Fig. 3: The proportion of uncertainty of yield changes (2069–2099 compared with 1980–2010, under SSP585) for different numbers of GGCMs used for wheat, maize, rice, and soybean. | Communications Earth & Environment

Fig. 3: The proportion of uncertainty of yield changes (2069–2099 compared with 1980–2010, under SSP585) for different numbers of GGCMs used for wheat, maize, rice, and soybean.

From: The optimization of model ensemble composition and size can enhance the robustness of crop yield projections

Fig. 3: The proportion of uncertainty of yield changes (2069–2099 compared with 1980–2010, under SSP585) for different numbers of GGCMs used for wheat, maize, rice, and soybean.The alternative text for this image may have been generated using AI.

The numbers 2–9 (and 2–8) represent different numbers of GGCMs used, including all of the possible combinations for wheat and maize (rice and soybean). The boxplot shows the distribution of uncertainty due to GGCM selection under different numbers of GGCMs used. The dark and light blue shaded areas respectively represent the thresholds of ±2.5% and ±5% when using the contribution of uncertainty estimated from all GGCMs as the benchmark. Box boundaries indicate the 25th and 75th percentiles. The black line within each box indicates the median value. GGCM is a global gridded crop model; GCM is a global climate model. S3 is the scenario when one model was randomly selected from each cluster (total 3 crop models). S4 is the scenario when four models were selected ensuring that at least one model belonged to each cluster.

Back to article page