Drought severity identification and classification of the land pattern using landsat 8 data based on spectral indices and maximum likelihood algorithm

Abstract

The manual survey of drought severity is very hectic and time-consuming task. This paper reports the study to assess the adeptness of satellite-based drought indices for observing the spatiotemporal extent of agricultural drought events. The Land Use Land Cover (LULC) has been categorized into six classes such as Vegetation, Settlement, Barren land, Harvested land, Hill with rocks and Water bodies and computed using Maximum Likelihood (ML) supervised algorithm. Moreover, an attempt has been made to analyze the drought condition using multi-date Landsat 8 images of Vaijapur taluka which falls in drought-prone zones. The severity of drought was determined and defined based on the Normalized Difference Vegetation Index (NDVI) with a good outcome. The drought severity was classified into three groups viz severe, moderate, and normal. The present study shows that the entire study area was affected by the worst drought condition during the period of 2013 and 2014. The experimental results examine that, the overall accuracy of ML classifier was 81.31% with kappa coefficient 0.81 for the year 2013 and it was 78.02% with a kappa value of 0.73 for the year 2014. The present study is essential for the assessment of drought condition with advanced technology before the drought get severe.