A MARKOV AND RUNS ANALYSIS OF DROUGHT IN THE CENTRAL UNITED STATES

Abstract

The most common measure of drought, the Palmer Drought Severity Index (PDSI), is combined with the theory of runs in a Markov chain analysis. Markov-derived transition and equilibrium probability estimates can be used to derive expected number of runs (droughts) and mean run length (drought duration). After a review of the theory of the Markov chain and runs, the techniques are applied to the PDSI from five states in the central United States. The results show very different spatial patterns given the degree of drought severity. Extreme droughts, as defined by the PDSI, occur more frequently, are more persistent and last longer in southern Nebraska and north central Kansas. By contrast, severe droughts are more frequent and persistent in southern Kansas and north-central Oklahoma. In general, climate divisions with high persistence for extreme drought tend to have low persistence for severe drought and vice versa. Overall, the central areas of Nebraska and Kansas tend to have the greatest drought hazard, while the hazard tends to decrease to the south and east. [Key words: climatology, drought, Markov chain, central United States.]