Matthieu Guimberteau

Assessing climate change impacts on sorghum and millet yields in the Sudanian and Sahelian savannas of West Africa

Sub-Saharan West Africa is a vulnerable region where a better quantification and understanding of the impact of climate change on crop yields is urgently needed. Here, we have applied the process-based crop model SARRA-H calibrated and validated over multi-year field trials and surveys at eight contrasting sites in terms of climate and agricultural practices in Senegal, Mali, Burkina Faso and Niger. The model gives a reasonable correlation with observed yields of sorghum and millet under a range of cultivars and traditional crop management practices. We applied the model to more than 7000 simulations of yields of sorghum and millet for 35 stations across West Africa and under very different future climate conditions. We took into account 35 possible climate scenarios by combining precipitation anomalies from 20% to 20% and temperature anomalies fromC0 toC6 C. We found that most of the 35 scenarios (31/35) showed a negative impact on yields, up to 41% forC6 C= 20% rainfall. Moreover, the potential future climate impacts on yields are very different from those recorded in the recent past. This is because of the increasingly adverse role of higher temperatures in reducing crop yields, irrespective of rainfall changes. When warming exceedsC2 C, negative impacts caused by temperature rise cannot be counteracted by any rainfall change. The probability of a yield reduction appears to be greater in the Sudanian region (southern Senegal, Mali, Burkina Faso, northern Togo and Benin), because of an exacerbated sensitivity to temperature changes compared to the Sahelian region (Niger, Mali, northern parts of Senegal and Burkina Faso), where crop yields are more sensitive to rainfall change. Finally, our simulations show that the photoperiod-sensitive traditional cultivars of millet and sorghum used by local farmers for centuries seem more resilient to future climate conditions than modern cultivars bred for their high yield potential ( 28% versus 40% for theC4 C= 20% scenario). Photoperiod-sensitive cultivars counteract the effect of temperature increase on shortening cultivar duration and thus would likely avoid the need to shift to cultivars with a greater thermal time requirement. However, given the large difference in mean yields of the modern versus traditional varieties, the modern varieties would still yield more under optimal fertility conditions in a warmer world, even if they are more affected by climate change.

Variation in stem mortality rates determines patterns of above-ground biomass in Amazonian forests: implications for dynamic global vegetation models

Abstract Understanding the processes that determine above‐ground biomass (AGB) in Amazonian forests is important for predicting the sensitivity of these ecosystems to environmental change and for designing and evaluating dynamic global vegetation models (DGVMs). AGB is determined by inputs from woody productivity [woody net primary productivity (NPP)] and the rate at which carbon is lost through tree mortality. Here, we test whether two direct metrics of tree mortality (the absolute rate of woody biomass loss and the rate of stem mortality) and/or woody NPP, control variation in AGB among 167 plots in intact forest across Amazonia. We then compare these relationships and the observed variation in AGB and woody NPP with the predictions of four DGVMs. The observations show that stem mortality rates, rather than absolute rates of woody biomass loss, are the most important predictor of AGB, which is consistent with the importance of stand size structure for determining spatial variation in AGB. The relationship between stem mortality rates and AGB varies among different regions of Amazonia, indicating that variation in wood density and height/diameter relationships also influences AGB. In contrast to previous findings, we find that woody NPP is not correlated with stem mortality rates and is weakly positively correlated with AGB. Across the four models, basin‐wide average AGB is similar to the mean of the observations. However, the models consistently overestimate woody NPP and poorly represent the spatial patterns of both AGB and woody NPP estimated using plot data. In marked contrast to the observations, DGVMs typically show strong positive relationships between woody NPP and AGB. Resolving these differences will require incorporating forest size structure, mechanistic models of stem mortality and variation in functional composition in DGVMs.

Water Balance in the Amazon Basin from a Land Surface Model Ensemble

AbstractDespite recent advances in land surface modeling and remote sensing, estimates of the global water budget are still fairly uncertain. This study aims to evaluate the water budget of the Amazon basin based on several state-of-the-art land surface model (LSM) outputs. Water budget variables (terrestrial water storage TWS, evapotranspiration ET, surface runoff R, and base flow B) are evaluated at the basin scale using both remote sensing and in situ data. Meteorological forcings at a 3-hourly time step and 1° spatial resolution were used to run 14 LSMs. Precipitation datasets that have been rescaled to match monthly Global Precipitation Climatology Project (GPCP) and Global Precipitation Climatology Centre (GPCC) datasets and the daily Hydrologie du Bassin de l’Amazone (HYBAM) dataset were used to perform three experiments. The Hydrological Modeling and Analysis Platform (HyMAP) river routing scheme was forced with R and B and simulated discharges are compared against observations at 165 gauges. Simu...

