Developing robust Vegetation Health Indices by optimizing the weights for vegetation and temperature condition indices using principal component analysis

Study region: Punjab Province and different agro-ecological zones of Pakistan. Study focus: This study aimed to develop three robust Vegetation Health Indices (VHIs) to validate the assumption that weights assigned to Vegetation Condition Index (VCI) and Temperature Condition Index (TCI) to calculate VHI should be optimized rather than fixed at 0.5 to account for climate variability and land use changes. To this end, this study develops two novel VHIs, VHIa and VHIg, using weighted arithmetic and geometric means of VCI and TCI, where Principal Component Analysis (PCA) is used to compute the spatially varying weights for VCI and TCI at a local scale (0.25° × 0.25°). Further, the VHImin takes the minima of VCI and TCI considering the bucket effect. The performance of newly developed VHIs are assessed against other drought indices and their impacts on wheat and rice yields across Punjab province. New hydrological insights for the region: Results reported better performance of VHIa with PCA weights and VHImin. In various agro-ecological zones (AEZs) in Pakistan, VHIa performs better across humid/sub-humid regions having significant vegetation cover and agricultural activities. In contrast, VHImin is suited for arid and hyper-arid regions. Bootstrap Quantile Regression reported robust performance of VHIa and VHImin in analyzing the impact of drought on crop yield. This study recommends that weights of VCI and TCI should be optimized to account for climate variability and land use changes.