Climate model uncertainty in impact assessments for agriculture: A multi‐ensemble case study on maize in sub‐Saharan Africa. Issue 3 (22nd March 2017)

• Record Type: Journal Article
• Authors: Dale, Amy; Fant, Charles; Strzepek, Kenneth; Lickley, Megan; Solomon, Susan
• Is Part Of: Earth's future. Volume 5:Issue 3(2017)
• Page Start: 337
• Page End: 353
• Publication Date: 2017-03-22
• Subjects: climate change -- maize yields -- sub‐Saharan Africa -- uncertainty -- robustness; Environmental sciences -- Periodicals; Environmental sciences; Periodicals; 550
• DOI: 10.1002/2017EF000539
• Languages: English
• ISSNs: 2328-4277

Abstract

Abstract: We present maize production in sub‐Saharan Africa as a case study in the exploration of how uncertainties in global climate change, as reflected in projections from a range of climate model ensembles, influence climate impact assessments for agriculture. The crop model AquaCrop‐OS (Food and Agriculture Organization of the United Nations) was modified to run on a 2° × 2° grid and coupled to 122 climate model projections from multi‐model ensembles for three emission scenarios (Coupled Model Intercomparison Project Phase 3 [CMIP3] SRES A1B and CMIP5 Representative Concentration Pathway [RCP] scenarios 4.5 and 8.5) as well as two "within‐model" ensembles (NCAR CCSM3 and ECHAM5/MPI‐OM) designed to capture internal variability (i.e., uncertainty due to chaos in the climate system). In spite of high uncertainty, most notably in the high‐producing semi‐arid zones, we observed robust regional and sub‐regional trends across all ensembles. In agreement with previous work, we project widespread yield losses in the Sahel region and Southern Africa, resilience in Central Africa, and sub‐regional increases in East Africa and at the southern tip of the continent. Spatial patterns of yield losses corresponded with spatial patterns of aridity increases, which were explicitly evaluated. Internal variability was a major source of uncertainty in both within‐model and between‐model ensembles and explained the majority of the spatial distribution of uncertainty in yield projections. Projected climate change impacts on maize production in different regions and nations ranged from near‐zero or positive (upper quartile estimates) to substantially negative (lower quartile estimates), highlighting a need for risk management strategies that are adaptive and robust to uncertainty. Key Points: We estimate the influence of global climate model uncertainty on projected maize yield changes in sub‐Saharan Africa due to climate change Five different GCM ensembles all project yield losses in the Sahel region and Southern Africa and sub‐regional increases in East Africa Irreducible internal variability is a major cause of uncertainty in crop model projections even out to 2090