundary value can be considered
a good approximation for European forests (Schenk &
Jackson, 2005). Soil texture as a percentage of clay, silt,
and sand was estimated from the pedological map of
the region (year 2005).
The meteorological and geographic information
were all re-projected using the same coordinate refer-
ence system WGS84/UTM zone 33 North (EPSG:
32633), and then resampled onto the regional grid at
1 km resolution. The main data analyses were per-
formed using the computing language R (R Core
Team 2021). Key packages used for data preprocessing
included “terra” (Hijmans et al., 2022) and “rgdal”
(Bivand et al., 2015).
Model simulations
Model initialization
The 3D-CMCC-FEM model (v.5.6) was applied on the
regional grid at 1 km spatial resolution to simulate the
forest carbon dynamic starting in January 2005 until
December 2019. The model requires the description of
the forest attributes at the beginning of simulation in
order to be initialized. The initial forest state was set
according to a simplified model initialization of the
forest aboveground structural complexity, i.e. for each
grid cell, we determined the dominant forest species
and estimated the average structural data: the average
tree diameter at breast height (DBH), the average tree
height (H), the stand density and the average age class
which represent the mandatory initial data for model
initialization.
The following six key species were considered:
European beech (F. sylvatica L.), Black pine (P. nigra
J.F. Arnold), Sweet chestnut (C. sativa Mill.), Turkey
oak (Q. cerris L.), Aleppo pine (P. halepensis Mill.) and
Holm oak (Q. ilex L.) as representative of the most
common forest types in the study area. The regional
forest map was used to define for each 1 km grid cell
the dominant forest species as the one covering the
highest forest fraction (Figure 1(b)) and the average
age class based on the development stage (provided by
the regional forest map along with the forest class).
Areas with dominance of maquis and other minor
forest species (the latest accounts for ~ 3.9% of the
region and are not currently parameterized in the
model) were masked out from the regional gird.
The final dominant forest age classes result in a total
of ~2800 km
2
, corresponding to ~ 80% of the entire
regional forested area.
Tree density data of the NFI field plots were used to
provide a representative estimate of the forest density
in each grid cell. To do so, we zonally averaged the
density data according to the regional grid and the
dominant forest classes. The basal area map was then
used in combination with the density map data to
calculate an average cell-level DBH and to provide a
1 km resolution DBH map for each dominant species
(data not shown).
To be consistent with the model processing and
inherent logic, the H was calculated from the average
DBH by applying the calibrated Chapman-Richards
4D. DALMONECH ET AL.
equation (Richards, 1959), which links DBH and H,
for each forest class with each cell. Starting from these
mandatory structural variables, i.e. DBH and H, the
model self-initializes the other state variables: i.e. leaf,
stem, branches, coarse and fine root, reserves (non-
structural carbon, NSC; which includes starch and
sugars) carbon and nitrogen pools using a species-
specific parameterization at the beginning of the
model simulations. Species-specific model parameters
(e.g. specific leaf area or maximum stomatal conduc-
tance; see Table S1) were retrieved and calibrated from
literature data. Specifically, the species-specific allo-
metric equations linking DBH to H and DBH to
stem biomass were calibrated in this study using the
second NFI tree-level data.
Simulations settings
Due to the relatively short time period covered by the
simulations, i.e. 2005–2019, we did not consider any
change in dominant species or land use. For the same
reason, and because of the spatial resolution, a con-
stant thinning rate implemented each year was con-
sidered as the only silvicultural intervention as
similarly as in Gutsch et al. (2018). The thinning rate
was an approximation derived from the forest man-
agement guidelines of the region as set to a yearly
removal rate of 1%, corresponding to ~ 20% of bio-
mass removed in 20 years. Forests in protected areas of
the Natura2000 network of the region are character-
ized by lower disturbance extension compared to the
other forested areas according to the disturbance maps
produced by Francini et al. (2021), and are assumed to
be interested by a lower level of tree harvesting. For
simplicity, we assume, thus, that mortality in protected
area is mainly caused by natural and background
mortality alone. Grid cells interested in fire events
over the period 2005–2019 were identified using the
national dataset of burnt areas from forest fires, pro-
duced by the Italian Forest Service (Comando Unità
Forestali, Ambientali e Agroalimentari of Carabinieri).
This dataset is acquired through a ground survey using
Global Navigation Satellite System receiver (GNSS)
and is available from 2005 to 2019. Grid cells where
more than 30% of the forest area was interested in fire
events were excluded from the analysis as the model
does not simulate extended natural disturbances (i.e.
fire and pests). This threshold was a compromise
between excluding too many grid cells and including
too many fire-disturbed areas.
Remote sensing – GPP datasets
Datasets based on RS-based data (or modeled by for-
cing with remote sensed data) are the most suitable
candidate to assess the overall model capability at
reproducing the GPP over large areas, due to their
continuous spatial and temporal coverage. In order
to make a more complete and comprehensive agree-
ment assessment, we selected gridded GPP estimates
from different independent sources as reference data-
sets, which are:
GOSIF-GPP
The GOSIF GPP dataset (Li & Xiao, 2019b) is a recently
developed GPP product that is based on the global
OCO-2 based SIF product (Li & Xiao, 2019a). The
GOSIF setup combines in a data-driven approach the
remotely sensed sun-induced fluorescence (SIF),
observed by the Orbiting Carbon Observatory-2
(OCO-2), the enhanced vegetation index (EVI) from
the Moderate Resolution Imaging Spectroradiometer
(MODIS) satellite data, meteorological data, i.e. photo-
synthetically active radiation (PAR), vapor pressure
deficit (VPD) and air temperature obtained from the
NASA reanalysis MERRA-2 data set to return a gridded
SIF dataset (Li & Xiao, 2019a). Established relationships
between the original OCO-2 SIF and flux tower GPP (Li
et al., 2018; Xiao et al., 2019) were then used to provide
the final gridded GPP product. For the model-data
comparison in this study we used the monthly and
annually aggregated ensemble mean of eight different
GPP estimates resulting from different GPP-SIF rela-
tionships, (http://globalecology.unhedu). This GPP
product, hereinafter simply referred as “GOSIF”, pro-
vides GPP estimates at 0.05 degree (corresponding to
~5 km) spatial resolution aggregated on a monthly time
step, over the years 2005–2019.
CFIX-GPP
The CFIX dataset provides gridded GPP values cover-
ing Italy. The estimates are obtained combining
meteorological and remotely sensed data within a
Light Use Efficiency modeling approach (Maselli et al.,
2006; Veroustraete et al., 2002). The original model
version was further modified to simulate the GPP in
water-limited, Mediterranean forest ecosystems by
Maselli et al. (2009), who introduced a short-term
water stress factor based on daily meteorological data.
In particular, the authors utilized the 1 km normalized
difference vegetation index (NDVI) from the Spot-
VEGETATION imagery to linearly retrieve the fraction
of absorbed photosynthetically active radiation
(APAR), and the downscaled E-OBS meteorological
data (both air temperature and precipitation) (Fibbi et
al., 2016; Maselli et al., 2012) to retrieve the GPP of all
Italian forests at 1 km spatial resolution. The accuracy
of the product was assessed by comparison to Eddy-
Covariance GPP data (Pastorello et al., 2020) collected
at several EC sites spread over the Italian peninsula and
for different forest types, showing satisfactory perfor-
mances (Chirici, 2016). The currently utilized product
provides forest GPP estimates at 1 km spatial resolution
aggregated on a monthly time step, over the years 2005–
2013.
EUROPEAN JOURNAL OF REMOTE SENSING 5
FLUXCOM-GPP
The FLUXCOM dataset is an upscaled product derived
from different Machine Learning-based approaches
(ML) combining data from the FLUXNET network of
eddy covariance towers with RS and meteorological data
as predictors (Jung et al., 2019; Pastorello et al., 2020). In
this study we used the FLUXCOM-RS dataset which
embeds in its statistical processing RS land products at
8-day resolution from the MODIS instrument, such as
EVI, fAPAR and the land surface temperature (see Jung
et al. (2019), for a thorough description of the
FLUXCOM products). Compared to other products of
the FLUXCOM-database, the FLUXCOM-RS dataset has
a finer spatial resolution, 0.083 degree (corresponding to
~8 km) spatial resolution, allowing to better deal with the
complex topography of the region under study.
The FLUXCOM-RS dataset, hereinafter simply
referred to as FLUXCOM, has been widely used as a
reference dataset in several inter-comparison studies
with both model and other reference datasets
(O’Sullivan et al., 2020; Wang et al., 2021; Zhang &
Ye, 2021). The selected dataset is provided as the
ensemble mean of monthly data computed from mul-
tiple ML algorithms covering the period 2000–2015
(www.fluxcom.org).
3D-CMCC-FEM GPP versus RS-based GPP
For comparability with the 3D-CMCC-FEM GPP
results, all the RS-based datasets were resampled to
the regional grid at 1 km resolution and reprojected
using the WGS84/UTM zone 33 North coordinate
reference system (EPSG: 326633) and compared at
both spatial and temporal levels against the 3D-
CMCC-FEM GPP data. Results at species-level are
first shown as aggregated data over regional level and
then shown at species-level.
Spatial variability analyses
The spatial agreement between the 3D-CMCC-FEM
GPP and the RS-based GPP data was assessed by
considering: Root Mean Square Error (RMSE),
Relative Difference (RD, expressed as: (modelled –
observed)/observed *100, where “observed” stands
for RS-based data and “modeled” for 3D-CMCC-
FEM GPP), Pearson’s correlation (r) and the SPAtial
EFficiency (SPAEF) metrics. The SPAEF (see Eq. 1) is
an integrated evaluation index which considers the
Pearson’s spatial correlation (r), the fraction of the
coefficient of variation β¼σsim =μsim
�=σobs=μobs
�
and the model-data histogram intersection
γ¼Pn
j¼1minðKj=LjÞ=Pn
j¼1K, where n is the number
of bins in the histogram and K and L are the histogram
for observed and simulated data (Koch et al., 2018).
SPAEF is equal to 1 in the case of a perfect match and 0
in case of mismatch.
SPAEF ¼1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r1ð Þ2þβ1ð Þ2þγ1ð Þ2
q(1)
The statistics were computed on the mean annual GPP
and seasonal GPP: i.e. winter (December, January,
February: DJF), spring (March, April, May: MAM),
summer (June, July, August: JJA) and autumn
(September, October, November: SON) for 3D-
CMCC-FEM and RS-based data.
Temporal variability analyses
The temporal agreement between the 3D-CMCC-
FEM GPP and the RS-based GPP data was assessed
by considering RMSE, RD, the Fractional Variance
(FV), which returns values bounded between −2 and
2 and it is equal to 0 when “modelled” and “observed”
have the same variance (Janssen & Heuberger, 1995).
The Pearson’s correlation coefficient r was also used at
grid-cell level, which returns the direction of the cor-
relation i.e. model-data correspondence of the sign of
year to year variability. Conversely to the spatial ana-
lysis the mean seasonal cycle (MSC) was also consid-
ered here. Anomalies were calculated for the model
and data by first removing the long-term linear trend,
and normalized by their standard deviation.
Both the 3D-CMCC-FEM and the RS-based GPP
climate sensitivity in the summer period have been eval-
uated through the Standardized Precipitation
Evaporation Index, SPEI (Vicente-Serrano et al., 2010).
The SPEI drought index helps to highlight periods of
wetter or drier conditions. This is a multi-scalar meteor-
ological drought index based on a statistical transforma-
tion of the climatic water balance, i.e. precipitation minus
potential evapotranspiration. We computed the SPEI
using the simplistic Hargreaves equations for the poten-
tial evapotranspiration calculation and considered differ-
ent time scales, with the aim to cover the growing season
period before the month of August. Following a similar
approach as in Mahnken et al. (2022), we regressed the
residuals of the 3D-CMCC-FEM and RS-based dataset
anomalies against the SPEI values as the predictor, and
the slope of the linear regression computed. Values of the
slope close to 0 indicate that the 3D-CMCC-FEM shows a
GPP-sensitivity to SPEI like the RS-based dataset.
The overall analyses were carried out considering
the entire RS-based dataset available years, thus
2005–2013 for CFIX, 2005–2015 for FLUXCOM
and 2005–2019 for GOSIF (i.e. 32 years in total).
Results
Spatial variability analysis
The 3D-CMCC-FEM simulated average annual GPP
for the period 2005–2019 is shown in Figure 2, with
overall values ranging from ~ 600 up to ~ 2200 gC m
−2
yr
−1
. 3D-CMCC-FEM GPP follows a west-east gradient
with higher productivity over the western side of the
6D. DALMONECH ET AL.
region in correspondence with the more humid and the
more productive beech-dominated areas; the forested
areas over the plain zones are mainly dominated by
Mediterranean pine species, which show the lowest
GPP values (Figure 2). At annual level (multiyear
mean annual) better correlations between 3D-CMCC-
FEM and RS-based GPP data are with GOSIF and
FLUXCOM (r = 0.77 and 0.67, respectively; Table 1)
as also for SPAEF metric (SPAEF = 0.62 and 0.59,
respectively), while lower RMSE and RD are when
3D-CMCC-FEM GPP is compared with GOSIF and
CFIX (RMSE = 235.8 and 221.9 gC m
−2
yr
−1
, and
RD = −4.3 and 0.01%, respectively). At seasonal level,
3D-CMCC-FEM GPP better correlates with GOSIF and
CIFX during summer (RMSE = 181.3 and 123.8 g
C m
−2
yr
−1
, RD = −13.7 and 10.7% and r = 0.79 and
0.8, respectively). Similarly, also in spring 3D-CMCC-
FEM better correlates, although with lower values, with
GOSIF and CIFX (RMSE = 104.4 and 103.6 gC m
−2
yr
−1
, RD = 9 and −7.9% and r = 0.4 and 0.28, respec-
tively). In winter 3D-CMCC-FEM GPP better corre-
lates with CIFX and FLUXCOM (r = 0.55 and 0.5, and
SPAEF = −1.26 and 0.53, respectively) but with slightly
lower RMSE values when compared with GOSIF and
FLUXCOM (RMSE = 54.3 and 39.97 gC m
−2
yr
−1
).
