Estimation des bilans hydriques présent et futur ainsi que de leur impact sur les rendements agricoles en Afrique du Nord et de l'Ouest à l'aide d'un modèle de végétation dynamique

Climate change is affecting agriculture all over the world with dramatic consequences in developing countries already affected by limited water resources. Unfortunately, the lack of a consistent coverage of hydrological data (precipitation, soil water content, etc.), especially in those regions, makes difficult a good knowledge of the current conditions and, thus, prevents any planning for the future. Therefore, ecosystem process-based models are interesting tools to represent the water balance of terrestrial ecosystems and provide an overview of the current and future hydrological conditions. Here, we use the CARAIB (CARbon Assimilation In the Biosphere; Dury et al., 2011) dynamic vegetation model (DVM) to evaluate current and future soil water availability and assess potential impacts of its reduction on crop yields. CARAIB is a state-of-the-art DVM with various modules dealing with (i) soil hydrology, (ii) photosynthesis/stomatal regulation, (iii) carbon allocation and biomass growth, (iv) litter/soil carbon dynamics, (v) vegetation cover dynamics, (vi) seed dispersal, and (vii) vegetation fires. Climate and atmospheric CO2 are the primary inputs. Within the VOTES project (Fontaine et al., 2014) and the GGCMI intercomparison (Global Gridded Crop Model Intercomparison), the model has been improved to include crops and meadows (Minet et al., 2015), and some modules have been written to translate model outputs into quantitative indicators of ecosystem services (e.g., to evaluate crop yield from net primary productivity, to calculate soil erosion from runoff and slope or to apply irrigation water).

In this contribution, we focus on four crop species (maize, rice, soy and spring wheat) over North and West Africa. We use historical climate and future projections from the ISI-MIP2 model intercomparison (Inter-Sectoral Impact Model Intercomparison Project). Historical climate is PGMFD v.2.1 (NCEP/NCAR Reanalysis 1) 1901-2012 time series. Future change in temperature and precipitation are produced over the 21st century by CMIP5 general circulation model simulations that were bias-corrected in the framework of the ISI-MIP2 project. Simulations of the vegetation model are performed at 0.5° spatial resolution under two water scenarios: a water-limited (precipitation only) and a full irrigation (water supplied to reach field capacity). In the model, the soil water available for plant (ASW) is expressed as the fraction between the actual soil water and the water available for plant growth at field capacity (FC): ASW = (SW-WP)/(FC-WP) with SW being the actual soil water and WP the wilting point. Under the first scenario, the soil water deficit and its impacts on crop production are evaluated. The second scenario allows an evaluation of the amount of irrigation water necessary to fulfill crop requirements. This kind of study can help in the identification of agricultural areas currently vulnerable or which might be exposed in the future. It can also be useful to optimise water resource management in African countries.