Abstract: Taking the main highland barley growing areas in Tibet as study region, a comprehensive risk assessment model of drought disasters was established based on hazard index of disaster-causing factors, exposure index of disaster-bearing body, vulnerability index of disaster-bearing body and disaster prevention and reduction ability of disaster-bearing body. The pure insurance premium rate of highland barley was determined by non-parametric method, and the pure insurance rate was revised based on the results of risk regionalization, the modified GM (1,1) model and R/S method were combined to predict the future revised pure insurance premium rate. The results show that: (1) based on the risk index of drought hazard factor and vulnerability index of disaster-bearing body, risk grade shows a trend of high on both sides of the middle low, the risk grade based on the exposure index, the ability of disaster prevention and mitigation, and the comprehensive risk index of drought disaster has a tendency to increase gradually from east to west. (2) The pure insurance premium rate of highland barley insurance in each station has little difference (between 1.07%-9.79%), but the modified pure insurance premium rate has a big difference (between 1.86%-17.02%). (3) Based on the risk index, exposure index, and vulnerability index of drought, the spatial distribution of pure insurance premium rate shows the characteristics of high middle, low two sides, however, the pure insurance premium rate based on the drought disaster prevention and mitigation capability correction and the drought disaster comprehensive risk index shows the characteristics of middle high, local high and both sides low. This showed that the scientific and reasonable determination of pure insurance rates should consider the combined effects of multiple drought indices. (4) For the first time, the modified GM (1,1) model and R/S method were used to predict the future revised pure insurance rate of highland barley in Tibet (the rate of increase was 0.21 percent points·10y⁻¹). The comprehensive risk index can objectively reflect the drought risk level of highland barley in Tibet, and the result of regionalization can provide basic support for revising the pure insurance premium rate, and then to improve the precision of highland barley agricultural insurance services to provide a scientific basis.

Key words: MCI index, Comprehensive risk assessment model of drought disaster, Nonparametric, GM（1,1）model, R/S method