During autumn 3D-CMCC-FEM is in agreement with
GOSIF and CIFX (RMSE = 64.47 and 52.41 gC m
−2
yr
−1
, RD = 2.4 and 15% and r = 0.74 and 0.67,
respectively).
Temporal variability analysis
Generally, the 3D-CMCC-FEM GPP shows overall
lower correlations of the temporal (which corresponds
to the interannual variability, IAV) vs. spatial compar-
ison for the mean annual GPP. Similarly, as in the
spatial analysis, 3D-CMCC-FEM GPP shows better
agreement with GOSIF (RMSE = 231.6 gC m
−2
yr
−1
,
RD = −4.6% and r = 0.4, see Table 2 for overall statis-
tics and Figures 3, 4, 5, 6 for the maps) and the lowest
Fractional Variance between the RS-based dataset con-
sidered (FV = 0.83, see also Standard Deviation analysis
Figure S1). At seasonal level, during summer, 3D-
CMCC-FEM GPP correlates better with the
FLUXCOM (r = 0.62 and FV = 1.64) although with the
highest RMSE (182.53 gC m
−2
yr
−1
) among the RS-
based dataset considered. Conversely, in spring 3D-
CMCC-FEM GPP correlates better with CFIX
Figure 2. 3D-CMCC-FEM mean annual GPP values (gC m
−2
yr
−1
) for the period 2005–2019 at 1 km spatial resolution. White areas
indicate areas not simulated by the 3D-CMCC-FEM.
EUROPEAN JOURNAL OF REMOTE SENSING 7
(r = 0.31) and with the lowest RMSE (86.81 gC m
−2
yr
−1
) and RD (−7.9%). In winter modeled GPP values
from 3D-CMCC-FEM better correlates with CFIX
(r = 0.69) but with lower RMSE, RD and FV values
when compared with FLUXCOM (RMSE = 36.24 gC
m
−2
yr
−1
, RD = −24.1 and FV = 1.77). Also during
autumn season, the 3D-CMCC-FEM GPP better corre-
lates with CFIX (r = 0.3) and with the lowest RMSE
(57.97 gC m
−2
yr
−1
) although the lowest Relative
Difference is when compared with GOSIF (RD = 1.7%).
At the Mean Seasonal Cycle level (thus comparing
month per month average GPP values) 3D-CMCC-
FEM GPP well correlates with all RS-based dataset
with r varying from 0.95 (when compared with
CFIX) to 0.91 (when compared with GOSIF) (see
Figure 7, Table 2 and Figure S2)
SPEI
We first computed the SPEI at a time-scale of 5
months (SPEI5) representing the spring and summer
period where the correlation between the RS-based
summer GPP and SPEI was stronger. We estimate
whether the 3D-CMCC-FEM data and RS-based data
Table 1. Spatial comparison of the 3D-CMCC-FEM GPP vs. RS –
based GPP. RMSE= root mean square error (gC m
−2
yr
−1
),
RD = relative difference (%), SPAEF = SPatial EFficiency,
r = Pearson’s correlation. DJF, winter months; MAM, spring
months, JJA, summer months; SON, autumn months. In bold
values with the best agreement between 3D-CMCC-FEM GPP
and RS-based GPP.
RMSE (gC m
−2
yr
−1
) RD (%) SPAEF r
3D-CMCC-FEM GPP vs. GOSIF GPP
ANNUAL 235.81 −4.30 0.62 0.77
DJF 54.34 −31.3 −1.50 0.39
MAM 104.43 9.0 0.22 0.40
JJA 181.35 −13.7 0.61 0.79
SON 64.47 2.4 0.46 0.74
3D-CMCC-FEM GPP vs. CFIX GPP
ANNUAL 221.91 0.01 −0.07 0.59
DJF 65.26 −47.8 −1.26 0.55
MAM 103.61 −7.9 −0.31 0.28
JJA 123.78 10.7 0.38 0.80
SON 52.41 15.1 0.48 0.67
3D-CMCC-FEM GPP vs. FLUXCOM GPP
ANNUAL 477.42 40.0 0.59 0.67
DJF 39.97 −13.4 −0.53 0.50
MAM 198.47 45.9 −0.45 −0.09
JJA 203.44 37.2 0.57 0.72
SON 118.25 57.7 0.40 0.68
Figure 3. Root mean square error (RMSE, gC m
−2
yr
−1
) between mean annual 3D-CMCC-FEM GPP (gC m
−2
yr
−1
) and the RS-based
GPP: a) 3D-CMCC-FEM GPP vs. GOSIF GPP for the period 2005–2019, b) 3D-CMCC-FEM GPP vs. CFIX GPP for the period 2005–2013,
c) 3D-CMCC-FEM GPP vs. FLUXCOM GPP for the period 2005–2015; d) histogram of the relative difference, dashed lines indicate the
median value. White areas on the maps indicate areas not simulated by the 3D-CMCC-FEM.
8D. DALMONECH ET AL.
differ in GPP sensitivity toward interannual varia-
tion in SPEI by computing the slope of the regres-
sion between 3D-CMCC-FEM vs. RS-based
differences of GPP and the SPEI. Generally, results
show how, when considering CFIX and FLUXCOM,
the slope is not significantly different from 0 for
almost the entire simulation domain (Figure 8).
Such a behavior indicates that 3D-CMCC-FEM and
RS-based GPP summer anomalies respond to SPEI5
similarly. When comparing with the GOSIF the
slope is still not significant over large areas.
However, in clustered areas, corresponding to
about a 20% of the simulation domain, the slope is
positive and significant showing that the 3D-CMCC-
FEM response to aridity is stronger than the GOSIF
data (Figure 8).
Species-level comparison
Data analysis at species level reveals as 3D-CMCC-
FEM model GPP tends to correlate better for some
species for some RS-based datasets than other (Table 3
and Figure 9). 3D-CMCC-FEM GPP correlates well
for Q. cerris and Q. ilex with GOSIF (r = 0.73 and 0.82)
but with low RMSE and RD for CFIX (RMSE = 141.09
and 141.99 gC m
−2
yr
−1
, and RD = −2.69 and 3.39%,
respectively). In all cases 3D-CMCC-FEM GPP for F.
sylvatica are slightly far from results from all RS-based
datasets with better correlations with CFIX (r = 0.43)
and lower RMSE and RD values for GOSIF (193.35 gC
m
−2
yr
−1
and 3.43%). Satisfactorily correlations are
shown for C. sativa with the GPP values of
FLUXCOM (r = 0.61) but with lower RMSE and RD
with GOSIF (200.45 gC m
−2
yr
−1
and −1.71%). For
Pinus species (both as P. halpensis and P. nigra), 3D-
CMCC-FEM GPP shows to better correlates and with
lower RMSE and RD values with GOSIF (r = 0.84,
RMSE = 170.84 gC m
−2
yr
−1
and RD = −7.91% for P.
halepensis; and r = 0.65, RMSE = 223.92 gC m
−2
yr
−1
and RD = 7.85% for P. nigra) than with CFIX and
FLUXCOM (Table 3 and Figure 9).
Discussion
In this study, we applied the 3D-CMCC-FEM spatial
explicitly over a Mediterranean region characterized
Figure 4. Relative difference (RD, %) between mean annual 3D-CMCC-FEM GPP and the RS-based GPP: a) 3D-CMCC-FEM GPP vs.
GOSIF GPP for the period 2005–2019, b) 3D-CMCC-FEM GPP vs. CFIX GPP for the period 2005–2013, c) 3D-CMCC-FEM GPP vs.
FLUXCOM GPP for the period 2005–2015; d) histogram of the relative difference, dashed lines indicate the median value. White
areas on the maps indicate areas not simulated by the 3D-CMCC-FEM.
EUROPEAN JOURNAL OF REMOTE SENSING 9
by elevated bio-geographical and climatological het-
erogeneity. Initializing the forest structure, combining
ground data from the NFI and the RS-based wall-to-
wall basal area map, allows having a more realistic
initial state of the carbon pools and forest structure,
reducing, thus, uncertainties related to spin-up proce-
dures which often translate into significant differences
in the carbon fluxes (Carvalhais et al., 2008, 2010;
Lindeskog et al., 2021; Massoud et al., 2019). The
calibration of the allometric equations, to which the
model has shown to be sensitive (Collalti et al., 2019),
sets an additional constraint on the modeled initial
growing biomass, which in turn influences the simu-
lated GPP via the amount of sapwood. By means of the
use of the basal area map to build the forest character-
istics at the beginning of simulation, the model is able
to retain part of the spatial information of the initial
aboveground biomass and average forests structure,
contributing, for instance, to the very satisfactorily
spatial correlations of the modeled annual GPP with
2 out of 3 RS-based datasets. While aggregating basal
area data at the 1 km resolution might reduce random
uncertainties in basal area, additional uncertainty
might stem from the tree density data, a data which
is less constrained. Yet, performed tests (not shown
here) indicate no significant sensitivity of the simu-
lated GPP to stand density.
Overall, 3D-CMCC-FEM performances in simulat-
ing GPP when compared to other large-scale and
independent RS-based data are shown to be generally
satisfactory at both spatial and temporal scales. In
particular in the summer period across the seasons
3D-CMCC-FEM GPP show satisfactorily correlations
and low relative differences and RMSE values when
most of the vegetation activity takes place and it is thus
the most robust signal across model and RS-based
datasets. Yet, the comparative analyzed pinpointed
some important challenges in applying the 3D-
CMCC-FEM in Mediterranean areas and highlighted
sources of uncertainties explaining the residual 3D-
CMCC-FEM and RS-based data differences across RS-
based datasets, which are grouped as follows:
Figure 5. Fractional variance (FV) between mean annual 3D-CMCC-FEM GPP and the RS-based GPP: a) 3D-CMCC-FEM GPP vs.
GOSIF GPP for the period 2005–2019, b) 3D-CMCC-FEM GPP vs. CFIX GPP for the period 2005–2013, c) 3D-CMCC-FEM GPP vs.
FLUXCOM GPP for the period 2005–2015; d) histogram of the relative difference, dashed lines indicate the median value. White
areas on the maps indicate areas not simulated by the 3D-CMCC-FEM.
10 D. DALMONECH ET AL.
3D-CMCC-FEM GPP vs. RS-based GPP
The 3D-CMCC-FEM GPP is close to the GOSIF and
CFIX GPP estimates, while a general systematic positive
overestimation emerge when comparing the 3D-
CMCC-FEM GPP to FLUXCOM GPP (although for
not all species, Figure 9). The 3D-CMCC-FEM and
RS-based data differences resulting from the analyses
are here discussed in relation to the different nature of
the RS-based datasets used and their underlying algo-
rithms and adopted approaches. FLUXCOM is an up-
scaled product of the local EC-tower GPP estimates
which is often used as benchmark RS-based dataset in
model evaluation analyses (e.g. Byrne et al., 2018). Part
of its worldwide application relies on the capability to
sample in the entire climate-vegetation space. However,
some particular areas at the rear edge of the
Mediterranean forests might not be covered by this
product (Jung et al., 2020). In addition, FLUXCOM
operates also at coarse spatial resolution (Zhang et al.,
2022; Zheng et al., 2020). Compared to GOSIF and
CFIX, the spatial, seasonal and interannual variability
in FLUXCOM relies on the RS-based dataset alone,
without including any climatic drivers. Unfortunately,
climate has been shown to play a significant role in the
local, regional and global GPP. The lack of climatic
drivers in FLUXCOM, might, thus, partly explain the
systematic differences observed when compared to the
3D-CMCC-FEM GPP, which, at the opposite, showed
to be sensitive to climate (Mahnken et al., 2022).
However, the noticeable differences found, although
some good correlations for some species, between the
3D-CMCC-FEM GPP and FLUXCOM, is common to
other process-based and dynamic vegetation models
comparative studies (Jung et al., 2020; Li & Xiao,
2019b; Zhang & Ye, 2021), and recent studies found
that the GPP may be underestimated in the
FLUXCOM-GPP in temperate areas (e.g. Bacour et al.,
2019; Norton et al., 2019; Wild et al., 2022).
Figure 6. Pearson’s correlation (r) between mean annual 3D-CMCC-FEM GPP and the RS-based GPP: a) 3D-CMCC-FEM GPP vs.
GOSIF GPP for the period 2005–2019, b) 3D-CMCC-FEM GPP vs. CFIX GPP for the period 2005–2013, c) 3D-CMCC-FEM GPP vs.
