Linking Satellite Imagery, Surveying and Crop Modeling to Assess Impacts of Climate Change on Maize Production at District Level in South Africa.

Although most dynamic crop models have been developed and tested for the scale of a homogeneous plot, applications related to climate change are often at broader spatial scales that can incorporate considerable heterogeneity. The approach followed in this study was to base crop model simulations on a field level maize crop mask that was developed using satellite imagery and crop type classification. The impact of projected climate change on maize production was assessed in three districts, Bloemfontein, Bethlehem and Bothaville in the Free State Province of South Africa using the DSSAT crop model. Crop management such as row spacing, plant population and planting dates were derived from objective yield surveys and associated with the fields proportionally to their occurrence. GIS and peodo-transfer functions were used to derive soil profile descriptions for each field based on land types. Fertilization was based on the yield potential of each field. Past (1980-2010) and future (5 GCMs for the time period 2040-2070, with RCP 8.5 and CO2 of 571 ppm) maize productivity was simulated for each field. Field level simulations have the advantage that they can be summarized to different levels such as, quinary catchments or districts and can be presented in map or graph format.