马尾松毛虫发生量空间定点预报模型的研究

利用1983年9月至1989年9月马尾松毛虫的虫情调查资料，研究仙居县各测报点的马尾松毛虫有虫面积、虫口密度的时空变化规律，分别建立三维自回归趋势面模型、预测克立格模型、空间-时间序列预测之自回归模型和广义时空回归模型4种发生量空间定点预报模型。结果表明，所建立的4种发生量空间定点预报模型均具有较好的预报效果，对1989年3代的虫口密度、有虫面积进行预报检验，预测克立格模型的预报准确率为100％；空间-时间序列预测之自回归模型和广义时空回归模型对两者的预报准确率则均为66.7％；三维自回归趋势面模型虫口密度预报准确率为33.3％，有虫面积的预报准确率为66.7％。

Study on Space Fixed-point Prediction Model of Dendrolimus Punctatus Walker's Occurrence Quantity

Using the insect situation investigation data from September,1983 to September,1989,the study researched the space time variation law of Dendrolimus punctatus Walker's insect appearance area and density at each forecast sites in Xianju country,and set up 4 kinds of space fixed-point prediction models of occurrence quantity including tri-dimensional auto-regression trend surface model,predicted kriging model,space-time series prediction of autoregressive model and generalized space regression model.The results showed that 4 kinds of space fixed-point prediction models had a good predication effects on forecast verification of insect density and appearance area of 3-generation in 1989,the forecasting accuracy rate of Kriging prediction model was 100%;while the forecasting accuracy rate of space-time series prediction of autoregressive and generalized space regression model were 66.7% respectively;the insect density's forecasting accuracy rate of tri-dimensional auto-regression trend surface model was 33.3%,and the forecasting accuracy rate of insect appearance area was 66.7%