Statistical Modeling of the Monthly Palmer Drought Severity Index

Abstract

This paper presents a new approach for the modeling of dry period interarrival times. The new model targets dry period interarrival times directly and uses a combination of the empirical model and results from extreme-value theory to achieve good fits. Existing approaches initially model dry period duration and successive wet period duration, then get a model for dry period interarrival times as a by-product. We perform a statistical study and analyze various duration characteristics of dry and wet periods based on monthly Palmer drought severity indices for climate divisions in the United States. Data from the states of Nebraska, Nevada, Arizona, and Florida are used to show that previous models are often inadequate. We also estimate return periods showing that our proposed approach for modeling dry period interarrival times is necessary and that designs and actuarial calculations based on the results produced by fitting previously published models would be severely flawed.