Future changes in precipitation and impacts on extreme streamflow over Amazonian sub-basins

Because of climate change, much attention is drawn to the Amazon River basin, whose hydrology has already been strongly affected by extreme events during the past 20 years. Hydrological annual extreme variations (i.e. low/high flows) associated with precipitation (and evapotranspiration) changes are investigated over the Amazon River sub-basins using the land surface model ORCHIDEE and a multimodel approach. Climate change scenarios from up to eight AR4 Global Climate Models based on three emission scenarios were used to build future hydrological projections in the region, for two periods of the 21st century. For the middle of the century under the SRESA1B scenario, no change is found in high flow on the main stem of the Amazon River (Obidos station), but a systematic discharge decrease is simulated during the recession period, leading to a 10% low-flow decrease. Contrasting discharge variations are pointed out depending on the location in the basin. In the western upper part of the basin, which undergoes an annual persistent increase in precipitation, high flow shows a 7% relative increase for the middle of the 21st century and the signal is enhanced for the end of the century (12%). By contrast, simulated precipitation decreases during the dry seasons over the southern, eastern and northern parts of the basin lead to significant low-flow decrease at several stations, especially in the Xingu River, where it reaches -50%, associated with a 9% reduction in the runoff coefficient. A 18% high-flow decrease is also found in this river. In the north, the low-flow decrease becomes higher toward the east: a 55% significant decrease in the eastern Branco River is associated with a 13% reduction in the runoff coefficient. The estimation of the streamflow elasticity to precipitation indicates that southern sub-basins (except for the mountainous Beni River), that have low runoff coefficients, will become more responsive to precipitation change (with a 5 to near 35% increase in elasticity) than the western sub-basins, experiencing high runoff coefficient and no change in streamflow elasticity to precipitation. These projections raise important issues for populations living near the rivers whose activity is regulated by the present annual cycle of waters. The question of their adaptability has already arisen.

Evaluation of the ORCHIDEE ecosystem model over Africa against 25 years of satellite-based water and carbon measurements

Few studies have evaluated land surface models for African ecosystems. Here we evaluate the Organizing Carbon and Hydrology in Dynamic Ecosystems (ORCHIDEE) process-based model for the interannual variability (IAV) of the fraction of absorbed active radiation, the gross primary productivity (GPP), soil moisture, and evapotranspiration (ET). Two ORCHIDEE versions are tested, which differ by their soil hydrology parameterization, one with a two-layer simple bucket and the other a more complex 11-layer soil-water diffusion. In addition, we evaluate the sensitivity of climate forcing data, atmospheric CO2, and soil depth. Beside a very generic vegetation parameterization, ORCHIDEE simulates rather well the IAV of GPP and ET (0.5 < r < 0.9 interannual correlation) over Africa except in forestlands. The ORCHIDEE 11-layer version outperforms the two-layer version for simulating IAV of soil moisture, whereas both versions have similar performance of GPP and ET. Effects of CO2 trends, and of variable soil depth on the IAV of GPP, ET, and soil moisture are small, although these drivers influence the trends of these variables. The meteorological forcing data appear to be quite important for faithfully reproducing the IAV of simulated variables, suggesting that in regions with sparse weather station data, the model uncertainty is strongly related to uncertain meteorological forcing. Simulated variables are positively and strongly correlated with precipitation but negatively and weakly correlated with temperature and solar radiation. Model-derived and observation-based sensitivities are in agreement for the driving role of precipitation. However, the modeled GPP is too sensitive to precipitation, suggesting that processes such as increased water use efficiency during drought need to be incorporated in ORCHIDEE.

