关键词: 玉米, 天气指数保险, 干旱指数保险, 农业保险, CERES-Maize作物模型, 概率分布模型

Abstract: The policy-based agricultural insurance of maize in Shaanxi province was impeded due to its moral hazard, adverse selection and high management cost. Weather index agricultural insurance takes specific meteorological index as trigger which can avoid the defects of the traditional agricultural insurance, and is one of the effective solutions to the current predicament of agricultural insurance. Changwu county located in Shaanxi, an important maize production region, drought stress in the water critical period is the main limiting factor that inhibits maize growth and yield. Therefore, constructing drought stress model for maize in its water critical period and studying the drought index insurance are of great significance for designing the weather index insurance and solving the dilemma of current traditional agricultural insurance. To isolate the influence of a single meteorological factor at a specific crop growth stage, a field experiment of rain-fed maize was conducted in Dryland Agriculture Experiment Station of Northwest A&F University in Changwu from 2011 to 2013.The field experimental data of weather variables, soil, management practices and maize growth and development, were used to calibrate and validate CERES-Maize model. The accumulated precipitation from June 21 to August 31 in Changwu was defined as the drought index (DI) of maize during the water critical period. The DI in 2013 was treated with ±20, ±40, ±60, ±80, ±100, ±120, ±140, ±160, ±180, ±200, ±220 and ±240mm and then distributed daily to simulate the maize yield using CERES-Maize model. The water treatment corresponding to the maximum simulated yield was set as the critical point, and the water treatment less than the critical point was set as the drought treatment. The drought stress model was constructed based on the data of drought treatment and the corresponding simulated yield, in combination with the disaster grade of yield reduction rate, the drought levels and the corresponding drought index thresholds were determined. The optimal distribution model of drought index was selected through EasyFit software using the meteorological data of Changwu from 1990 to 2019, and the occurrence probability of different drought levels in the water critical period of maize were estimated by the selected model. The drought index insurance rate of maize in water critical period was determined by ratemaking method based on the occurrence probability of each drought grade and the corresponding yield reduction rate. The compensation scheme of drought index insurance was designed using the projection pursuit regression method. The results showed that the average absolute relative error (ARE) and relative root mean square error (RRMSE) of CERES-Maize simulation were less than 10%, which was in line with the requirements of crop model simulation accuracy. A linear relationship was showed between maize drought index (DI) during the water critical period and the simulated yield loss（y, %）, that was y=-0.55DI+107.17. The Log-logistic model performed best for the drought index distribution, and the Anderson-Darling (AD) test value was only 0.20. The occurrence probability of light, moderate, severe and excessive drought was 9.75%, 5.90%, 3.71% and 3.50%, respectively. The premium rate of drought index insurance of maize in water critical period was 5.6%. The compensation will start when DI is ≤185 and be graded as the maize under drought stress.

Key words: Maize, Weather index insurance, Drought index insurance, Agricultural insurance, CERES-Maize crop model, Probability distribution model