FLUXCOM GPP for the period 2005–2015; d) histogram of the relative difference, dashed lines indicate the median value. Grey
areas on the maps indicate where correlations are not significant. White areas on the maps indicate areas not simulated by the 3D-
CMCC-FEM.
EUROPEAN JOURNAL OF REMOTE SENSING 11
Figure 7. Pearson’s correlation (r) between mean seasonal cycle (MSC) 3D-CMCC-FEM GPP and the RS-based GPP: a) 3D-CMCC-FEM
GPP vs. GOSIF GPP for the period 2005–2019, b) 3D-CMCC-FEM GPP vs. CFIX GPP for the period 2005–2013, c) 3D-CMCC-FEM GPP
vs. FLUXCOM GPP for the period 2005–2015, d) histogram of the r, dashed lines indicate the median value. White areas on the
maps indicate areas not simulated by the 3D-CMCC-FEM.
Table 2. Temporal comparison of the 3D-CMCC-FEM GPP vs. RS – based GPP. RMSE=
root mean square error (gC m
−2
yr
−1
); RD = relative difference (%); FV = fractional
variance; r = Pearson’s correlation. Metrics are first computed at grid cell level and
reported as the median value. DJF, winter months; MAM, spring months, JJA, summer
months; SON, autumn months; MSC, mean seasonal cycle. In bold values with the best
agreement between 3D-CMCC-FEM GPP and RS-based GPP.
RMSE (gC m
−2
yr
−1
) RD (%) FV r
3D-CMCC-FEM GPP vs. GOSIF GPP
ANNUAL 231.60 −4.6 0.83 0.40
DJF 54.64 −40.6 1.70 0.51
MAM 105.61 10.4 0.77 0.15
JJA 146.72 −12.7 0.97 0.53
SON 77.33 1.7 0.38 0.17
MSC / / / 0.91
3D-CMCC-FEM GPP vs. CFIX GPP
ANNUAL 179.55 −1.3 1.17 0.12
DJF 66.67 −58.7 1.09 0.69
MAM 86.81 −7.9 1.36 0.31
JJA 86.12 7.5 0.68 0.65
SON 57.97 15.2 0.65 0.30
MSC / / / 0.95
(Continued)
12 D. DALMONECH ET AL.
Table 2. (Continued).
RMSE (gC m
−2
yr
−1
) RD (%) FV r
3D-CMCC-FEM GPP vs. FLUXCOM GPP
ANNUAL 486.65 40.6 1.69 0.39
DJF 36.24 −24.1 1.77 0.55
MAM 209.14 49.4 1.66 0.26
JJA 182.53 37.9 1.64 0.62
SON 120.33 54.7 1.41 0.17
MSC / / / 0.92
Figure 8. Spatial distribution of the slope of the linear regression of the summer GPP residuals and SPEI5 for: a) 3D-CMCC-FEM vs.
GOSIF, b) 3D-CMCC-FEM vs. CFIX, c) 3D-CMCC-FEM vs. FLUXCOM. Grey areas indicate the slope is not significantly different from 0
at p < 0.05. White areas on the maps indicate areas not simulated by the 3D-CMCC-FEM.
Table 3. Comparison of the 3D-CMCC-FEM GPP vs. RS-based GPP at single species
level. RMSE= root mean square error (gC m
−2
yr
−1
), RD = relative differences (%), r =
Pearson’s correlation. Metrics are first computed at grid cell level and reported as the
median value. In bold values with the best agreement between 3D-CMCC-FEM GPP
and RS-based GPP.
RMSE (gC m
−2
yr
−1
) RD (%) r
3D-CMCC-FEM GPP vs. GOSIF GPP
Quercus cerris 245.66 −6.33 0.73
Fagus sylvatica 193.35 3.43 0.28
Quercus ilex 267.65 3.96 0.83
Pinus halepensis 170.84 −7.91 0.85
Pinus nigra 223.92 7.86 0.66
Castanea sativa 200.45 −1.72 0.25
3D-CMCC-FEM GPP vs. CFIX GPP
Quercus cerris 141.1 −2.7 0.60
Fagus sylvatica 477.4 31.1 0.43
Quercus ilex 141.9 3.4 0.70
Pinus halepensis 355.1 (28%) −26.8 0.65
Pinus nigra 314.5 (21%) 16.9 0.25
Castanea sativa 203.3 7.0 0.36
3D-CMCC-FEM GPP vs. FLUXCOM GPP
Quercus cerris 440.4 39.1 0.66
Fagus sylvatica 777.8 63.1 0.10
Quercus ilex 462.1 40.0 0.75
Pinus halepensis 140.7 7.2 0.68
Pinus nigra 656.1 57.4 0.57
Castanea sativa 584.1 49.0 0.61
EUROPEAN JOURNAL OF REMOTE SENSING 13
The GOSIF and the CFIX datasets, which 3D-
CMCC-FEM GPP seems to reply better than
FLUXCOM, have the advantage of having a finer spatial
resolution (~5 and 1 km versus ~8 km in FLUXCOM)
and, thus, contain more information, with the GOSIF
having a globally higher continuous coverage via the
original SIF data, which is the proxy of the photosyn-
thetic activity at canopy scales (Li & Xiao, 2019a; Sun et
al., 2017). CFIX is instead driven by a Light Use
Efficiency model driven by NDVI, which is a vegetation
index more suitable to investigate plant greenness
rather than purely photosynthetic activities, as high-
lighted by Camps-Valls (2021) and Walther et al.
(2019). This inherent characteristic for NDVI might
explain the lack of spatial correlation but also lower
spatial correlations when compared to FLUXCOM
and GOSIF. In addition, the Light Use Efficiency mod-
els are known to not saturate at increasing solar radia-
tion, but that would lead CFIX having higher GPP
values than 3D-CMCC-FEM GPP, which uses the bio-
chemical model of Farquhar et al. (1980). However,
while in some species CFIX has higher GPP values
than 3D-CMCC-FEM GPP (e.g. F. sylvatica and Q.
Ilex) in other species values are lower (P. halepensis
and Q. cerris)(Figure 9). Differences between 3D-
CMCC-FEM and CFIX may thus more probably rely
on different models’ parameterization and not just on
the different approach used to simulate photosynthesis.
Indeed, as outlined in other studies carried at global
scales, there is a higher uncertainty in interannual varia-
bility even across different RS-based datasets
(Butterfield et al., 2020; Zhang & Ye, 2021) and even
in its correct (of GPP) calculation across different mod-
els (Dunkl et al., 2023).
The differences in standard deviation between the
3D-CMCC-FEM and the RS-based GPP (see Figure
S1), despite high in absolute terms, are still compar-
able to the range shown for instance in Zhang and Ye
(2021) and Zheng et al. (2020). Similarly, the partial
lack of agreement in the winter (but this is worth also
for the spring and autumn) months that we found in
this study has been observed also in other RS-based
data and process-based models comparative analyses
(Zhang and Ye, 2021). However, this it is not surpris-
ing, at least for deciduous species, given that 3D-
CMCC-FEM simulates dominant vegetation only
and no photosynthesis when there are no leaves on
at the opposite to RS-based data which may account
for underneath vegetation and this might explain the
better correlations with evergreen species than for the
deciduous for some RD-based datasets. Spring and
autumn also depend on the spatial and temporal varia-
bility of bud breaks and leaf falls which control photo-
synthetic activity during and across the years.
However, mismatches and asynchronies for the begin-
ning and the end of the growing season largely vary
between species and RS-based datasets considered as
also found in the literature for other models (Peano et
al., 2019, 2021). In addition, the RS-based datasets
capture the fluxes embedding the entire sub-grid
variability, including the contribution of the vegeta-
tion not simulated by the model (e.g. crops, maquis
and understory which in any case comprise about one
fifth of the entire vegetation area only), in particular in
areas where the forest cover is indeed low because of
low stand density, such as in the plain areas of the
region under study or in degraded areas. Spatial scale
mismatch between data and models may explain, thus,
likely part of the low performances in some areas of
the region. Zhao and Zhu (2022) showed how GOSIF
has apparently a similar interannual variability and
trends when compared to the TRENDY simulations
(an ensemble of simulations from Dynamic Global
Vegetation Models; Sitch, 2015), as similarly as we
found here using a process-based, stand-level, forest
model.
On the other hand, some have raised doubts and
concerns on the robustness of the relationship
Figure 9. Mean annual 3D-CMCC-FEM GPP vs. RS-based GPP scatter plot (gC m
−2
yr
−1
) at the species-level: a) the 3D-CMCC-FEM
GPP vs. GOSIF b) 3D-CMCC-FEM GPP vs. CFIX and c) 3D-CMCC-FEM GPP vs. FLUXCOM estimates (black line is the 1:1 line, dashed
line is the linear fit). Each point represents data at grid cell level, different colors indicate the different species considered.
14 D. DALMONECH ET AL.
between GPP and remotely sensed sun-induced fluor-
escence (SIF) (Bacour et al., 2019; Chen et al., 2021;
Wohlfahrt et al., 2018) showing that the SIF might
overestimate the GPP values at least in temperate
broadleaved forests (Qiu et al., 2020). However, to
our knowledge, there are no specific indication that
this retains also for Mediterranean forests. Yet the SIF
data from RS remain one of the best proxies for photo-
synthesis over large areas to date (Li & Xiao, 2019a;
Sun et al., 2017), making the GOSIF estimates poten-
tially the more robust estimates compared to both
CFIX and FLUXCOM.
Differences in the original spatial resolution of RS-
based data or land cover used to drive the GOSIF and
CFIX datasets, might additionally contribute to
explain the 3D-CMCC-FEM GPP and RS-based data-
sets – but also between RS-based datasets – differ-
ences. Interestingly, when compared RS-based
dataset vs. RS-based dataset GPP (e.g. GOSIF vs.
FLUXCOM), and not just 3D-CMCC-FEM GPP vs.
RS-based dataset, low correlations and discrepancies
(in both in the relative and in the absolute sense),
although minor than against 3D-CMCC-FEM, across
the results (see Table S3 and Table S4), at both the
spatial and the temporal scales (including at the spe-
cies level, see for example the large RMSE for all
deciduous species in Table S4), have been found.
That was, however, expected given that RS-based
methods are diagnostic and of similar nature tools
rather than inherently prognostic (and with greater
uncertainties) tools as potentially all models are. A
recent large review on RS-based products intercom-
parison confirmed the large variability and different
sensitivity to climatic factors between different
approaches and criteria when compared at site-level
(Sun et al., 2019).
3D-CMCC-FEM-related uncertainties
The 3D-CMCC-FEM model has been extensively eval-
uated at the site level all over Europe and evaluated
against measurements of carbon fluxes from the
FLUXNET database (Collalti et al., 2016, 2018;
Marconi et al., 2017). The evaluation has been carried
out in the past over a broad spectrum of climate, forest
management, and species at stand level (e.g. Dalmonech
et al., 2022; Testolin et al., 2023). In some recent model
inter-comparison studies, the 3D-CMCC-FEM was
shown to be able to simulate, among other things, C
fluxes, e.g. GPP, and key structural variables, e.g. dia-
meter, basal area, etc, with very good performance when
compared to measured data and to other state-of-the-art
forest models (Engel et al., 2021; Mahnken et al., 2022).
However, when applying the 3D-CMCC-FEM at a dif-
ferent spatial scales model parameters, structural, and
input-related uncertainties might amplify or even buffer
the error in simulating GPP as observed for other models
(e.g. Dalmonech et al., 2015; Dunkl et al., 2023; Zhang et
al., 2022). In any case, the overall, 3D-CMCC-FEM GPP
(~600 - ~2200 gC m
−2
yr
−1
) is well in the bounds of the
~ 600 - ~2500 gC m
−2
yr
−1
GPP values described in
Collalti and Prentice (2019) for temperate deciduous
and coniferous forests in Europe.
The seasonal cycle of GPP describes the seasonal
pattern of carbon gross assimilation by plants and the
beginning and end of the growing season for the decid-
uous forests. For two out of the three datasets used here
the 3D-CMCC-FEM may simulate a slightly earlier
beginning of the growing season (e.g. for Q. cerris)
(Figure S2). This might be partially attributable to the
budburst parameterization, which is based on the
Thermic sum and the growing degree days metric
which is a trigger for the leaf flushing. Such a parameter
is kept constant (as in many other models, see Collalti et
al., 2019; Peano et al., 2019; Peaucelle et al., 2019) as a
species-specific parameter, irrespective of any local cli-
matic adaptation. The observed discrepancy in the
beginning of the growing season, compared to the
delay projected by the 3D-CMCC-FEM model, may
be partly attributed also to the influence of under-
growth vegetation (such as grass or shrubs) on RS-
based data. These understory plants frequently begin
photosynthesis earlier, a phenomenon not accounted
for by the 3D-CMCC-FEM model. The seasonality of
the GPP in the evergreen species is instead apparently
more shaped by the direct environmental effect on the
photosynthetic process at a seasonal time scale (and less
affected by underneath vegetation), suggesting that the
3D-CMCC-FEM GPP and RS-based datasets differ-
ences might be also attributable to some bias in the
meteorological data used or for the other physiological
parameters adopted or in the below-canopy vegetation.
