A control chart for severity index to detect drought compares favourably to logistic regression

摘要

We describe the discovery of how a traditional control chart for the Palmer Drought SeveritynIndex (PDSI) to detect drought compares favourably to a theoretically appropriate statisticaln(logistic regression) model of drought as a function of PDSI. Our empirical results are based onnmonthly observations of PDSI, precipitation and temperature made in Kansas since 1895. Resultsnfrom the study suggest that a relatively simple statistical approach based on Shewhart controlncharts may provide a more accessible method for relevant government agencies to predictndroughts, improving resource management and preparation. Moreover, utilizing such an approachnover more sophisticated methods may come at little expense regarding prediction errors.