Development of PCA-based composite drought index for agricultural drought assessment using remote- sensing

Abstract

The present study was conducted in the Saurashtra region of Gujarat to demonstrate the development and validation of location and crop-specific composite drought index (CDI) using a linear combination of three parameters including meteorological drought index, vegetation drought index and inverse of maximum consecutive dry days%for major Kharif crops of the region i.e. cotton and groundnut. The performance of nine drought indices including six meteorological and three remote sensing-based vegetation indices wasevaluated in terms of correlation with district scale crop yields.The district-wise expressions of CDI were developed by assigning principal component analysis (PCA) based weights to parameters.Standardized Precipitation Evapotranspiration Index (SPEI)/ Reconnaissance Drought Index (RDI) among meteorological indices and NDVI Anomaly Index (NDVIA)/ Vegetation Condition Index (VCI) among vegetation indices were found suitable for generating district specific CDI expressions. The developed CDI showed higher correlation with Kharif cotton and Groundnut crop yields as compared to various meteorological as well as vegetation indices used in the study and effectively quantified major historic agricultural droughts.The average correlation coefficients of developed CDI with cotton and groundnut yields were 0.71 and 0.77 respectively. The correlations of CDI and crop yields for all CDI expression were highly significant with p<0.01. The method developed in the study will be useful to generate crop and region-specific multi-scalar drought indices by the amalgamation of multiple drought indices for assessing crop production losses.

Keywords

• Drought monitoring
• Cotton
• Groundnut
• SPEI
• NDVI
• VCI