Application of the Hidden Markov Model to Agricultural Drought Disaster Risk Assessment

Liyuan Zhang,Wentong Yang,Chunlei Liang

DOI: https://doi.org/10.2139/ssrn.3975980

2021-01-01

SSRN Electronic Journal

Abstract:Drought disaster is one of the meteorological disasters that hinders agricultural development, and risk assessment of this disaster can reduce losses. Although there are many kinds of risk assessment methods, most of them are static. For this reason, a risk assessment model combining static assessment and dynamic assessment is proposed. In this model, firstly, the indicator system method and the method of Comprehensive Percentage of Production Loss (CPPL) are used to construct a static risk assessment model, which can get the year-by-year drought disaster information. Secondly, based on the static risk assessment results, the Hidden Markov Model (HMM) is used to achieve dynamic risk assessment. Finally, the applicability of the risk assessment model is validated by using data from the three provinces in Northeast China. The results show that since the 21st century, the meteorological conditions in Heilongjiang Province were gradually suitable for the growth of crops. Although the meteorological risk in Liaoning Province had been high in recent years, the drought disaster in the area tended to be slight. The agricultural meteorological situation and drought disaster situation in Liaoning Province hadn’t been optimistic. In the next three years, agriculture in the Northeast is more likely to suffer from a slight drought disaster, and the severe disaster is more likely to be caused by moderate-risk meteorology. The HMM can be applied to dynamic drought disaster risk assessment, and the HMM combined with the static risk assessment results can guide agricultural drought disaster risk prevention work.