Soil data aggregation effects in regional yield simulations

Large-scale yield simulations often use data of coarse spatial resolution as input for process-based models. However, using aggregated data as input for process-based models entails the risks of introducing errors due to aggregation (AE). Such AE depend on the aggregation method, on the type of aggregated data as well as on its spatial heterogeneity. However, previous studies indicated that AE in Central Europe might be largely driven by aggregating soil data. AE in yield could therefore be assessed prior to simulation for those regions with a distinct relationship between spatial yield variability and soil heterogeneity. The present study investigates the AE for soil data and its contribution to the total AE for soil and climate data for a range of different crop models. Soil data is aggregated by area majority in order to maintain physical consistency among soil variables. AE are assessed for climate and soil data in North Rhine-Westphalia, German, upscaling from 1 to 100 km resolution. We present a model comparison on AE for a range of environmental conditions differing in climate and soil for two crops grown under water-limited conditions. Winter wheat and silage maize yields of 1982-2011 were simulated with crop models after calibration to average regional sowing date, harvest date and crop yield. Results point to the importance of estimating AE for soil data. Ways to generalize from these results to other regions are discussed.