Twan van Noije’s research while affiliated with Koninklijk Nederlands Meteorologisch Instituut and other places

Given the importance of aerosols and clouds and their interactions in the climate system, it is imperative that the global Earth system models accurately represent processes associated with them. This is an important prerequisite if we are to narrow the uncertainties in future climate projections. In practice, this means that continuous model evaluations and improvements grounded in observations are necessary. Numerous studies in the past few decades have shown both the usability and the limitations of utilizing satellite-based observations in understanding and evaluating aerosol–cloud interactions, particularly under varying meteorological and satellite sensor sensitivity paradigms. Furthermore, the vast range of spatio-temporal scales at which aerosol and cloud processes occur adds another dimension to the challenges faced when evaluating climate models. In this context, the aim of this study is two-fold. (1) We evaluate the most recent, significant changes in the representation of aerosol and cloud processes implemented in the EC-Earth3-AerChem model in the framework of the EU project FORCeS compared with its previous CMIP6 version (Coupled Model Intercomparison Project Phase 6; https://pcmdi.llnl.gov/CMIP6/, last access: 13 February 2019). We focus particularly on evaluating cloud physical properties and radiative effects, wherever possible, using a satellite simulator. We report on the overall improvements in the EC-Earth3-AerChem model. In particular, the strong warm bias chronically seen over the Southern Ocean is reduced significantly. (2) A statistical, maximum covariance analysis is carried out between aerosol optical depth (AOD) and cloud droplet (CD) effective radius based on the recent EC-Earth3-AerChem/FORCeS simulation to understand to what extent the Twomey effect can manifest itself in the larger spatio-temporal scales. We focus on the three oceanic low-level cloud regimes that are important due to their strong net cooling effect and where pollution outflow from the nearby continent is simultaneously pervasive. We report that the statistical covariability between AOD and CD effective radius is indeed dominantly visible even at the climate scale when the aerosol amount and composition are favourably preconditioned to allow for aerosol–cloud interactions. Despite this strong covariability, our analysis shows a strong cooling/warming in shortwave cloud radiative effects at the top of the atmosphere in our study regions associated with an increase/decrease in CD effective radius. This cooling/warming can be attributed to the increase/decrease in low cloud fraction, in line with previous observational studies.

Anthropogenic aerosols play a major role in the Earth–atmosphere system by influencing the Earth's radiative budget and precipitation and consequently the climate. The perturbation induced by changes in anthropogenic aerosols on the Earth's energy balance is quantified in terms of the effective radiative forcing (ERF). In this work, the present-day shortwave (SW), longwave (LW), and total (i.e., SW plus LW) ERF of anthropogenic aerosols is quantified using two different sets of experiments with prescribed sea surface temperatures (SSTs) from Earth system models (ESMs) participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6): (a) time-slice pre-industrial perturbation simulations with fixed SSTs (piClim) and (b) transient historical simulations with time-evolving SSTs (histSST) over the historical period (1850–2014). ERF is decomposed into three components for both piClim and histSST experiments: (a) ERFARI, representing aerosol–radiation interactions; (b) ERFACI, accounting for aerosol–cloud interactions (including the semi-direct effect); and (c) ERFALB, which is due to temperature, humidity, and surface albedo changes caused by anthropogenic aerosols. We present spatial patterns at the top-of-atmosphere (TOA) and global weighted field means along with inter-model variability (1 standard deviation) for all SW, LW, and total ERF components (ERFARI, ERFACI, and ERFALB) and for every experiment used in this study. Moreover, the inter-model agreement and the robustness of our results are assessed using a comprehensive method as utilized in the IPCC Sixth Assessment Report. Based on piClim experiments, the total present-day (2014) ERF from anthropogenic aerosol and precursor emissions is estimated to be -1.11 ± 0.26 Wm-2, mostly due to the large contribution of ERFACI to the global mean and to the inter-model variability. Based on the histSST experiments for the present-day period (1995–2014), similar results are derived, with a global mean total aerosol ERF of -1.28 ± 0.37 Wm-2 and dominating contributions from ERFACI. The spatial patterns for total ERF and its components are similar in both the piClim and histSST experiments. Furthermore, implementing a novel approach to determine geographically the driving factor of ERF, we show that ERFACI dominates over the largest part of the Earth and that ERFALB dominates mainly over the poles, while ERFARI dominates over certain reflective surfaces. Analysis of the inter-model variability in total aerosol ERF shows that SW ERFACI is the main source of uncertainty predominantly over land regions with significant changes in aerosol optical depth (AOD), with eastern Asia contributing mostly to the inter-model spread of both ERFARI and ERFACI. The global spatial patterns of total ERF and its components from individual aerosol species, such as sulfates, organic carbon (OC), and black carbon (BC), are also calculated based on piClim experiments. The total ERF caused by sulfates (piClim-SO2) is estimated at -1.11 ± 0.31 Wm-2, and the OC ERF (piClim-OC) is -0.35 ± 0.21 Wm-2, while the ERF due to BC (piClim-BC) is 0.19 ± 0.18 Wm-2. For sulfates and OC perturbation experiments, ERFACI dominates over the globe, whereas for BC perturbation experiments ERFARI dominates over land in the Northern Hemisphere and especially in the Arctic. Generally, sulfates dominate ERF spatial patterns, exerting a strongly negative ERF especially over industrialized regions of the Northern Hemisphere (NH), such as North America, Europe, and eastern and southern Asia. Our analysis of the temporal evolution of ERF over the historical period (1850–2014) reveals that ERFACI clearly dominates over ERFARI and ERFALB for driving the total ERF temporal evolution. Moreover, since the mid-1980s, total ERF has become less negative over eastern North America and western and central Europe, while over eastern and southern Asia there is a steady increase in ERF magnitude towards more negative values until 2014.