Indeed, model ecophysiological parameter values are
derived from the literature, yet not calibrated for the
specific sites or regions, in order to allow the 3D-
CMCC-FEM general applicability as reported in
Mahnken et al. (2022). In addition, other potential
source of uncertainty, although minor, for the apparent
mismatches in the beginning and in the end of the
growing season between 3D-CMCC-FEM and RS-
based datasets may stem from the different temporal
resolutions, indeed, e.g. FLUXCOM is an 8-day product
while 3D-CMCC-FEM is a daily one.
To capture the summer GPP sensitivity to hydro-
logical variability, i.e. wetter or drier conditions, we
use a simple GPP-SPEI diagnostics. The 3D-CMCC-
FEM was first shown to be able to simulate compar-
ably to the RS-based datasets the interannual variabil-
ity (see Table 2) and then of the summer GPP and
aridity signal over a large extent (see Figure 8). In large
areas of the region, the 3D-CMCC-FEM and the RS-
based datasets have similar responses in terms of GPP-
SPEI5. Yet, in some areas, a too-strong model
response to SPEI5 also emerges, indicating that the
EUROPEAN JOURNAL OF REMOTE SENSING 15
3D-CMCC-FEM has a higher summer GPP-response
to negative SPEI5 (i.e. drier conditions) compared to
the RS-based datasets. In particular, results indicate a
higher model sensitivity of GPP to SPEI for oaks-
dominated forests clustered on higher elevations.
Oak’s species are known to be an isohydric species, i.
e. they weakly regulate stomatal openness under
drought conditions and are more resilient to drought
(Castellaneta et al., 2022; Ripullone et al., 2020), and it
is possible that the high sensitivity of GPP to SPEI5 in
the 3D-CMCC-FEM is a result either of a too strong
control of VPD on stomatal conductance (Grossiord
et al., 2020) or of a too strong soil moisture control on
the photosynthesis (Crow et al., 2020; Fang et al.,
2021). Summer period is the season where most of
the annual vegetation activity takes place and reaches
its maximum, and which variability is the most pro-
minent feature in Mediterranean areas. As a matter of
fact, in a previous study (see Collalti et al., 2016) was
shown how, at site level in beech and coniferous for-
ests, the 3D-CMCC-FEM was able to sufficiently
simulate the sign of the year-to-year variability of the
annual GPP anomalies, and in particular in years
affected by important summer drought (i.e. 2003).
Yet interannual variability is the most uncertain signal
event when comparing different data and often con-
sidered as an “acid test” in vegetation modeling
(Collalti et al., 2016; Dunkl et al., 2023; Keenan et
al., 2012; Yang et al., 2022).
However, stand-level studies showed how the 3D-
CMCC-FEM model realistically simulates the response
of GPP to atmospheric VPD and temperature (Mahnken
et al., 2022). In the model soil hydrology is simulated via
a single-layer bucket model and this might be another
factor contributing to the stronger than observed mod-
eled response, as it has been shown as overall that the
majority of models with a soil bucket hydrology tend to
limit GPP more than models with other soil conceptual e.
g. multilayered schemes (Hanson et al., 2004). This
higher sensitivity and the apparent stronger drop of
modeled summer GPP might partially explain the higher
standard deviation of the modeled signal compared to
RS-based datasets (see Figure S1). One potential addi-
tional explanation to the modeled stronger GPP sensitiv-
ity to SPEI, would be the detected, although the narrow,
earlier onset of the growing season in the model com-
pared to the GOSIF dataset. In fact, leaf development not
only drives how earlier or later the uptake of carbon from
the atmosphere starts in the season, but also how earlier/
later other processes, such as the leaf transpiration occurs
in spring, potentially affecting the summer GPP response
(Bastos et al., 2021; Chen et al., 2023; Peano et al., 2019).
Meteorological and soil data uncertainties
Low accuracy in the reconstructed meteorological for-
cing (see Bandhauer et al., 2022) might play an additional
important role in studies that are conducted at high
spatial resolution. The impact of the meteorological
data used in data-driven and PBFMs was already
shown to be important to accurately simulate GPP (e.g.
Wu et al., 2017; Zhang & Ye, 2022; Zhang et al., 2022).
The 3D-CMCC-FEM was shown to be able to simulate
GPP closer to observations at the site level in northern,
central and southern Europe, where the downscaling and
bias correction of the climate data was facilitated by the
local geography (Collalti et al., 2018; Testolin et al., 2023).
Potential positive biases in the original E-OBS tem-
perature dataset, as its accuracy relies also on the den-
sity of the termopluviometric stations and altitudinal
cover, might translate into higher temperatures at
higher elevations where, indeed, there are few stations
as in southern Italy. A positive bias in temperature over
areas where the temperature should be a limiting factor
for vegetation may partially explain the high modeled
GPP values in beech forests leading to an anticipated
onset or at thermic stress during the summer. Beech in
the Italian Apennine Mountain regions represents the
upper limit of forest vegetation, and other studies (Fang
& Lechowicz, 2006; Marchand et al., 2023) showed that
the most limiting factor for the presence and growth of
the beech in Europe is, indeed, temperature. The simu-
lated values of GPP in beech-dominated forests are, on
average higher than the three reference dataset, with
simulated values in some areas larger than 2000 gC m
−2
yr
−1
. However these values are overestimated as
observed average annual values of GPP in a beech forest
in central Italy, a monitoring site with fluxes measured
by means of the eddy covariance technique, are in the
order of 1600 gC m
−2
yr
−1
with range between ~ 550
and 2500 gC m
−2
yr
−1
from Reyer et al. (2020). When
applied in the same beech forest, the 3D-CMCC-FEM
was shown to be able to fall within the observed range
of GPP (Mahnken et al., 2022), indicating how the
source of bias might be indeed attributable to uncer-
tainties stemming from the up-scaling and affecting
mostly the medium to higher elevation areas.
Simulated values for the deciduous oaks are indeed
more comparable to EC-estimates in a site in central
Italy (Roccarespampani, www.fluxnet.com, Tedeschi
et al., 2006), with annual values of the order of 1600
gC m
−2
yr
− 1
for model and observations. However, a
west-east gradient in the model-data difference is
apparent at the regional level when comparing the
3D-CMCC-FEM to CFIX and GOSIF. In the former
case, differences correlate with elevation and tempera-
ture, very likely a results of the lack of accuracy in the
E-OBS dataset. Spatial differences in the 3D-CMCC-
FEM GPP emerge in this study, which cannot be
explained only by uncertainties in the modeling
setup alone. The amount of water available for plants
is determined in the 3D-CMCC-FEM by the balance
between the precipitation (the inflow) and the evapo-
transpiration (the outflow) as well as the soil
16 D. DALMONECH ET AL.
characteristics, such as soil depth and texture. The
water availability and the soil texture translate into
the matric potential, which directly affects the stoma-
tal regulation, leaf transpiration and the photosynth-
esis. When the model is applied spatially in a
Mediterranean area with orographic and geographical
complexity, the capability of the climate to retain the
spatial and temporal variability of precipitation might
have important effects in modulating the modeled
spatial and temporal GPP variability (see Zhang &
Ye, 2022), and this will be the object of further
investigations.
The last source of uncertainties in the 3D-CMCC-
FEM GPP and RS-based GPP might stem from the soil
characteristics and, in particular, the soil depth. The
dataset used in the analyses as a proxy of depth avail-
able to root does not show any sensible spatial varia-
bility, e.g. with elevation, which is not realistic
considering the topography and plasticity of plants
roots (Fan et al., 2017; Stocker et al., 2023). The 3D-
CMCC-FEM at this scale of application would benefit
from coupling with more advanced hydrological
scheme or a soil depth parameterization, to better
couple the vegetation with the effective soil water
availability (Kollet & Maxwell, 2008; Niu et al., 2011)
or from soil moisture assimilation from remote sen-
sing products (Crow et al., 2020; Kumar et al., 2014).
Limitations and further considerations
In this study we did not consider the land use change
and the model does not simulate any spatial interaction
from neighboring grid cells, not allowing for example
simulating forest expansion. We additionally aggregated
the forest species according to macro classes, hence
neglecting intra-specific differences, if any. However,
some compromises between input data requirements
and time for model runs were needed to keep simula-
tions and calculations feasible. In particular, at the spatial
resolution of the study, we were more interested in
capturing trends and spatial variability in GPP than the
finest sub-grid variability within the forest ecosystem.
No nitrogen cycle is included; however, the pre-
cipitation is commonly identified as the main limit-
ing factor for photosynthesis in Mediterranean areas
(Flexas et al., 2014; Fyllas et al., 2017; Keenan et
al., 2009).
Conclusions and outlook
PBFMs offer a complementary tool to ground forest
inventory networks and satellite-based records in mon-
itoring carbon sequestration forest growth and the cli-
mate change impacts on forests, on many other variables
which are otherwise difficult to measure extensively or
monitor continuously. However, any model’s reliability
needs to be verified and tested even in high heterogeneity
contexts and not only to site-level ones.
With this aim, we used RS-based data to initialize,
apply, and test for the first time the biogeochemical,
biophysical, process-based model 3D-CMCC-FEM on
a regular grid at high spatial resolution under typical
Mediterranean climate. We compare the obtained gross
primary production over a large area (~80% of the
Basilicata region) against a suite of different-in-nature
independent RS-based data. In spite of the simplified
initial forest setup and the underlined uncertainties, the
3D-CMCC-FEM was shown capable of capturing both
the spatial and the temporal variability of the RS-based
data, even at species-level. Further tests are needed, yet
the very promising results open the possibility of using
the PBFMs to investigate the spatio-temporal dynamics
of forest growth over larger spatial scales and under
drought conditions and future climate scenarios,
shaped by the spatial climatic and ecological heteroge-
neity such as in the Mediterranean areas.
Acknowledgments
The work was carried out in the framework of the project
‘Advanced EO Technologies for studying Climate Change
impacts on the environment – OT4CLIMA’ (D.D. 2261 -
6.9.2018, PON R&I 2014–2020 and FSC). This work has
been partially supported by MIUR Project (PRIN 2020)
“Unraveling interactions between WATER and carbon
cycles during drought and their impact on water
resources and forest and grassland ecosySTEMs in the
Mediterranean climate (WATERSTEM)” (Project num-
ber: 20202WF53Z), “WAFER” at CNR (Consiglio
Nazionale delle Ricerche) and D.D., E.V., A.C. were sup-
ported by PRIN 2020 (cod 2020E52THS) - Research
Projects of National Relevance funded by the Italian
Ministry of University and Research entitled: “Multi-
scale observations to predict Forest response to pollution
and climate change” (MULTIFOR, project number
2020E52THS). D.D., E.V., A.C. acknowledge also funding
by the project OptForEU H2020 research and innovation
programme under grant agreement No. 101060554. We
also acknowledge the project funded under the National
Recovery and Resilience Plan (NRRP), Mission 4
Component 2 Investment 1.4 - Call for tender No. 3138
of 16 December 2021, rectified by Decree n.3175 of 18
December 2021 of Italian Ministry of University and
Research funded by the European Union –
NextGenerationEU under award Number: Project code
CN_00000033, Concession Decree No. 1034 of 17 June
2022 adopted by the Italian Ministry of University and
Research, CUP B83C22002930006, Project title “National
Biodiversity Future Centre - NBFC”. We acknowledge the
E-OBS dataset from the EU-FP6 project UERRA (https://
www.uerra.eu) and the Copernicus Climate Change
Service, and the data providers in the ECA&D project
(https://www.ecad.eu)”. The FLUXCOM products were
obtained from the Data Portal at https://www.fluxcom.
org/. D.D. thanks C. Trotta for valuable discussions about
tree allometric relationships and thanks M. Willeit and
E. Grieco for providing comments on a previous draft of
the manuscript.
EUROPEAN JOURNAL OF REMOTE SENSING 17
Disclosure statement
No potential conflict of interest was reported by the author(s).
Funding
This work was supported by the Consiglio Nazionale delle
Ricerche ['WAFER' @ CNR]; Horizon 2020 Framework
Programme [OptForEU H2020 No. 101060554]; Ministero
dell’Istruzione, dell’Università e della Ricerca
['WATERSTEM ' Project number: 20202WF53Z].
ORCID
D. Dalmonech http://orcid.org/0000-0002-1932-5011
E. Vangi http://orcid.org/0000-0002-9772-2258
M. Chiesi http://orcid.org/0000-0003-3459-6693
G. Chirici http://orcid.org/0000-0002-0669-5726
L. Fibbi http://orcid.org/0000-0001-6985-6809
F. Giannetti http://orcid.org/0000-0002-4590-827X
G. Marano http://orcid.org/0000-0003-2600-984X
C. Massari http://orcid.org/0000-0003-0983-1276
A. Nolè http://orcid.org/0000-0002-5144-3421
J. Xiao http://orcid.org/0000-0002-0622-6903
A. Collalti http://orcid.org/0000-0002-4980-8487
Author contribution
Daniela Dalmonech: Conceptualization, Data curation,
Formal analysis, Investigation, Methodology, Resources,
Software, Validation, Visualization, Writing – original
draft. Elia Vangi, Gherardo Chirici, and Francesca
Giannetti: Resources, Writing – review & editing.
Jingfend Xiao: Resources, Writing – review & editing;
Marta Chiesi and Luca Fibbi: Resources, Writing – review
& editing. Gina Marano: Software, Writing – review &
editing Christian Massari: Writing – review & editing.
