Spectral predictors of crop development and yield

Remote sensing enters still wider into its application stage when the goal is to bring the up-to-now investigation results to an operational use. Agricultural monitoring is among the priorities of remote sensing observations supplying early information on crop growth. Interest is rapidly spreading over the past years in the application of hyperspectral data to precision farming. In this paper, we propose and investigate the performance of an approach for providing crop state assessment and yield forecasts. In order crop information to be obtained from remotely sensed data the approach comprised: development of models between plant spectral reflectance and biophysical parameters for estimation of crop state variables from radiometric data; development of yield forecasting models from single-date and time-series spectral data; verification of remote sensing predictions through comparison with estimations from yield relationships with crop agronomical parameters. The algorithm was realized on winter wheat. In-situ high- resolution visible and near-infrared reflectance data were acquired throughout the growing season, along with detailed datasets of crop parameters. Spectral-biophysical models were developed relating crop variables and yield to different spectral predictors. The algorithm was tested and validated using airborne remote sensing data. A good correspondence was found between predicted and actual yield.