Publications

Prasad, NR; Patel, NR; Danodia, A (2021). Cotton Yield Estimation Using Phenological Metrics Derived from Long-Term MODIS Data. JOURNAL OF THE INDIAN SOCIETY OF REMOTE SENSING, 49(11), 2597-2610.

Abstract
Crop phenology plays a vital role in crop productivity and has been emphasized as the key indicator of crop growth stages. Time-series 16 day composite of Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) data from 2011 to 2018 were quantified to study the spatiotemporal and inter-annual changes in crop phenology with smoothening of NDVI curve by Savitzky-Golay filtering method to drive phenological metrics and predict the yield of cotton for the part of Maharashtra state. Two dates of Landsat-8 datasets with supervised classification-based maximum-likelihood classifier were used to generate crop inventory map. The overall accuracy of the map was 89.1% with a 0.79 kappa coefficient and further used in the generation of cotton crop mask. A correlation was established between cotton yield and the phenological metrics. A multiple regression model constructed with phenological metrics from the year 2011-2016 produced a correlation with 0.85 adjusted R-2, 46.97 standard estimated error, 10(-6)p value. The fitted model was used to estimate the cotton yield of 2016-2018 with 82.29 kg ha(-1) root mean square error, 36.13% mean absolute percentage error, and 0.81 Index of Agreement. The results have shown the potential of the model in capturing the variation in the sowing time, end of the crop season, length of growing period, peak NDVI time, peak NDVI value and amplitude in spatiotemporal scale for drought and normal years on a long-term basis.

DOI:
10.1007/s12524-021-01414-6

ISSN:
0255-660X