Improving the DSSAT Crop Models for Response to Climate, Soils, and Management: Past, Present, and Future.

The chronology of code improvements to the DSSAT (Decision Support System for Agrotechnology Transfer) models is reviewed, giving present status, and future needs and plans.  The initial strategy in DSSAT beginning in the mid-1980s was to have common input files of weather, soils, and management for different models, CERES for cereals and the GRO models for three legumes.  The next step in the early 1990s was to create a generic template model, CROPGRO, in which one common FORTRAN code was sufficient for many crops, where any crop-specific parameterization was removed from the code, and placed into read-in species files for each crop.  During this time, CO₂ response functions were parameterized.  Early in 2000, the group decided that the needs of crop rotation/sequence dictated that the crops share the same land unit where the soil water, N, and C simulations carried over to the next crop or fallow.  This version was named the cropping system model (CSM) and it incorporated a daily-time step of CENTURY as an organic C module.  From about 2000 to present, additional crops were added to the DSSAT, many template crops such as tomato, cotton, fababean, chickpea, macuna, pigeonpea, safflower, and perennial tropical forages (the latter in a standalone environment).  During the decade from 2006 to present, model coding for responses to temperature and CO₂ were re-evaluated, and code improvements made.   Recent emphasis has been to add phosphorus and potassium response to the models.  The future needs include:  rigorous testing against evapotranspiration data and water deficit experiments, improvement of elevated temperature stress effects, creation of more mechanistic rooting in soil, and linkage of genetic coefficients to genetic markers (QTLs).  Model improvement is benefitting from voluntary collaboration among crop modelers, open and sharing of data, and open-source code development.