Estimating Crop Development Stages from Multispectral Data

Accurate prediction of crop yields requires knowledge of the crop's develop-ment stage at critical times during the growing season. Various meteorological models have been developed to estimate crop development. However, the crop planting date, or equivalently, the date the crop reaches a particular crop devel-· opment stage generally is needed to ini-· tialize these models. Recently, efforts have been made towards developing methods for estimating crop development stage from Landsat MSS data. We have developed a technique that is designed to estimate the calendar day that a crop reaches a parti-' cular development stage early or late in the growing season. The method requires Landsat observations from the first half of the growing season to make the early season estimate and from the last half of the season for the late season estimate. Following a description of the crop model employed we describe the application of our method to Purdue Agronomy Farm data collected by a truck-mounted Exotech-lOO radiometer and to Landsat MSS data. Preliminary results indicate that the method has good potential for making accurate estimates of the calendar day that a crop reaches a particular crop development stage (depending on training data).