Abstract
Nowadays, drought monitoring with various probabilistic indices has become common. However, the interpretation and applicability issues of multi-scalar drought indices are the main problems in establishing accurate drought mitigation policies. In addition, the spatial structure of environmental variables such as rainfall, and the spatial distribution of meteorological stations have a vital role in the precise and accurate analysis. In this paper, a comprehensive drought index "the Multi-Scalar Aggregative Standardized Precipitation Temperature Index (MASPTI)" is proposed. In MASPTI procedure, temporal vectors of various time scales of SPTI index are accumulated by giving long term transient weights. These weights are determined from the steady state probabilities of drought classification states in each time scale. Application of the proposed index is based on spatio-temporal data of SPTI index at its various time scales. However, before proceeding to evaluate MASPTI, we first observed the spatial relevancy of important time scale of SPTI index using the machine learning wrapper Boruta algorithm. The preliminary evaluation of MASPTI is based on four meteorological stations located in different homogeneous climatic clusters in Pakistan. The comparative analysis includes the ordinal association, where historical qualitative series of drought classes attained from MASPTI are compared with existing SPTI time scales. Outcomes show that MASPTI has the ability to capture joint characterization of drought by incorporating long term probabilities as a transient weight.