Composition du jury

Présidente : Agnès Ducharne, CNRS, METIS/IPSL
Directeur de thèse : Olivier Boucher, CNRS, IPSL
Co-directeur de thèse : Philippe Ciais, CEA, LSCE/IPSL
Rapporteur: Stephen Sitch, University of Exeter
Rapporteur: Sönke Zaehle, Max Planck Institute
Examinateur: Roland Séférian, CNRM, Météo-France/CNRS
Examinatrice: Lingli Liu, Chinese Academy of Sciences
Invité : Laurent Li, CNRS, LMD/IPSL

Résumé

Anthropogenic atmospheric aerosols have been recognized to have significantly affected the climate system through their interactions with radiation and cloud during the last decades. Besides these well-known but poorly-understood physical processes in the atmosphere, recent studies reported strong influences of aerosols on the carbon cycle, especially its terrestrial component. The changes in carbon cycle will further alter the climate through the climate-carbon feedback. It remains uncertain how much anthropogenic aerosols perturb the land carbon cycle. This thesis aims to quantify and attribute the impacts of anthropogenic aerosols on the terrestrial cycle using a modeling approach.
In Chapter 2, a set of offline simulations using the ORCHIDEE land surface model driven by climate fields from different CMIP5 generation climate models were performed to investigate the impacts of anthropogenic aerosols on the land C cycle through their impacts on climate. The results indicate an increased cumulative land C sink of 11.6-41.8 PgC during 1850-2005 due to anthropogenic aerosols. The increase in net biome production (NBP) is mainly found in the tropics and northern mid latitudes. Aerosol-induced cooling is the main factor driving this NBP changes. At high latitudes, aerosol-induced cooling caused a stronger decrease in gross primary production (GPP) than in total ecosystem respiration (TER), leading to lower NBP. At mid latitudes, cooling‐induced decrease in TER is stronger than for GPP, resulting in a net NBP increase. At low latitudes, NBP was also enhanced due to the cooling‐induced GPP increase, but regional precipitation decline in response to anthropogenic aerosol emissions may negate the effect of temperature. As climate models currently disagree on how aerosol emissions affect tropical precipitation, the precipitation change in response to aerosols becomes the main source of uncertainty in aerosol-caused C flux changes. The results suggest that better understanding and simulation of how anthropogenic aerosols affect precipitation in climate
models is required for a more accurate attribution of aerosol effects on the terrestrial carbon cycle.
Chapter 3 presents the development and evaluation of a new version of ORCHIDEE model named ORCHIDEE_DF. Compared with the standard ORCHIDEE model (ORCHIDEE trunk), ORCHIDEE_DF includes a new light partitioning module to separate the downward shortwave radiation into direct and diffuse components, as well as a new canopy radiative transmission module to simulate the transmission of diffuse and direct radiation, and the light absorption of sunlit and shaded leaves separately. The new model ORCHIDEE_DF was evaluated using flux observations from 159 eddy covariance sites over the globe. Compared with the original model, ORCHIDEE_DF improves the GPP simulation under sunny conditions and captures the observed higher photosynthesis under cloudier conditions for most plant functional types (PFTs). The results from ORCHIDEE_DF and standard ORCHIDEE together indicate that the larger GPP under cloudy conditions compared to sunny conditions is mainly driven by increased diffuse light in the morning and in the afternoon, and by the decreased water vapor pressure deficit (VPD) and air temperature at midday. The strongest positive effects of diffuse light on photosynthesis are found in the range 5-20 °C and VPD<1 kPa. This effect is found to decrease when VPD becomes too large, or temperature falls outside that range, likely because stomatal conductance takes control of photosynthesis. ORCHIDEE_DF underestimates the diffuse light effect at low temperature in all PFTs and overestimates this effect at high temperature and high VPD in grasslands and croplands. This bias is likely due to the parameterization in the original model. The new model has the potential to better investigate the impact of large-scale aerosol changes on the terrestrial carbon budget, both in the historical period and in the context of future air quality policies and/or climate engineering.
Using empirically tuned ORCHIDEE_DF, Chapter 4 performed two sets of simulations based on the observation-based CRUJRA climate dataset, and the climate fields from IPSL-CM6A-LR simulations to systematically investigate the impacts of aerosol-induced changes in diffuse radiation and other factors. The two sets of simulations find an enhanced cumulative land C sink of 6.8 PgC (CRUJRA climate) and 15.9 PgC (IPSL-CM6A-LR climate) in response to the anthropogenic aerosol-caused diffuse radiation fraction changes during the historical period and this enhancement mainly occurs after the 1950s.A series of factorial simulations driven by IPSL climate show that globally, the anthropogenic aerosol-induced land C sink increase is mainly due to the diffuse light fertilization effect but there is also a contribution from the cooling effect. Furthermore, a comparison of different methods reconstructing the diffuse radiation field under no anthropogenic aerosol scenario indicates that correctly considering the variability of diffuse radiation fraction is essential to obtain unbiased carbon fluxes.
Although this thesis gained a relatively systematic understanding in aerosol impacts on terrestrial ecosystems, there remain uncertainties due to the limit of current modeling tools and experimental designs. To reduce these uncertainties, future work needs to include representation of currently missing mechanisms (e.g., deposition of nutrients associated with aerosols) into land surface models, collect more relevant observations for calibration, and design experiments to investigate aerosol impacts in fully coupled simulations. With the help of reliable future scenarios in aerosol emissions (such as those from the SSP-RCP dataset), improved modeling tools are expected to better evaluate the aerosol-related air quality policies.