The grid interval of a global climate model (GCM) is generally hundreds of kilometers in latitude and longitude. The spatial heterogeneity of crop condition (e.g., phenology and yields) at that scale could be substantial. Because the atmosphere-cropland exchanges of energy, water, and materials are sensitive to crop condition, this issue poses a question: How can we simulate the condition of a crop of interest on a GCM grid scale while taking into account the spatial heterogeneity of crop condition at a sub-grid scale? We therefore proposed an ensemble approach that uses stochastic parameter values to represent the spatial variation in the phenological and biophysical characteristics of a crop within a given GCM grid box, and tested its feasibility with simulation experiments. The combination of the Soil and Water Assessment Tool (SWAT) applied to maize in the Central Great Plains, United States, and coarse-resolution (2.5° × 2.5°) reanalysis data was taken as the example. The ensemble simulations successfully captured the spatial variation in the phenology and yield. Our conclusion is that the ensemble approach is feasible and expected to benefit large-area crop modeling when extending those models to include more information on the spatial heterogeneity of crop condition than ever.
- Large-area crop model
- Markov Chain Monte Carlo method
- Soil and Water Assessment Tool (SWAT)
- Spatial heterogeneity
ASJC Scopus subject areas
- Agronomy and Crop Science
- Atmospheric Science