Probabilistic assessment of meteorological drought over South Korea under RCP scenarios using a hidden Markov model

Most drought indices are evaluated based on pre-defined thresholds, which are inadequate for demonstrating the inherent uncertainty of drought. This study employed a hidden Markov model-based drought index (HMM-DI) for probabilistic assessment of meteorological drought in South Korea. The HMM-DI was developed to take into account the inherent uncertainty embedded in daily precipitation and to assess drought severity without using pre-defined thresholds. Daily rainfall data recorded during 1973–2015 at 56 stations over South Korea were aggregated with 6- and 12-month windows to develop HMM-DIs for various time scales. The HMM-DIs were extended to assess future droughts in South Korea using synthesized monthly rainfall data (2016–2100) under Representative Concentration Pathway (RCP) 4.5 and 8.5 scenarios. The overall results indicated that the HMM-DI can classify drought conditions considering inherent uncertainty embedded in observations and can also demonstrate the probabilistic drought occurrence in the future.