Browsing by Subject "Multi-model"

Now showing 1 - 1 of 1

Reducing uncertainty in prediction of climate change impacts on crop production in Ethiopia
(2024) Rettie, Fasil Mequanint; Streck, Thilo
Ethiopia, with an economy heavily reliant on agriculture, is among the countries most vulnerable to climate change. It faces recurrent climate extreme events that result in devastating impacts and acute food shortages for millions of people. Studies that focus on their influence on agriculture, especially crop productivity, are of particular importance. However, only a few studies have been conducted in Ethiopia, and existing studies are spatially limited and show considerable spatial invariance in predicted impacts, as well as discrepancies in the sign and direction of impacts. Therefore, a robust, regionally focused, and multi-model assessment of climate change impacts is urgently needed. To guide policymaking and adaptation strategies, it is essential to quantify the impacts of climate change and distinguish the different sources of uncertainty. Against this backdrop, this study consisted of several key components. Using a multi-crop model ensemble, we began with a local climate change impact assessment on maize and wheat growth and yield across three sites in Ethiopia . We quantified the contributions of different sources of uncertainty in crop yield prediction. Our results projected a of 36 to 40% reduction in wheat grain yield by 2050, while the impact on maize was modest. A significant part of the uncertainty in the projected impact was attributed to differences in the crop growth models. Importantly, our study identified crop growth model-associated uncertainty as larger than the rest of the model components. Second, we produced a high-resolution daily projections database for rainfall and temperature to serve the requirement for impact modeling at regional and local levels using a statistical downscaling technique based on state-of-the-art GCMs under a range of emission scenarios called Shared Socioeconomic Pathways (SSPs). The evaluated results suggest that the downscaling strategy significantly reduced the biases between the GCM outputs and the observation data and minimized the errors in the projections. Third, we explored the magnitude and spatial patterns of trends in observed and projected changes in climate extremes indices based on downscaled high-resolution daily climate data to serve as a baseline for future national or regional-level impact assessment. Our results show largely significant and spatially consistent trends in temperature-derived extreme indices, while precipitation-related extreme indices are heterogeneous in terms of spatial distribution, magnitude, and statistical significance coverage. The projected changes in temperature-related indices are dominated by the uncertainties in the GCMs, followed by uncertainties in the SSPs. Unlike the temperature-related indices, the uncertainty from internal climate variability constitutes a considerable proportion of the total uncertainty in the projected trends. Fourth, we examined the regional-scale impact of climate change on maize and wheat yields by crop modeling, in which we calibrated and validated three process-based crop models to guide the design of national-level adaptation strategies in Ethiopia. Our analysis showed that under a high-emissions scenario, the national-level median wheat yield is expected to decrease by 4%, while maize yield is expected to increase by 2.5% by the end of the century. The CO2 fertilization effect on the crop simulations would offset the projected negative impact. Crop model spread followed by GCMs was identified as the largest contributor to overall uncertainty to the estimated yield changes. In summary, our study quantifies the impact of climate change and demonstrates the importance of a multi-model ensemble approach. We highlight the significant impacts of climate change on wheat yield in Ethiopia and the importance of crop model improvements to reduce overall uncertainty in the projected impact.