Angelo Nole: Resources, Writing – review & editing.
Alessio Collalti: Conceptualization, Formal analysis,
Investigation, Methodology, Resources, Software,
Validation, Visualization, Writing – review & editing,
Funding aquisition.
Data availability statement
The 3D-CMCC-FEM model code version 5.6 is publicly
available under the GNU General Public Licence v3.0 (GPL)
and can be found on the GitHub platform at: https://github.
com/Forest-Modelling-Lab/3D-CMCC-FEM). All data and
model executable, and scripts to perform analyses and figures
presented in this work are provided open access in the
Zenodo server (https://doi.org/10.5281/zenodo.8060401).
Correspondence and requests for additional materials should
be addressed to the corresponding author.
References
Allen, C. D., Breshears, D. D., & McDowell, N. G. (2015).
On underestimation of global vulnerability to tree mor-
tality and forest die-off from hotter drought in the
anthropocene. Ecosphere, 6(8), 1–55. https://doi.org/10.
1890/ES15-00203.1
Bacour, C., Maignan, F., MacBean, N., Porcar-Castell,
A., Flexas, J., Frankenberg, C., Peylin, P., Chevallier,
F., Vuichard, N., & Bastrikov, V. (2019). Improving
estimates of gross primary productivity by assimilat-
ing solar-induced fluorescence satellite retrievals in a
terrestrial biosphere Model using a process-based SIF
Model. Journal of Geophysical Research:
Biogeosciences, 124(11), 3281–3306. https://doi.org/
10.1029/2019JG005040
Bandhauer, M., Isotta, F., Lakatos, M., Lussana, C., Båserud,
L., Izsák, B., Szentes, O., Tveito, O. E., & Frei, C. (2022).
Evaluation of daily precipitation analyses in E-OBS
(v19.0e) and ERA5 by comparison to regional high-reso-
lution datasets in European regions. International Journal
of Climatology, 42(2), 727–747. https://doi.org/10.1002/
joc.7269
Barbati, A., Corona, P., & Marchetti, M. (2007). A forest
typology for monitoring sustainable forest management:
The case of European forest types. Plant Biosystems - an
International Journal Dealing with All Aspects of Plant
Biology, 141(1), 93–103. https://doi.org/10.1080/
11263500601153842
Bastos, A., Orth, R., Reichstein, M., Ciais, P., Viovy, N., Zaehle,
S., Anthoni, P., Arneth, A., Gentine, P., Joetzjer, E., Lienert, S.,
Loughran, T., McGuire, P. C., Sungmin, O., Pongratz, J., &
Sitch, S. (2021). Vulnerability of European ecosystems to two
compound dry and hot summers in 2018 and 2019. Earth
System Dynamics, 12(4), 1015–1035. https://doi.org/10.5194/
esd-12-1015-2021
Bivand, R., Keitt, T., Rowlingson, B., Pebesma, E., Sumner,
M., Hijmans, R., Baston, D., Rouault, E., Warmerdam, F.,
Ooms, J., & Rundel, C. (2015). Package ‘rgdal’. Bindings
for the Geospatial Data Abstraction Library. Available
online: https://cran.r-project.org/web/packages/rgdal/
index.html
Bugmann, H., & Seidl, R. (2022). The evolution, complexity
and diversity of models of long-term forest dynamics.
Journal of Ecology, 110(10), 2288–2307. https://doi.org/
10.1111/1365-2745.13989
Butterfield, Z., Buermann, W., & Keppel-Aleks, G. (2020).
Satellite observations reveal seasonal redistribution of
northern ecosystem productivity in response to interann-
ual climate variability. Remote Sensing of Environment,
242(June 2019), 111755. https://doi.org/10.1016/j.rse.
2020.111755
Byrne, B., Wunch, D., Jones, D. B. A., Strong, K., Deng, F.,
Baker, I., Köhler, P., Frankenberg, C., Joiner, J., Arora, V.
K., Badawy, B., Harper, A. B., Warneke, T., Petri, C., Kivi,
R., & Roehl, C. M. (2018). Evaluating GPP and respira-
tion estimates over northern midlatitude ecosystems
using solar-induced fluorescence and atmospheric CO 2
measurements. Journal of Geophysical Research:
Biogeosciences, 123(9), 2976–2997. https://doi.org/10.
1029/2018JG004472
Camps-Valls, G. (2021). A unified vegetation index for
quantifying the terrestrial biosphere. Science Advances, 7
(9). https://doi.org/10.1126/sciadv.abc7447
Carvalhais, N., Reichstein, M., Ciais, P., Collatz, G. J.,
Mahecha, M. D., Montagnani, L., Papale, D., Rambal, S.,
& Seixas, J. (2010). Identification of vegetation and soil
carbon pools out of equilibrium in a process model via
eddy covariance and biometric constraints. Global
Change Biology, 16(10), 2813–2829. https://doi.org/10.
1111/j.1365-2486.2010.02173.x
Carvalhais, N., Reichstein, M., Seixas, J., Colltaz, G. J.,
Pereira, J. S., Berbigier, P., Carrara, A., Granier, A.,
Montagnani, L., Papale, D., Rambal, S., Sanz, M. J., &
18 D. DALMONECH ET AL.
Valentini, R. (2008). Implications of the carbon cycle
steady state assumption for biogeochemical modeling
performance and inverse parameter retrieval. Global
Biogeochemical Cycles, 22(2), 1–16. https://doi.org/10.
1029/2007GB003033
Castellaneta, M., Rita, A., Camarero, J. J., Colangelo, M., &
Ripullone, F. (2022). Declines in canopy greenness and
tree growth are caused by combined climate extremes
during drought-induced dieback. Science of the Total
Environment, 813, 152666. https://doi.org/10.1016/j.scito
tenv.2021.152666
Chen, Keski-Saari, S., Kontunen-Soppela, S., Zhu, X., Zhou,
X., Hänninen, H., Pumpanen, J., Mola-Yudego, B., Wu,
D., & Berninger, F. (2023). Immediate and carry-over
effects of late-spring frost and growing season drought
on forest gross primary productivity capacity in the
Northern Hemisphere. Global Change Biology, 2022(14),
3924–3940. October. https://doi.org/10.1111/gcb.16751
Chen, A., Mao, J., Ricciuto, D., Lu, D., Xiao, J., Li, X.,
Thornton, P. E., & Knapp, A. K. (2021). Seasonal changes
in GPP/SIF ratios and their climatic determinants across
the Northern Hemisphere. Global Change Biology, 27(20),
5186–5197. https://doi.org/10.1111/gcb.15775
Chiesi, M., Moriondo, M., Maselli, F., Gardin, L., Fibbi, L.,
Bindi, M., & Running, S. W. (2010). Simulation of
Mediterranean forest carbon pools under expected envir-
onmental scenarios. Canadian Journal of Forest Research,
40(5), 850–860. https://doi.org/10.1139/X10-037
Chirici, G. (2016). Estimating daily forest carbon fluxes
using a combination of ground and remotely sensed
data. JGR Biogeosciences, 121(2), 266–279. https://doi.
org/10.1002/2015JG003019
Chirici, G., Giannetti, F., McRoberts, R. E., Travaglini,
D., Pecchi, M., Maselli, F., Chiesi, M., & Corona, P.
(2020). Wall-to-wall spatial prediction of growing
stock volume based on Italian National Forest
Inventory plots and remotely sensed data.
International Journal of Applied Earth Observation
and Geoinformation, 84(September 2019), 101959.
https://doi.org/10.1016/j.jag.2019.101959
Collalti, A., Biondo, C., Buttafuoco, G., Maesano, M.,
Caloiero, T., Lucà, F., Pellicone, G., Ricca, N., Salvati,
R., Veltri, A., Scarascia-Mugnozza, G., & Matteucci, G.
(2017). Simulation, calibration and validation protocols
for the model 3D-CMCC-CNR-FEM: A case study in the
Bonis’ watershed (Calabria, Italy). Forest@ - Rivista di
Selvicoltura ed Ecologia Forestale, 14(4), 247–256.
https://doi.org/10.3832/efor2368-014
Collalti, A., Dalmonech, D., Marano, G., Vangi, E., Puchi, P.,
Grieco, E., & Orrico, M. (2022). “3D-CMCC-FEM User’s
Guide”. https://doi.org/10.32018/3D-CMCC-FEM-2022
Collalti, A., Marconi, S., Ibrom, A., Trotta, C., Anav, A.,
D’andrea, E., Matteucci, G., Montagnani, L., Gielen,
B., Mammarella, I., Grünwald, T., Knohl, A.,
Berninger, F., Zhao, Y., Valentini, R., & Santini, M.
(2016). Validation of 3D-CMCC Forest Ecosystem
Model (v.5.1) against eddy covariance data for 10
European forest sites. Geoscientific Model
Development, 9(2), 479–504. https://doi.org/10.5194/
gmd-9-479-2016
Collalti, A., Perugini, L., Santini, M., Chiti, T., Nolè, A.,
Matteucci, G., & Valentini, R. (2014). A process-based
model to simulate growth in forests with complex struc-
ture: Evaluation and use of 3D-CMCC forest ecosystem
model in a deciduous forest in Central Italy. Ecological
Modelling, 272, 362–378. https://doi.org/10.1016/j.ecol
model.2013.09.016
Collalti, A., & Prentice, I. C. (2019). Is NPP proportional to
GPP? Waring’s hypothesis twenty years on. Tree
Physiology, 39(8), 1473–1483. https://doi.org/10.1093/
treephys/tpz034
Collalti, A., Thornton, P. E., Cescatti, A., Rita, A., Borghetti,
M., Nolè, A., Trotta, C., Ciais, P., & Matteucci, G. (2019).
The sensitivity of the forest carbon budget shifts across
processes along with stand development and climate
change. Ecological Applications, 29(2), 1–18. https://doi.
org/10.1002/eap.1837
Collalti, A., Tjoelker, M. G., Hoch, G., Mäkelä, A.,
Guidolotti, G., Heskel, M., Petit, G., Ryan, M. G.,
Battipaglia, G., Matteucci, G., & Prentice, I. C. (2020).
Plant respiration: Controlled by photosynthesis or bio-
mass? Global Change Biology, 26(3), 1739–1753. https://
doi.org/10.1111/gcb.14857
Collalti, A., Trotta, C., Keenan, T. F., Ibrom, A., Bond-
Lamberty, B., Grote, R., Vicca, S., Reyer, C. P. O.,
Migliavacca, M., Veroustraete, F., Anav, A., Campioli,
M., Scoccimarro, E., Šigut, L., Grieco, E., Cescatti, A., &
Matteucci, G. (2018). Thinning can reduce losses in car-
bon use efficiency and carbon stocks in managed forests
under warmer climate. Journal of Advances in Modeling
Earth Systems, 10(10), 2427–2452. https://doi.org/10.
1029/2018MS001275
Cornes, R. C., van der Schrier, G., van den Besselaar, E. J. M.,
& Jones, P. D. (2018). An ensemble version of the E-OBS
temperature and precipitation data sets. Journal of
Geophysical Research: Atmospheres, 123(17), 9391–9409.
https://doi.org/10.1029/2017JD028200
Costantini, G., Bellotti, A., Mancino, G., & Borghetti, M.
(2006). Ferrara - Carta Forestale della Basilicata. Atlante -
Re- gione Basilicata, Istituto Nazionale di Economia
Agraria.
Crow, W. T., Gomez, C. A., Sabater, J. M., Holmes, T., Hain, C.
R., Lei, F., Dong, J., Alfieri, J. G., & Anderson, M. C. (2020).
Soil moisture–evapotranspiration overcoupling and l-band
brightness temperature assimilation: Sources and forecast
implications. Journal of Hydrometeorology, 21(10), 2359–
2374. https://doi.org/10.1175/JHM-D-20-0088.1
Dalmonech, D., Marano, G., Amthor, J. S., Cescatti, A.,
Lindner, M., Trotta, C., & Collalti, A. (2022). Feasibility
of enhancing carbon sequestration and stock capacity in
temperate and boreal European forests via changes to
management regimes. Agricultural and Forest
Meteorology, 327(April), 109203. https://doi.org/10.
1016/j.agrformet.2022.109203
Dalmonech, D., Zaehle, S., J, S. G., Brovkin, V., Reick, C., &
Schnur, R. (2015). Separation of the effects of land and
climate Model errors on simulated contemporary land
carbon cycle trends in the MPI earth system Model ver-
sion 1*. Journal of Climate, 28(1), 272–291. https://doi.
org/10.1175/JCLI-D-13-00593.110.1175/JCLI-D-13-
00593.s1
D’Andrea, E., Rezaie, N., Prislan, P., Gričar, J., Collalti, A.,
Muhr, J., & Matteucci, G. (2020). Frost and drought:
Effects of extreme weather events on stem carbon
dynamics in a Mediterranean beech forest. Plant Cell
and Environment, 43(10), 2365–2379. https://doi.org/10.
1111/pce.13858
D’Andrea, E., Scartazza, A., Battistelli, A., Collalti, A.,
Proietti, S., Rezaie, N., Matteucci, G., Moscatello, S., &
Meinzer, F. (2021). Unravelling resilience mechanisms
in forests: Role of non-structural carbohydrates in
responding to extreme weather events. Tree
Physiology, 41(10), 1808–1818. https://doi.org/10.1093/
treephys/tpab044
EUROPEAN JOURNAL OF REMOTE SENSING 19
De Marco, A., Sicard, P., Feng, Z., Agathokleous, E., Alonso,
R., Araminiene, V., Augustatis, A., Badea, O., Beasley, J.
