USING MULTISPECTRAL DIGITAL IMAGERY TO EXTRACT BIOPHYSICAL VARIABILITY OF RICE TO REFINE NUTRIENT PRESCRIPTION MODELS.

In this paper we report on two components of research towards incorporating remotely sensed field images with a crop growth and development model to predict spatial variability in yield. Initial regression analysis between crop biomass and normalised difference vegetation index (NDVI) images of a rice crop yield R2 values of up to 0.73 at the mid-tillering stage of crop development. A rice crop growth and development model specifically modified to accept crop biomass data has predicted crop yield to within an average 0.72 t/ha of the measured values. The potential of combining both the image-derived biomass estimates and the modified crop growth and development model to predict spatial variability of crop yield is also discussed