Simulating regional inter-annual crop yield variability over multiple decades with the Community Land Model (CLM5)

Global climate change with a predicted increase in weather extremes entails vulnerability and new challenges to regional agriculture. While the general impacts of climate change on global food security are a much studied topic, the implications for regional inter-annual yield variability remain unclear. In this study, we analysed the effects of weather trends on regional crop productivity within two agriculturally managed regions in different climate zones, simulated with the latest version of the Community Land Model (version 5.0) over two decades (1999-2019). We evaluated the models’ potential to represent the inter-annual variability of crop yield in comparison to recorded yield variability and different weather indicators, e.g., drought index and growing season length and evaluated which variables (i.e., temperature, precipitation, initial soil moisture content) dominantly drive changes in CLM5-predicted yield variability. The simulation results were able to reproduce the sign of crop yield anomalies, and thus provide a basis on which to study the effects of different weather patterns on inter-annual yield variability. However, the simulations showed limitations in correctly capturing inter-annual differences of crop yield in terms of total magnitudes (up to 10 times lower than in official records). Our results indicate that these limitation arise mainly from uncertainties in the representation of the subsurface soil moisture regime and a corresponding lack of sensitivity towards drought stress. Insights from this work were used to summarize implications for future analysis of CLM5-BGC simulation results over agriculturally managed land and allowed us to discuss and investigate possible technical model improvements.