C., Branquinho, C., Bruckman, V. J., Collalti, A., David-
Schwartz, R., Domingos, M., Du, E., Garcia Gomez, H.,
Hashimoto, S., Hoshika, Y., Jakovljevic, T., McNulty, S.,
and Paoletti, E. (2022). Strategic roadmap to assess forest
vulnerability under air pollution and climate change.
Global Change Biology, 28(17), 5062–5085. https://doi.
org/10.1111/gcb.16278
De Pury, D. G. G., & Farquhar, G. D. (1997). Simple scaling
of photosynthesis from leaves to canopies without the
errors of big-leaf models. Plant, Cell and Environment,
20(5), 537–557. https://doi.org/10.1111/j.1365-3040.1997.
00094.x
Dosio, A., & Paruolo, P. (2011). Bias correction of the
ENSEMBLES high-resolution climate change projections
for use by impact models: Evaluation on the present
climate. Journal of Geophysical Research Atmospheres,
116(16), 1–22. https://doi.org/10.1029/2011JD015934
Dubrovský, M., Hayes, M., Duce, P., Trnka, M., Svoboda,
M., & Zara, P. (2014). Multi-GCM projections of future
drought and climate variability indicators for the
Mediterranean region. Regional Environmental Change,
14(5), 1907–1919. https://doi.org/10.1007/s10113-013-
0562-z
Dunkl, I., Lovenduski, N., Collalti, A., Arora, V. K., Ilyina,
T., & Brovkin, V. (2023). Gross primary productivity and
the predictability of CO
2
: More uncertainty in what we
predict than how well we predict it. Biogeosciences, 20
(16), 3523–3538. https://doi.org/10.5194/bg-20-3523-
2023
Engel, M., Vospernik, S., Toigo, M., Morin, X., Tomao, A.,
Trotta, C., Steckel, M., Barbati, A., Pretzsch, N. A., Del
Rio, H., Skrzyszewski, M., Ponette, J., Lof, Q., Jansons, M.,
& Brazaitis, A. (2021). Simulating the effects of thinning
and species mixing on stands of oak (Quercus petraea
(Matt.) Liebl./Quercus robur L.) and pine (Pinus sylvestris
L.) across Europe. Ecological Modelling, 442(15), 109406.
https://doi.org/10.1016/j.ecolmodel.2020.109406
Fang, J., & Lechowicz, M. J. (2006). Climatic limits for the
present distribution of beech (Fagus L.) species in the
world. Journal of Biogeography, 33(10), 1804–1819.
https://doi.org/10.1111/j.1365-2699.2006.01533.x
Fang, Y., Leung, L. R., Wolfe, B. T., Detto, M., Knox, R. G.,
McDowell, N. G., Grossiord, C., Xu, C., Christoffersen, B.
O., Gentine, P., Koven, C. D., & Chambers, J. Q. (2021).
Disentangling the effects of vapor pressure deficit and soil
water availability on canopy conductance in a seasonal
tropical forest during the 2015 El niño drought. Journal of
Geophysical Research: Atmospheres, 126(10), 1–20.
https://doi.org/10.1029/2021JD035004
Fan, Y., Miguez-Macho, G., Jobbágy, E. G., Jackson, R. B., &
Otero-Casal, C. (2017). Hydrologic regulation of plant
rooting depth. Proceedings of the National Academy of
Sciences of the United States of America, 114(40), 10572–
10577. https://doi.org/10.1073/pnas.1712381114
Farquhar, G. D., Caemmerer, S., & Berry, J. A. (1980). A
biochemical model of photosynthetic CO
2
assimilation in
leaves of C
3
species. Planta, 149(1), 78-90–90. https://doi.
org/10.1007/BF00386231
Fernández-Martínez, M., Peñuelas, J., Chevallier, F., Ciais,
P., Obersteiner, M., Rödenbeck, C., Sardans, J., Vicca, S.,
Yang, H., Sitch, S., Friedlingstein, P., Arora, V. K., Goll,
D. S., Jain, A. K., Lombardozzi, D. L., McGuire, P. C., &
Janssens, I. A. (2023). Diagnosing destabilization risk in
global land carbon sinks.Nature, 615(November 2021).
https://doi.org/10.1038/s41586-023-05725-1
Fibbi, L., Chiesi, M., Moriondo, M., Bindi, M., Chirici, G.,
Papale, D., Gozzini, B., & Maselli, F. (2016). Correction of
a 1 km daily rainfall dataset for modelling forest ecosys-
tem processes in Italy. Meteorological Applications, 23(2),
294–303. https://doi.org/10.1002/met.1554
Flexas, J., Diaz-Espejo, A., Gago, J., Gallé, A., Galmés, J.,
Gulías, J., & Medrano, H. (2014). Photosynthetic limita-
tions in Mediterranean plants: A review. Environmental
and Experimental Botany, 103, 12–23. https://doi.org/10.
1016/j.envexpbot.2013.09.002
Francini, S., McRoberts, R. E., Giannetti, F., Marchetti, M.,
Scarascia Mugnozza, G., & Chirici, G. (2021). The three
Indices three Dimensions (3I3D) algorithm: A new
method for forest disturbance mapping and area estima-
tion based on optical remotely sensed imagery.
International Journal of Remote Sensing, 42(12), 4697–
4715. https://doi.org/10.1080/01431161.2021.1899334
Fyllas, N. M., Christopoulou, A., Galanidis, A., Michelaki, C.
Z., Dimitrakopoulos, P. G., Fulé, P. Z., & Arianoutsou, M.
(2017). Tree growth-climate relationships in a forest-plot
network on Mediterranean mountains. Science of the
Total Environment, 598, 393–403. https://doi.org/10.
1016/j.scitotenv.2017.04.145
Gauquelin, T., Michon, G., Joffre, R., Duponnois, R., Génin,
D., Fady, B., Bou Dagher-Kharrat, M., Derridj, A.,
Slimani, S., Badri, W., Alifriqui, M., Auclair, L., Simenel,
R., Aderghal, M., Baudoin, E., Galiana, A., Prin, Y.,
Sanguin, H., Fernandez, C., & Baldy, V. (2018).
Mediterranean forests, land use and climate change: A
social-ecological perspective. Regional Environmental
Change, 18(3), 623–636. https://doi.org/10.1007/s10113-
016-0994-3
Giannetti, F., Chirici, G., Vangi, E., Corona, P., Maselli, F.,
Chiesi, M., D’Amico, G., & Puletti, N. (2022). Wall-to-
wall mapping of Forest Biomass and wood volume incre-
ment in Italy. Forests, 13(12), 1989. 13(12. https://doi.org/
10.3390/f13121989
Grossiord, C., Buckley, T. N., Cernusak, L. A., Novick, K. A.,
Poulter, B., Siegwolf, R. T. W., Sperry, J. S., & McDowell,
N. G. (2020). Plant responses to rising vapor pressure
deficit. New Phytologist, 226(6), 1550–1566. https://doi.
org/10.1111/nph.16485
Grünig, M., Seidl, R., & Senf, C. (2023). Increasing aridity
causes larger and more severe forest fires across Europe.
Global Change Biology, 29(6), 1648–1659. https://doi.org/
10.1111/gcb.16547
Gutsch, Lasch-Born, P., Kollas, C., Suckow, F., & Reyer, C.
P. O. (2018). Balancing trade-offs between ecosystem
services in Germany’s forests under climate change.
Environmental Research Letters, 13(4), 045012. https://
doi.org/10.1088/1748-9326/aab4e5
Hanson, P. J., Amthor, J. S., Wullschleger, S. D., Wilson, K.
B., Grant, R. F., Hartley, A., Hui, D., Hunt Jr, E. R.,
Johnson, D. W., Kimball, J. S., King, A. W., Luo, Y.,
McNulty, S. G., Sun, G., Thornton, P. E., Wang, S.,
Williams, M., Baldocchi, D. D., & Cushman, R. M.
(2004). Oak forest carbon and water simulations: Model
intercomparisons and evaluations against independent
data. Ecological Monographs, 74(3), 443–489. https://doi.
org/10.1890/03-4049
Herrera, S., Margarida Cardoso, R., Matos Soares, P.,
Espírito-Santo, F., Viterbo, P., & Gutiérrez, J. M. (2019).
Iberia01: A new gridded dataset of daily precipitation and
temperatures over Iberia. Earth System Science Data, 11
(4), 1947–1956. https://doi.org/10.5194/essd-11-1947-
2019
20 D. DALMONECH ET AL.
Hiederer, R. (2013). Mapping soil properties for Europe -
spatial representation of soil database attributes. JRC,
Technical Reports. https://publications.jrc.ec.europa.eu/
repository/handle/JRC83425
Hijmans, R. J., Bivand, R., Forner, K., Ooms, J., Pebesma, E.,
& Sumner, M. D. (2022). Package ‘terra’. Vienna, Austria.
Janssen, P. H. M., & Heuberger, P. S. C. (1995). Calibration
of process-oriented models. Ecological Modelling, 83(1–
2), 55–66. https://doi.org/10.1016/0304-3800(95)00084-9
Jung, M., Koirala, S., Weber, U., Ichii, K., Gans, F., Camps-
Valls, G., Papale, D., Schwalm, C., Tramontana, G., &
Reichstein, M. (2019). The FLUXCOM ensemble of glo-
bal land-atmosphere energy fluxes. Scientific Data, 6(1),
1–14. https://doi.org/10.1038/s41597-019-0076-8
Jung, M., Schwalm, C., Migliavacca, M., Walther, S., Camps-
Valls, G., Koirala, S., Anthoni, P., Besnard, S., Bodesheim,
P., Carvalhais, N., Chevallier, F., Gans, F., Goll, S.,
Haverd, D., Köhler, V., Ichii, P., Jain, K., Liu, A.,
Lombardozzi, J., & Reichstein, M. (2020). Scaling carbon
fluxes from eddy covariance sites to globe: Synthesis and
evaluation of the FLUXCOM approach. Biogeosciences,
17(5), 1343–1365. https://doi.org/10.5194/bg-17-1343-
2020
Kannenberg, S. A., Schwalm, C. R., Anderegg, W. R. L., &
Rejmanek, M. (2020). Ghosts of the past: How drought
legacy effects shape forest functioning and carbon
cycling. Ecology Letters, 23(5), 891–901. https://doi.org/
10.1111/ele.13485
Kattge, J., & Knorr, W. (2007). Temperature acclimation in
a biochemical model of photosynthesis: A reanalysis of
data from 36 species. Plant, Cell and Environment, 30(9),
1176–1190. https://doi.org/10.1111/j.1365-3040.2007.
01690.x
Keenan, T., Baker, I., Barr, A., Ciais, P., Davis, K., Dietze,
M., Dragoni, D., Gough, C. M., Grant, R., Hollinger, D.,
Hufkens, K., Poulter, B., McCaughey, H., Raczka, B., Ryu,
Y., Schaefer, K., Tian, H., Verbeeck, H., Zhao, M., &
Richardson, A. D. (2012). Terrestrial biosphere model
performance for inter-annual variability of land-atmo-
sphere CO 2 exchange. Global Change Biology, 18(6),
1971–1987. https://doi.org/10.1111/j.1365-2486.2012.
02678.x
Keenan, T., García, R., Friend, A. D., Zaehle, S., Gracia,
C., & Sabate, S. (2009). Improved understanding of
drought controls on seasonal variation in
Mediterranean forest canopy CO2 and water fluxes
through combined in situ measurements and ecosystem
modelling. Biogeosciences, 6(8), 1423–1444. https://doi.
org/10.5194/bg-6-1423-2009
Koch, J., Demirel, M. C., & Stisen, S. (2018). The SPAtial
EFficiency metric (SPAEF): Multiple-component evalua-
tion of spatial patterns for optimization of hydrological
models. Geoscientific Model Development, 11(5), 1873–
1886. https://doi.org/10.5194/gmd-11-1873-2018
Kollet, S. J., & Maxwell, R. M. (2008). Capturing the influ-
ence of groundwater dynamics on land surface processes
using an integrated, distributed watershed model. Water
Resources Research, 44(2), 1–18. https://doi.org/10.1029/
2007WR006004
Kumar, S. V., Peters-Lidard, C. D., Mocko, D., Reichle, R.,
Liu, Y., Arsenault, K. R., Xia, Y., Ek, M., Riggs, G., Livneh,
B., & Cosh, M. (2014). Assimilation of remotely sensed
soil moisture and snow depth retrievals for drought esti-
mation. Journal of Hydrometeorology, 15(6), 2446–2469.
https://doi.org/10.1175/JHM-D-13-0132.1
Liaw, A., & Wiener, M. (2002). Classification and regression
by randomForest. R News, 2(3), 18–22.