Given the importance of aerosols, clouds and their interactions in the climate system, it is imperative that the global Earth system models accurately represent processes associated with them. This is an important prerequisite if we were to narrow the uncertainties in future climate projections. In practice, this means that the continuous model evaluations and improvements grounded in observations are necessary. Numerous studies in the last few decades have shown both the usability and the limitations of utilizing satellite-based observations in understanding and evaluating aerosol-cloud interactions, particularly under varying meteorological and satellite sensor sensitivity paradigms. Furthermore, the vast range of spatio-temporal scales at which aerosol and cloud processes occur adds another dimension to the challenges while evaluating climate models. In this context, the aim of this study is two-fold. 1) We evaluate the most recent, significant changes in the representation of aerosol and cloud processes implemented in the EC-Earth3-AerChem model in the framework of the EU project FORCeS compared to its previous CMIP6 version. We focus particularly on evaluating cloud physical properties and radiative effects, wherever possible, using a satellite simulator. We report overall improvements in EC-Earth3-AerChem model. In particular, the strong warm bias chronically seen over the Southern Ocean is reduced significantly. 2) A statistical, maximum covariance analysis is carried out between aerosol optical depth (AOD) and cloud droplet (CD) effective radius based on the recent EC-Earth3-AerChem/FORCeS simulation to understand to what extent the Twomey effect can manifest itself in the larger spatio-temporal scales. We focus on the three oceanic low-level cloud regimes that are important due to their strong net cooling effect and where pollution outflow from the nearby continent is simultaneously pervasive. We report that the statistical covariability between AOD and CD effective radius is indeed dominantly visible even at the climate scale when the aerosol amount and composition are favourably preconditioned for allowing aerosol-cloud interactions. Despite this strong covariability, our analysis shows a strong cooling/warming in shortwave cloud radiative effects at the top of the atmosphere in our study regions associated with an increase/decrease in CD effective radius. And this cooling/warming can be attributed to the increase/decrease in low cloud fraction, in line with the previous observational studies.

Long-term exposure to ambient ozone (O 3 ) is associated with excess respiratory mortality. Pollution emissions, demographic, and climate changes are expected to drive future ozone-related mortality. Here, we assess global mortality attributable to ozone according to an IPCC (Intergovernmental Panel on Climate Change) SSP (Shared Socioeconomic Pathway) scenario applied in CMIP6 (Coupled Model Intercomparison Project Phase 6) models, projecting a temperature increase of about 3.6°C by the end of the century. We estimated ozone-related mortality on a global scale up to 2090 following the Global Burden of Disease (GBD) 2019 approach, using bias-corrected simulations from three CMIP6 Earth System Models (ESMs) under the SSP3-7.0 emissions scenario. Based on the three ESMs simulations, global ozone-related mortality by 2090 will amount to 2.79M [95% CI 0.97M–5.23M] to 3.12M [95% CI 1.11M–5.75M] per year, approximately ninefold that of the 327K [95% CI 103K–652K] deaths per year in 2000. Climate change alone may lead to an increase of ozone-related mortality in 2090 between 42K [95% CI -37K–122K] and 217K [95% CI 68K–367K] per year. Population growth and ageing are associated with an increase in global ozone-related mortality by a factor of 5.34, while the increase by ozone trends alone ranges between factors of 1.48 and 1.7. Ambient ozone pollution under the high-emissions SSP3-7.0 scenario is projected to become a significant human health risk factor. Yet, optimizing living conditions and healthcare standards worldwide to the optimal ones today (application of minimum baseline mortality rates) will help mitigate the adverse consequences associated with population growth and ageing, and ozone increases caused by pollution emissions and climate change.