Multicriteria evaluation of discharge simulation in Dynamic Global Vegetation Models

In this study, we assessed the performance of discharge simulations by coupling the runoff from seven Dynamic Global Vegetation Models (DGVMs; LPJ, ORCHIDEE, Sheffield-DGVM, TRIFFID, LPJ-GUESS, CLM4CN, and OCN) to one river routing model for 16 large river basins. The results show that the seasonal cycle of river discharge is generally modeled well in the low and middle latitudes but not in the high latitudes, where the peak discharge (due to snow and ice melting) is underestimated. For the annual mean discharge, the DGVMs chained with the routing model show an underestimation. Furthermore, the 30 year trend of discharge is also underestimated. For the interannual variability of discharge, a skill score based on overlapping of probability density functions (PDFs) suggests that most models correctly reproduce the observed variability (correlation coefficient higher than 0.5; i.e., models account for 50% of observed interannual variability) except for the Lena, Yenisei, Yukon, and the Congo river basins. In addition, we compared the simulated runoff from different simulations where models were forced with either fixed or varying land use. This suggests that both seasonal and annual mean runoff has been little affected by land use change but that the trend itself of runoff is sensitive to land use change. None of the models when considered individually show significantly better performances than any other and in all basins. This suggests that based on current modeling capability, a regional-weighted average of multimodel ensemble projections might be appropriate to reduce the bias in future projection of global river discharge.

Diagnosing phosphorus limitations in natural terrestrial ecosystems in carbon cycle models

Abstract Most of the Earth System Models (ESMs) project increases in net primary productivity (NPP) and terrestrial carbon (C) storage during the 21st century. Despite empirical evidence that limited availability of phosphorus (P) may limit the response of NPP to increasing atmospheric CO2, none of the ESMs used in the previous Intergovernmental Panel on Climate Change assessment accounted for P limitation. We diagnosed from ESM simulations the amount of P need to support increases in carbon uptake by natural ecosystems using two approaches: the demand derived from (1) changes in C stocks and (2) changes in NPP. The C stock‐based additional P demand was estimated to range between −31 and 193 Tg P and between −89 and 262 Tg P for Representative Concentration Pathway (RCP) 2.6 and RCP8.5, respectively, with negative values indicating a P surplus. The NPP‐based demand, which takes ecosystem P recycling into account, results in a significantly higher P demand of 648–1606 Tg P for RCP2.6 and 924–2110 Tg P for RCP8.5. We found that the P demand is sensitive to the turnover of P in decomposing plant material, explaining the large differences between the NPP‐based demand and C stock‐based demand. The discrepancy between diagnosed P demand and actual P availability (potential P deficit) depends mainly on the assumptions about availability of the different soil P forms. Overall, future P limitation strongly depends on both soil P availability and P recycling on ecosystem scale.

Matrix‐Based Sensitivity Assessment of Soil Organic Carbon Storage: A Case Study from the ORCHIDEE‐MICT Model

Abstract Modeling of global soil organic carbon (SOC) is accompanied by large uncertainties. The heavy computational requirement limits our flexibility in disentangling uncertainty sources especially in high latitudes. We build a structured sensitivity analyzing framework through reorganizing the Organizing Carbon and Hydrology in Dynamic Ecosystems (ORCHIDEE)‐aMeliorated Interactions between Carbon and Temperature (MICT) model with vertically discretized SOC into one matrix equation, which brings flexibility in comprehensive sensitivity assessment. Through Sobols method enabled by the matrix, we systematically rank 34 relevant parameters according to variance explained by each parameter and find a strong control of carbon input and turnover time on long‐term SOC storages. From further analyses for each soil layer and regional assessment, we find that the active layer depth plays a critical role in the vertical distribution of SOC and SOC equilibrium stocks in northern high latitudes (>50°N). However, the impact of active layer depth on SOC is highly interactive and nonlinear, varying across soil layers and grid cells. The stronger impact of active layer depth on SOC comes from regions with shallow active layer depth (e.g., the northernmost part of America, Asia, and some Greenland regions). The model is sensitive to the parameter that controls vertical mixing (cryoturbation rate) but only when the vertical carbon input from vegetation is limited since the effect of vertical mixing is relatively small. And the current model structure may still lack mechanisms that effectively bury nonrecalcitrant SOC. We envision a future with more comprehensive model intercomparisons and assessments with an ensemble of land carbon models adopting the matrix‐based sensitivity framework.