论文部分内容阅读
本文建立了23个综合考虑水稻单产的时间效应、空间效应和时空交互效应的时空模型。利用湖北省1991-2007年县级水稻单产数据,借助WinBUGS进行Gibbs抽样估计,根据DIC准则进行选择最优模型,预测1992-2009年各县市的水稻单产。基于2008年预测单产及其后验分布,厘定县级水稻产量保险的纯费率。研究的主要结论:1992-2007年的预测值与实际值比较接近,且预测的水稻单产的标准差都比较小,表明所选择模型具有良好的短期预测能力;相邻地域的纯费率比较接近;所厘定的纯费率与蒙特卡洛(Monte Carlo,MC)误差正相关,Pearson相关系数为0.6793,MC误差包含来自时间、空间以及两者相互作用带来的不确定性。
In this paper, we established 23 spatio-temporal models that comprehensively consider the temporal effects, spatial effects and spatio-temporal interactions of rice yield. Based on the data of county-level paddy rice production in 1991-2007 in Hubei Province, Gibbs sampling was estimated by WinBUGS, and the optimal model was selected according to DIC criteria to predict the rice yield in all counties and cities in 1992-2009. Based on the projected yield in 2008 and its posterior distribution, the pure rate of county-level rice yield insurance is determined. The main conclusions of the study are as follows: The predicted value of 1992-2007 is close to the actual value, and the standard deviation of the predicted rice yield is relatively small, indicating that the selected model has good short-term forecasting ability; the pure premium rates of the neighboring regions are relatively close ; The pure rate determined is positively related to the error of Monte Carlo (MC), the Pearson correlation coefficient is 0.6793, and the MC error includes the uncertainty brought by the time, space and the interaction between the two.