利用投影寻踪回归技术进行草地产量预报的研究

本文利用1991～1992年在新疆阜康县不同类型草地上观测的牧草产量、环境因子和卫星遥感资料运用投影寻踪回归(PPR)技术,探讨了草地产量预报的原理与方法,建立了产量预报综合模型,其预报精度达到83.5%以上,并克服传统预报方法的一些不足,科学地提示其内在关系,进而对影响产量预报精度和产量形成的因素进行分析与实验验证。因而研究结果表明利用PPR技术基本实现了科学预报草地产量的目的。

A Study on the Grassland Forage Yield Forecasting Using Projection Pursuit Regression (PPR) Technology

The principles and methods of forecasting grassland forage yields were discussed and the comprehensive models of forecasting forage yields were established by the PPR technology using the actual forage yields, environment and satellite remote sensing data observed in 1991 - 1992 in different grassland types in Fukang county,Xinjiang. The forecasting forage yield accuracy was 83. 5%- The shortages of the routine forecasting forage yield methods were overcome and the correlation between forage yields and indices were revealed. Further, the factors of effecting on forecasting forage yield accruacy and formation were analysed and checked by the experiment. The objectives of scientific forecasting forage yields in different grassland types were realized by the PPR technology.