Forecasting crop yields to increase food security: a novel biophysical and social integrated systems approach
Abstract
Food security is of key importance in developing countries, especially as climate change worsens. Yield variability due to weather events impacts 20-49% of the total production each year, particularly in the regions with high levels of food insecurity. Real-time crop yield and land use data are scarce and, where available, uncertainty is large. Combining crop models and remote sensing can help to provide rapid country-level crop yield forecasts. Physically based crop models are well developed but, given high knowledge barriers, are rarely (with few exceptions such as the CropM and MARS initiatives) integrated into governments' yield monitoring, and forecasting operations. We evaluated the hypothesis that integrating local expertise, historical yields, physically based-modeling and earth observation data can provide food-insecure-regions with robust, in-season yield predictions. Informed by local knowledge and existing data, globally available weather and soils data initiate the System Approach for Land Use Sustainability (SALUS) model inputs for real-time and future prediction simulations. Splitting training and validation years and simulating yields over a suite of locally informed management parameters enables the autonomous selection of regionally specific model parameters. Historical climate conditions are used to predict</sp
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMGC45A..41F