Lindeskog, M., Smith, B., Lagergren, F., Sycheva, E., Ficko,
A., Pretzsch, H., & Rammig, A. (2021). Accounting for
forest management in the estimation of forest carbon
balance using the dynamic vegetation model LPJ-
GUESS (v4.0, r9710): Implementation and evaluation of
simulations for Europe. Geoscientific Model Development,
14(10), 6071–6112. https://doi.org/10.5194/gmd-14-
6071-2021
Li, X., & Xiao, J. (2019a). A global, 0.05-degree product of
solar-induced chlorophyll fluorescence derived from
OCO-2, MODIS, and reanalysis data. Remote Sensing,
11(5), 2563. https://doi.org/10.3390/rs11050517
Li, X., & Xiao, J. (2019b). Mapping photosynthesis solely
from solar-induced chlorophyll fluorescence: A global,
fine-resolution dataset of gross primary production
derived from OCO-2. Remote Sensing, 11(21), 2563.
Remote Sensing, 11(21. https://doi.org/10.3390/
rs11212563
Li, X., Xiao, J., He, B., Altaf Arain, M., Beringer, J., Desai, A.
R., Emmel, C., Hollinger, D. Y., Krasnova, A.,
Mammarella, I., Noe, S. M., Ortiz, P. S., Rey-Sanchez, A.
C., Rocha, A. V., & Varlagin, A. (2018). Solar-induced
chlorophyll fluorescence is strongly correlated with ter-
restrial photosynthesis for a wide variety of biomes: First
global analysis based on OCO-2 and flux tower observa-
tions. Global Change Biology, 24(9), 3990–4008. https://
doi.org/10.1111/gcb.14297
Lorenz, R., Stalhandske, Z., & Fischer, E. M. (2019).
Detection of a climate Change Signal in extreme heat,
heat stress, and cold in Europe from observations.
Geophysical Research Letters, 46(14), 8363–8374. https://
doi.org/10.1029/2019GL082062
Mahnken, M., Cailleret, M., Collalti, A., Trotta, C., Biondo,
C., D’Andrea, E., Dalmonech, D., Marano, G., Mäkelä, A.,
Minunno, F., Peltoniemi, M., Trotsiuk, V., Nadal-Sala,
D., Sabaté, S., Vallet, P., Aussenac, R., Cameron, D. R.,
Bohn, F. J., Grote, R., and Reyer, C. P. O. (2022).
Accuracy, realism and general applicability of European
forest models. Global Change Biology, 28(23), 6921–6943.
https://doi.org/10.1111/gcb.16384
Marchand, W., Buechling, A., Rydval, M., Čada, V.,
Stegehuis, A. I., Fruleux, A., Poláček, M., Hofmeister, J.,
Pavlin, J., Ralhan, D., Dušátko, M., Janda, P., Mikoláš, M.,
Vostarek, O., Bače, R., Frankovič, M., Kozák, D., Roibu,
C. C., Chaskovskyy, O. … Svoboda, M. (2023).
Accelerated growth rates of Norway spruce and
European beech saplings from Europe’s temperate pri-
mary forests are related to warmer conditions.
Agricultural and Forest Meteorology 329, 109280. March
2022. https://doi.org/10.1016/j.agrformet.2022.109280
Marconi, S., Chiti, T., Nolè, A., Valentini, R., & Collalti, A.
(2017). The role of respiration in estimation of net carbon
cycle: Coupling soil carbon dynamics and canopy turn-
over in a novel version of 3D-CMCC forest ecosystem
model. Forests, 8(6), 220. https://doi.org/10.3390/
f8060220
Maréchaux, I., Langerwisch, F., Huth, A., Bugmann, H.,
Morin, X., Reyer, C. P. O., Seidl, R., Collalti, A., Dantas
de Paula, M., Fischer, R., Gutsch, M., Lexer, M. J.,
Lischke, H., Rammig, A., Rödig, E., Sakschewski, B.,
Taubert, F., Thonicke, K., Vacchiano, G., & Bohn, F. J.
(2021). Tackling unresolved questions in forest ecology:
The past and future role of simulation models. Ecology
and Evolution, 11(9), 3746–3770. https://doi.org/10.1002/
ece3.7391
Maselli, F., Barbati, A., Chiesi, M., Chirici, G., & Corona, P.
(2006). Use of remotely sensed and ancillary data for
EUROPEAN JOURNAL OF REMOTE SENSING 21
estimating forest gross primary productivity in Italy.
Remote Sensing of Environment, 100(4), 563–575.
https://doi.org/10.1016/j.rse.2005.11.010
Maselli, F., Pasqui, M., Chirici, G., Chiesi, M., Fibbi, L.,
Salvati, R., & Corona, P. (2012). Modeling primary pro-
duction using a 1 km daily meteorological data set.
Climate Research, 54(3), 271–285. https://doi.org/10.
3354/cr01121
Maselli, P., Puletti, D., Chirici G, N., & Corona, P. (2009).
Combining remote sensing and ancillary data to monitor
the gross productivity of water-limited forest ecosystems.
Remote Sensing of Environment, 113(3), 657–667. https://
doi.org/10.1016/j.rse.2008.11.008
Massari, C., Brocca, L., Pellarin, T., Abramowitz, G.,
Filippucci, P., Ciabatta, L., Maggioni, V., Kerr, Y., &
Fernandez Prieto, D. (2020). A daily 25 km short-latency
rainfall product for data-scarce regions based on the
integration of the global precipitation measurement mis-
sion rainfall and multiple-satellite soil moisture products.
Hydrology and Earth System Sciences, 24(5), 2687–2710.
https://doi.org/10.5194/hess-24-2687-2020
Massoud, E. C., Xu, C., Fisher, R. A., Knox, R. G., Walker, A.
P., Serbin, S. P., Christoffersen, B. O., Holm, J. A.,
Kueppers, L. M., Ricciuto, D. M., Wei, L., Johnson, D.
J., Chambers, J.Q., Koven, C. D., McDowell, N. G., &
Vrugt, J. A. (2019). Identification of key parameters con-
trolling demographically structured vegetation dynamics
in a land surface model: CLM4.5(FATES). Geoscientific
Model Development, 12, 4133–4164. https://doi.org/10.
5194/gmd-12-4133-2019
Médail, F., Monnet, A. C., Pavon, D., Nikolic, T.,
Dimopoulos, P., Bacchetta, G., Arroyo, J., Barina, Z.,
Albassatneh, M. C., Domina, G., Fady, B., Matevski, V.,
Mifsud, S., & Leriche, A. (2019). What is a tree in the
Mediterranean basin hotspot? A critical analysis. Forest
Ecosystems, 6(1). https://doi.org/10.1186/s40663-019-
0170-6
Merganičová, K., Merganič, J., Lehtonen, A., Vacchiano, G.,
Sever, M. Z. O., Augustynczik, A. L. D., Grote, R.,
Kyselová, I., Mäkelä, A., Yousefpour, R., Krejza, J.,
Collalti, A., Reyer, C. P. O., & Polle, A. (2019). Forest
carbon allocation modelling under climate change. Tree
Physiology, 39(12), 1937–1960. https://doi.org/10.1093/
treephys/tpz105
Minunno, F., Peltoniemi, M., Härkönen, S., Kalliokoski, T.,
Makinen, H., & Mäkelä, A. (2019). Bayesian calibration of
a carbon balance model PREBAS using data from perma-
nent growth experiments and national forest inventory.
Forest Ecology and Management, 440(March), 208–257.
https://doi.org/10.1016/j.foreco.2019.02.041
Moreno, A., & Hasenauer, H. (2016). Spatial downscaling of
European climate data. International Journal of
Climatology, 36(3), 1444–1458. https://doi.org/10.1002/
joc.4436
Niu, G. Y., Yang, Z. L., Mitchell, K. E., Chen, F., Ek, M. B.,
Barlage, M., Kumar, A., Manning, K., Niyogi, D., Rosero,
E., Tewari, M., & Xia, Y. (2011). The community Noah
land surface model with multiparameterization options
(Noah-MP): 1. Model description and evaluation with
local-scale measurements. Journal of Geophysical
Research Atmospheres, 116(12), 1–19. https://doi.org/10.
1029/2010JD015139
Noce, S., Collalti, A., Valentini, R., & Santini, M. (2016). Hot
spot maps of forest presence in the Mediterranean basin.
IForest - Biogeosciences & Forestry, 9(5), 766–774. https://
doi.org/10.3832/ifor1802-009
Nord-Larsen, T., & Schumacher, J. (2012). Estimation of
forest resources from a country wide laser scanning sur-
vey and national forest inventory data. Remote Sensing of
Environment, 119, 148–157. https://doi.org/10.1016/j.rse.
2011.12.022
Norton, A. J., Rayner, P. J., Koffi, E. N., Scholze, M., Silver, J.
D., & Wang, Y. P. (2019). Estimating global gross primary
productivity using chlorophyll fluorescence and a data
assimilation system with the BETHY-SCOPE model.
Biogeosciences, 16(15), 3069–3093. https://doi.org/10.
5194/bg-16-3069-2019
O’Sullivan, M., Smith, W. K., Sitch, S., Friedlingstein, P.,
Arora, V. K., Haverd, V., Jain, A. K., Kato, E., Kautz, M.,
Lombardozzi, D., Nabel, J. E. M. S., Tian, H., Vuichard,
N., Wiltshire, A., Zhu, D., & Buermann, W. (2020).
Climate-driven variability and trends in plant productiv-
ity over recent decades based on three global products.
Global Biogeochemical Cycles, 34(12). https://doi.org/10.
1029/2020GB006613
Pastorello, G., Trotta, C., Canfora, E., Chu, H., Christianson,
D., Cheah, Y. W., Poindexter, C., Chen, J., Elbashandy,
A., Humphrey, M., Isaac, P., Polidori, D., Ribeca, A., van
Ingen, C., Zhang, L., Amiro, B., Ammann, C., Arain, M.
A., Ardö, J. & Papale, D. (2020). The FLUXNET2015
dataset and the ONEFlux processing pipeline for eddy
covariance data. Scientific Data, 7(1), 225. https://doi.org/
10.1038/s41597-020-0534-3
Patacca, M., Lindner, M., Lucas-Borja, M. E., Cordonnier,
T., Fidej, G., Gardiner, B., Hauf, Y., Jasinevičius, G.,
Labonne, S., Linkevičius, E., Mahnken, M., Milanovic,
S., Nabuurs, G. J., Nagel, T. A., Nikinmaa, L., Panyatov,
M., Bercak, R., Seidl, R., Ostrogović Sever, M. Z. &
Schelhaas, M. J. (2023). Significant increase in natural
disturbance impacts on European forests since 1950.
Global Change Biology, 29(5), 1359–1376. https://doi.
org/10.1111/gcb.16531
Peano, D., Hemming, D., Materia, S., Delire, C., Fan, Y.,
Joetzjer, E., Lee, H., Nabel, J. E. M. S., Park, T., Peylin, P.,
Wårlind, D., Wiltshire, A., & Zaehle, S. (2021). Plant
phenology evaluation of CRESCENDO land surface
models – part 1: Start and end of the growing season.
Biogeosciences, 18(7), 2405–2428. https://doi.org/10.5194/
bg-18-2405-2021
Peano, D., Materia, S., Collalti, A., Alessandri, A., Anav, A.,
Bombelli, A., & Gualdi, S. (2019). Global variability of
simulated and observed vegetation growing season.
Journal of Geophysical Research: Biogeosciences, 124(11),
3569–3587. https://doi.org/10.1029/2018JG004881
Peaucelle, M., Janssens, I. A., Stocker, B. D., Descals
Ferrando, A., Fu, Y. H., Molowny-Horas, R., Ciais, P., &
Peñuelas, J. (2019). Spatial variance of spring phenology
in temperate deciduous forests is constrained by back-
ground climatic conditions. Nature Communications, 10
(1), 1–10. https://doi.org/10.1038/s41467-019-13365-1
Peñuelas, J., Sardans, J., Filella, I., Estiarte, M., Llusià, J.,
Ogaya, R., Carnicer, J., Bartrons, M., Rivas-Ubach, A.,
Grau, O., Peguero, G., Margalef, O., Pla-Rabés, S.,
Stefanescu, C., Asensio, D., Preece, C., Liu, L., Verger,
A., Barbeta, A. & Terradas, J. (2017). Impacts of global
change on Mediterranean forests and their services.
Forests, 8(12), 1–37. https://doi.org/10.3390/f8120463
Pretzsch, H., Grote, R., Reineking, B., Rötzer, T., & Seifert, S.
(2008). Models for forest ecosystem management: A
European perspective. Annals of Botany, 101(8), 1065–
1087. https://doi.org/10.1093/aob/mcm246
Qiu, R., Han, G., Ma, X., Xu, H., Shi, T., & Zhang, M. (2020).
A comparison of OCO-2 SIF, MODIS GPP, and GOSIF
22 D. DALMONECH ET AL.
data from gross primary production (GPP) estimation
and seasonal cycles in North America. Remote Sensing,
12(2), 258. https://doi.org/10.3390/rs12020258
R Core Team. (2021). R: A language and environment for
statistical computing. R Foundation for Statistical
Computing. https://www.R-project.org/
Reyer. (2015). Forest productivity under environmental
change—A review of stand-scale modeling studies.
Current Forestry Reports, 1(2), 53–68. https://doi.org/10.
1007/s40725-015-0009-5
Reyer, C. P. O., Gonzalez, S., Dolos, R., Hartig, K., Hauf, F.,
Noack, Y., Lasch-Born, M., Rötzer, P., Pretzsch, T.,
Meesenburg, H., Fleck, H., Wagner, S., Bolte, M.,
Sanders, A., Kolari, T. G. M., Mäkelä, P., Vesala, A.,
Mammarella, T., Pumpanen, I. G., & Frieler, K. (2020).