Paleo-proxy data indicate that a “Green Sahara” thrived in northern Africa during the early- to mid-Holocene (MH; 11 000 to 5000 years before present), characterized by more vegetation cover and reduced dust emissions. Utilizing a state-of-the-art atmospheric chemical transport model, TM5-MP, we assessed the changes in biogenic volatile organic compound (BVOC) emissions, dust emissions and secondary organic aerosol (SOA) concentrations in northern Africa during this period relative to the pre-industrial (PI) period. Our simulations show that dust emissions reduced from 280.6 Tg a-1 in the PI to 26.8 Tg a-1 in the MH, agreeing with indications from eight marine sediment records in the Atlantic Ocean. The northward expansion in northern Africa resulted in an increase in annual emissions of isoprene and monoterpenes during the MH, around 4.3 and 3.5 times higher than that in the PI period, respectively, causing a 1.9-times increase in the SOA surface concentration. Concurrently, enhanced BVOC emissions consumed more hydroxyl radical (OH), resulting in less sulfate formation. This effect counteracted the enhanced SOA surface concentration, altogether leading to a 17 % increase in the cloud condensation nuclei at 0.2 % super saturation over northern Africa. Our simulations provide consistent emission datasets of BVOCs, dust and the SOA formation aligned with the northward shift of vegetation during the “Green Sahara” period, which could serve as a benchmark for MH aerosol input in future Earth system model simulation experiments.

Absorbing aerosols emitted from biomass burning (BB) greatly affect the radiation balance, cloudiness, and circulation over tropical regions. Assessments of these impacts rely heavily on the modeled aerosol absorption from poorly constrained global models and thus exhibit large uncertainties. By combining the AeroCom model ensemble with satellite and in situ observations, we provide constraints on the aerosol absorption optical depth (AAOD) over the Amazon and Africa. Our approach enables identification of error contributions from emission, lifetime, and MAC (mass absorption coefficient) per model, with MAC and emission dominating the AAOD errors over Amazon and Africa, respectively. In addition to primary emissions, our analysis suggests substantial formation of secondary organic aerosols over the Amazon but not over Africa. Furthermore, we find that differences in direct aerosol radiative effects between models decrease by threefold over the BB source and outflow regions after correcting the identified errors. This highlights the potential to greatly reduce the uncertainty in the most uncertain radiative forcing agent.

Anthropogenic aerosols play a major role for the Earth-Atmosphere system by influencing the Earth’s radiative budget and climate. The effect of the perturbation induced by changes in anthropogenic aerosols on the Earth's energy balance is quantified in terms of the effective radiative forcing (ERF) which is the recommended metric for perturbations affecting the Earth’s top-of-atmosphere energy budget since it is a better way to link this perturbation to subsequent global mean surface temperature change. In this work, the present-day ERF of anthropogenic aerosols is quantified using simulations from Earth system models (ESMs) participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6). The ERFs of individual aerosol species, such as sulphates, organic carbon (OC), and black carbon (BC) are calculated along with the ERF due to all anthropogenic aerosols and the transient ERF over the historical period (1850–2014). Additionally, ERF is analyzed into three components: (a) ERFARI, representing aerosol-radiation interactions, (b) ERFACI, accounting for aerosol-cloud interactions, and (c) ERFALB, which is mainly due to the contribution of surface albedo changes caused by anthropogenic aerosols. Here, the total anthropogenic aerosol ERF (calculated using the piClim-aer experiment) is estimated to be -1.11 ± 0.26 W m-2, mostly due to the large contribution of ERFACI (-1.14 ± 0.33 W m-2), compared to ERFARI (-0.02 ± 0.20 W m-2) and ERFALB (0.05 ± 0.07 W m-2). The total ERF caused by sulphates (piClim-SO2) is estimated at -1.11 ± 0.31 W m-2, the OC ERF (piClim-OC) is -0.35 ± 0.21 W m-2, whereas the ERF exerted by BC (piClim-BC) is 0.19 ± 0.18 W m-2. On top of that, our analysis reveals that ERFACI clearly prevails over the largest part of the Earth except for the BC experiment where ERFARI prevails over land. By the end of the historical period (1995–2014), the global mean total aerosol ERF is estimated at -1.28 ± 0.37 W m-2 (calculated using the histSST experiment). We find that sulphates dominate both present-day and transient ERF spatial patterns at the top of the atmosphere, exerting a strongly negative ERF especially over industrialized regions of the Northern Hemisphere, such as North America, Europe, East and South Asia. Since the mid-1980s ERF has become less negative over Eastern North America and Western and Central Europe, while over East and South Asia there is a steady increase in ERF magnitude towards more negative values until 2014.

