Publication

Abstract : Remote sensing is one of the major techniques employed for studying agricultural patterns which tend to be highly dynamic these days. This paper delves into the possibility of finding the yield of coconut crop with hyper spectral remote sensing technology. Hyper spectral images with the advantage of having hundreds of contiguous spectral bands make crop yield monitoring a reality. Yield monitoring and crop growth assessment are very important to be carried out both at the state and national level of a country to estimate the production. In this work coconut yield is estimated by using hyper spectral images captured by the Hyperion sensor deployed on the EO-1 satellite of NASA. Unsupervised classification techniques have been used to identify the different classes present in the dataset. The process of identifying different classes extends from locating the crop under study among the different features on ground to identifying variations in the crop according to the health of the plant, pest infestation, and various other factors. The outcome of the work is evaluated in terms of the spectral reflectance of different classes. The canopy reflectance has a direct implication on crop chlorophyll and thereby their health throwing insight into the yield of the crop. Results obtained are compared with real time reflectance values measured using spectro radiometer and it is found that the hyper spectral image processing is performing on par with the hardware measurements. The health condition which is derived from these reflectance curves gives information about the yield (in lakh nuts) per acre for each class of the crop.