The PROFOUND database for evaluating vegetation
models and simulating climate impacts on European
forests. Earth System Science Data, 12(2), 1295–1320.
https://doi.org/10.5194/essd-12-1295-2020
Reyer, C., Lasch-Born, P., Suckow, F., Gutsch, M.,
Murawski, A., & Pilz, T. (2014). Projections of regional
changes in forest net primary productivity for different
tree species in Europe driven by climate change and
carbon dioxide. Annals of Forest Science, 71(2), 211–225.
https://doi.org/10.1007/s13595-013-0306-8
Richards, F. J. (1959). A flexible growth function for empiri-
cal use. Journal of Experimental Botany, 10(2), 290–301.
https://doi.org/10.1093/jxb/10.2.290
Ripullone, F., Camarero, J. J., Colangelo, M., Voltas, J., &
Cernusak, L. (2020). Variation in the access to deep soil
water pools explains tree-to-tree differences in drought-
triggered dieback of Mediterranean oaks. Tree Physiology,
40(5), 591–604. https://doi.org/10.1093/treephys/tpaa026
Rita, A., Camarero, J. J., Nolè, A., Borghetti, M., Brunetti,
M., Pergola, N., Serio, C., Vicente-Serrano, S. M.,
Tramutoli, V., & Ripullone, F. (2020). The impact of
drought spells on forests depends on site conditions:
The case of 2017 summer heat wave in southern
Europe. Global Change Biology, 26(2), 851–863. https://
doi.org/10.1111/gcb.14825
Rojas, R., Feyen, L., Dosio, A., & Bavera, D. (2011).
Improving pan-European hydrological simulation of
extreme events through statistical bias correction of
RCM-driven climate simulations. Hydrology and Earth
System Sciences, 15(8), 2599–2620. https://doi.org/10.
5194/hess-15-2599-2011
Sanchez-Ruiz, S., Chiesi, M., Fibbi, L., Carrara, A., Maselli,
F., & Gilabert, M. A. (2018). Optimized application of
biome-BGC for modeling the daily GPP of natural vege-
tation over Peninsular Spain. Journal of Geophysical
Research: Biogeosciences, 123(2), 531–546. https://doi.
org/10.1002/2017JG004360
Schenk, H. J., & Jackson, R. B. (2005). Mapping the global
distribution of deep roots in relation to climate and soil
characteristics. Geoderma, 126(1–2), 129–140. 1-2 SPEC.
ISS. https://doi.org/10.1016/j.geoderma.2004.11.018
Schuldt, B., Buras, A., Arend, M., Vitasse, Y., Beierkuhnlein, C.,
Damm, A., Gharun, M., Grams, T. E. E., Hauck, M., Hajek,
P., Hartmann, H., Hiltbrunner, E., Hoch, G., Holloway-
Phillips, M., Körner, C., Larysch, E., Lübbe, T., Nelson, D.
B., Rammig, A. & Kahmen, A. (2020). A first assessment of
the impact of the extreme 2018 summer drought on Central
European forests. Basic and Applied Ecology, 45, 86–103.
https://doi.org/10.1016/j.baae.2020.04.003
Senf, C., Buras, A., Zang, C. S., Rammig, A., & Seidl, R.
(2020). Excess forest mortality is consistently linked to
drought across Europe. Nature Communications, 11(1),
1–8. https://doi.org/10.1038/s41467-020-19924-1
Sitch, S. (2015). Recent trends and drivers of regional
sources and sinks of carbon dioxide. Biogeosciences, 12
(3), 653–679. https://doi.org/10.5194/bg-12-653-201510.
5194/bg-12-653-2015-supplement
Stocker, B. D., Tumber-Dávila, S. J., Konings, A. G.,
Anderson, M. C., Hain, C., & Jackson, R. B. (2023).
Global patterns of water storage in the rooting zones of
vegetation. Nature Geoscience, 16(3), 250–256. https://
doi.org/10.1038/s41561-023-01125-2
Suárez-Muñoz, M., Bonet-García, F. J., Navarro-Cerrillo, R.,
Herrero, J., & Mina, M. (2023). Forest management sce-
narios drive future dynamics of Mediterranean planted
pine forests under climate change. Landscape Ecology, 38
(8), 2069–2084. https://doi.org/10.1007/s10980-023-
01678-y
Sun, Y., Frankenberg, C., Wood, J. D., Schimel, D. S., Jung,
M., Guanter, L., Drewry, D. T., Verma, M., Porcar-
Castell, A., Griffis, T. J., Gu, L., Magney, T. S., Köhler,
P., Evans, B., & Yuen, K. (2017). OCO-2 advances photo-
synthesis observation from space via solar-induced chlor-
ophyll fluorescence. Science, 358(6360). https://doi.org/
10.1126/science.aam5747
Sun, Z., Wang, W., Zhag, X., Tai, H., Guo, H., Yin, S., &
Zhag, T. (2019). A recent large review on RS-based pro-
ducts intercomparison confirmed the large variability and
different sensitivity to climatic factors between different
approaches and criteria when compared at site-level.
Science of the Total Environment, 668, 696–713. https://
doi.org/10.1016/j.scitotenv.2019.03.025
Tarquini, T., Isola, I., Favalli, M., Mazzarini, F., Bisson, M.,
Pareschi, M. T., & Boschi, E. (2009). TINITALY/01: A
new triangular irregular network of Italy. Annals of
Geophysics, 50(3), 50(3. https://doi.org/10.4401/ag-4424
Tedeschi, V., Rey, A., Manca, G., Valentini, R., Jarvis, P. J., &
Borghetti, M. (2006). Soil respiration in a Mediterranean
oak forest at different developmental stages after coppi-
cing. Global Change Biology, 12(1), 110–121. https://doi.
org/10.1111/j.1365-2486.2005.01081.x
Testolin, R., Dalmonech, D., Marano, G., Bagnara, M.,
D’Andrea, E., Matteucci, G., Noce, S., & Collalti, A.
(2023). Simulating diverse forest management options
in a changing climate on a Pinus nigra subsp. laricio
plantation in Southern Italy. Science of the Total
Environment, 857(October 2022), 159361. https://doi.
org/10.1016/j.scitotenv.2022.159361
Vacchiano, G., Magnani, F., & Collalti, A. (2012). Modeling
Italian forests: State of the art and future challenges.
IForest - Biogeosciences & Forestry, 5(1), 113–120.
https://doi.org/10.3832/ifor0614-005
Vangi, E., D’Amico, G., Francini, S., Borghi, C., Giannetti,
F., Corona, P., Marchetti, M., Travaglini, D., Pellis, G.,
Vitullo, M., & Chirici, G. (2023). LARGE-SCALE high-
resolution yearly modeling of forest growing stock
volume and above-ground carbon pool. Environmental
Modelling and Software, 159, 105580. https://doi.org/10.
1016/j.envsoft.2022.105580
Vangi, E., D’amico, G., Francini, S., Giannetti, F., Lasserre,
B., Marchetti, M., McRoberts, R. E., & Chirici, G. (2021).
The effect of forest mask quality in the wall–to–wall
estimation of growing stock volume. Remote Sensing, 13
(5), 1–20. https://doi.org/10.3390/rs13051038
Verkerk, P. J., Delacote, P., Hurmekoski, E., Kunttu, J.,
Matthews, R., Mäkipää, R., Mosley, F., Perugini, L., Reyer,
C. P. O., Roe, S., & Trømborg, E. (2022). Forest-based climate
change mitigation and adaptation in Europe.
EUROPEAN JOURNAL OF REMOTE SENSING 23
Veroustraete, F., Sabbe, H., & Eerens, H. (2002). Estimation
of carbon mass fluxes over Europe using the C-Fix model
and euroflux data. Remote Sensing of Environment, 83(3),
376–399. https://doi.org/10.1016/S0034-4257(02)00043-3
Vicente-Serrano, S. M., Beguería, S., & López-Moreno, J. I.
(2010). A multiscalar drought index sensitive to global
warming: The standardized precipitation evapotranspira-
tion index. Journal of Climate, 23(7), 1696–1718. https://
doi.org/10.1175/2009JCLI2909.1
Vogel, J., Paton, E., Aich, V., & Bronstert, A. (2021).
Increasing compound warm spells and droughts in the
Mediterranean Basin. Weather and Climate Extremes, 32,
100312. https://doi.org/10.1016/j.wace.2021.100312
Walther, S., Duveiller, G., Jung, M., Guanter, L., Cescatti, A.,
& Camps-Valls, G. (2019). Satellite observations of the
contrasting response of trees and grasses to variations in
water availability. Geophysical Research Letters, 46(3),
1429–1440. https://doi.org/10.1029/2018GL080535
Wang, S., Zhang, Y., Ju, W., Qiu, B., & Zhang, Z. (2021).
Tracking the seasonal and inter-annual variations of glo-
bal gross primary production during last four decades
using satellite near-infrared reflectance data. Science of
the Total Environment, 755, 755. https://doi.org/10.1016/
j.scitotenv.2020.142569
Waser, L. T., Ginzler, C., & Rehush, N. (2017). Wall-to-wall
tree type mapping from countrywide airborne remote
sensing surveys. Remote Sensing, 9(8), 766. https://doi.
org/10.3390/rs9080766
Wild, B., Teubner, I., Moesinger, L., Zotta, R., Forkel, M.,
van der Schalie, R., Sitch, S., & Dorigo, W. (2022).
VODCA2GPP – a new, global, long-term (1988–2020)
gross primary production dataset from microwave
remote sensing. Earth System Science Data, 14(3), 1063–
1085. https://doi.org/10.5194/essd-14-1063-2022
Wohlfahrt, G., Gerdel, K., Migliavacca, M., Rotenberg, E.,
Tatarinov, F., Müller, J., Hammerle, A., Julitta, T.,
Spielmann, F. M., & Yakir, D. (2018). Sun-induced fluor-
escence and gross primary productivity during a heat
wave. Scientific Reports, 8(1), 1–9. https://doi.org/10.
1038/s41598-018-32602-z
Wu, Z., Ahlström, A., Smith, B., Ardö, J., Eklundh, L.,
Fensholt, R., & Lehsten, V. (2017). Climate data
induced uncertainty in model-based estimations of
terrestrial primary productivity. Environmental
Research Letters, 12(6), 064013. https://doi.org/10.
1088/1748-9326/aa6fd8
Xiao, J., Li, X., He, B., Arain, M. A., Beringer, J., Desai, A. R.,
Emmel, C., Hollinger, D. Y., Krasnova, A., Mammarella,
I., Noe, S. M., Serrano Ortiz, P., Rey-Sanchez, C., Rocha,
A. V., & Varlagin, A. (2019). Solar-induced chlorophyll
fluorescence exhibits a universal relationship with gross
primary productivity across a wide variety of biomes.
Global Change Biology, 25(4), e4–e6. https://doi.org/10.
1111/gcb.14565
Yang, R., Wang, J., Zeng, N., Sitch, S., Tang, W., McGrath,
M. J., Cai, Q., Liu, D., Lombardozzi, D., Tian, H., Jain, A.
K., & Han, P.(2022). Divergent historical GPP trends
among state-of-the-art multi-model simulations and
satellite-based products. Earth Syst. Dynam, 13(2), 833–
849. https://doi.org/10.5194/esd-13-833-202210.5194/
esd-13-833-2022-supplement
Zampieri, M., Grizzetti, B., Toreti, A., De Palma, P., &
Collalti, A. (2021). Rise and fall of vegetation annual
primary production resilience to climate variability
projected by a large ensemble of earth system mod-
els’ simulations. Environmental Research Letters, 16
(10), 105001. https://doi.org/10.1088/1748-9326/
ac2407
Zhang, F., Lu, X., Huang, Q., & Jiang, F. (2022). Impact of
different ERA reanalysis data on GPP simulation.
Ecological Informatics, 68(August 2021), 101520. https://
doi.org/10.1016/j.ecoinf.2021.101520
Zhang, Y., & Ye, A. (2021). Would the obtainable gross
primary productivity (GPP) products stand up? A critical
assessment of 45 global GPP products. Science of the Total
Environment, 783, 146965. https://doi.org/10.1016/j.scito
tenv.2021.146965
Zhang, Y., & Ye, A. (2022). Uncertainty analysis of multiple
terrestrial gross primary productivity products. Global
Ecology and Biogeography, 31(11), 2204–2218. https://
doi.org/10.1111/geb.13578
Zhao, W., & Zhu, Z. (2022). Exploring the Best-Matching
Plant Traits and Environmental Factors for Vegetation
Indices in estimates of global gross primary productivity.
Remote Sensing, 14(24), 6316. https://doi.org/10.3390/
rs14246316
Zheng, Y., Shen, R., Wang, Y., Li, X., Liu, S., Liang, S., Chen,
J. M., & Ju, W., Zhang, L., & Yuan, W. (2020). Improved
estimate of global gross primary production for reprodu-
cing its long-term variation, 1982–2017. Earth System
Science Data, 12(4), 2725–2746. https://doi.org/10.5194/
essd-12-2725-2020
24 D. DALMONECH ET AL.