Earth system models (ESMs) participating in the latest Coupled Model Intercomparison Project Phase 6 (CMIP6) simulate various components of fine particulate matter (PM2.5) as major climate forcers. Yet the model performance for PM2.5 components remains little evaluated due in part to lack of observational data. Here, we evaluate near-surface concentrations of PM2.5 and its five main components over China as simulated by fourteen CMIP6 models, including organic carbon (OC, available in 14 models), black carbon (BC, 14 models), sulfate (14 models), nitrate (4 models), and ammonium (5 models). For this purpose, we collect observational data between 2000 and 2014 from a satellite-based dataset for total PM2.5 and from 2469 measurement records in the literature for PM2.5 components. Seven models output total PM2.5 concentrations, and they all underestimate the observed total PM2.5 over eastern China, with GFDL-ESM4 (–1.5 %) and MPI-ESM-1-2-HAM (–1.1 %) exhibiting the smallest biases averaged over the whole country. The other seven models, for which we recalculate total PM2.5 from the available components output, underestimate the total PM2.5 concentrations, partly because of the missing model representations of nitrate and ammonium. Concentrations of the five individual components are underestimated in almost all models, except that sulfate is overestimated in MPI-ESM-1-2-HAM by 12.6 % and in MRI-ESM2-0 by 24.5 %. The underestimation is the largest for OC (by –71.2 % to –37.8 % across the 14 models) and the smallest for BC (–47.9 % to –12.1 %). The multi-model mean (MMM) reproduces fairly well the observed spatial pattern for OC (R = 0.51), sulfate (R = 0.57), nitrate (R = 0.70) and ammonium (R = 0.75), yet the agreement is poorer for BC (R = 0.39). The varying performances of ESMs on total PM2.5 and its components have important implications for the modeled magnitude and spatial pattern of aerosol radiative forcing.

The Radiative Forcing Model Intercomparison Project (RFMIP) allows estimates of effective radiative forcing (ERF) in the Coupled Model Intercomparison Project phase six (CMIP6). We analyze the RFMIP output, including the new experiments from models that use the same parameterization for anthropogenic aerosols (RFMIP‐SpAer), to characterize and better understand model differences in aerosol ERF. We find little changes in the aerosol ERF for 1970–2014 in the CMIP6 multi‐model mean, which implies greenhouse gases primarily explain the positive trend in the total anthropogenic ERF. Cloud‐mediated effects dominate the present‐day aerosol ERF in most models. The results highlight a regional increase in marine cloudiness due to aerosols, despite suppressed cloud lifetime effects in that RFMIP‐SpAer experiment. Negative cloud‐mediated effects mask positive direct effects in many models, which arise from strong anthropogenic aerosol absorption. The findings suggest opportunities to better constrain simulated ERF by revisiting the optical properties and long‐range transport of aerosols.

Paleo-proxy data indicates that a "Green Sahara" thrived in Northern Africa during the early- to mid-Holocene (MH; 11,000 to 5,000 years before present), characterized by more vegetation cover and reduced dust emission. Utilizing a state-of-the-art atmospheric chemical transport model TM5-MP, we assessed the changes in biogenic volatile organic compounds (BVOCs) emissions, dust emission and secondary organic aerosol (SOA) concentration in Northern Africa during this period relative to the pre-industrial (PI) period. Our simulations show that dust emissions reduced from 280.6 Tg a-1 in the PI to 26.8 Tg a-1 in the MH, agreeing with indications from eight marine sediment records in the Atlantic Ocean. The northward expansion in Northern Africa resulted in an increase in annual emissions of isoprene and monoterpenes during the MH, around 4.3 and 3.5 times higher than that in the PI period, respectively, causing 1.9 times increase in the SOA surface concentration. The enhanced SOA surface concentration and decreased sulfate surface concentration counteracted each other, leading to a 17 % increase in the cloud condensation nuclei at 0.2 % super saturation over Northern Africa. Our simulations provide consistent emission datasets of BVOCs, dust, and the SOA formation aligned with the northward shift of vegetation during the "Green Sahara" period, which could serve as a benchmark for MH aerosol input in future Earth system model simulation experiments.