Defining optimum spectral narrow bands and bandwidths for agricultural applications.

Abstract

In this study, an attempt was made to define the optimum set of spectral narrowbands and the required bandwidth for agricultural applications. Spectral observations were collected using ASD handheld spectroradiometer (325-1075 nm) from major kharif (rainy) and rabi (winter) season crops, two different soil types and crops under different agronomic treatments. To identify best bands, a stepwise discriminant analysis (SDA) was carried out for each data set. Aggregating the bands selected from individual SDA, 13 optimum narrow bands were identified for crop and soil assessment. To find optimum bandwidth, the measured reflectance was aggregated to different bandwidths (3, 5, 10, 15, 20, 25 and 30 nm). The reflectance values at different wavelength regions were compared with the original spectra using root mean square error. It was found that the optimum bandwidth required for crop discrimination differed for different wavelength regions.