Modified analogue forecasting in the hidden Markov framework for meteorological droughts

Abstract

An analogue method (AM) is a nonparametric approach that has been applied to predict the future states of a dynamic system by following the evolution of the analogues in the historical archive. In this study, we proposed a hidden Markov model (HMM) framework for a modified analogue forecasting (MAF) approach for meteorological droughts in Korea. The unobservable (hidden) state process in the framework aims to model the underlying drought state, while the observation process was formed from the time series of the standardized precipitation index (SPI) as a drought index. Within the framework, the likelihood estimator was used as the measure of similarity between past SPI analogues and current data. The MAF approach was conducted on the selected analogues to make forecasts at lead times of one and three months. The proposed model was applied to five selected stations in Korea using the SPI data from 1973 to 2016. The forecasting performance of the proposed model was tested during the validation period (2003-2016) using several statistical criteria and it was compared to a persistence-based benchmark model. The results showed significant improvement in the forecasting capacity, and satisfactory performance for numerical SPI forecasting and categorical drought forecasting. The results also suggested that the proposed model was able to provide useful information for determining future drought categories for early drought warning with a lead time of up to three months.