论文部分内容阅读
本文通过未基于分区与基于分区的水稻总产遥感拟合模型的比较分析,选取最优模型进行水稻遥感估产.以湖南省为研究区,在水稻遥感估产分区、水稻可能种植区识别的基础上,以县为单位,利用2000~2007年的统计产量与MODIS EVI建立未基于分区与基于分区的水稻总产遥感拟合模型,通过相对误差、RMSE,以及拟合结果与统计值比较散点图分析,选择最优遥感拟合模型,并用此模型对2008年湖南省水稻总产进行预测.研究结果表明,基于分区的水稻总产遥感拟合模型要比未基于分区的要好,最优模型为二次非线性模型和逐步回归模型,且生育期主要集中在孕穗期到乳熟期.水稻总产拟合及预测结果与统计值相比省级相对误差都小于5%,且拟合结果的误差总体上比预测结果的误差要小.基于分区的水稻总产遥感估产模型有效地提高了水稻遥感估产的精度.
Based on the comparative analysis of non-zone-based and district-based total rice remote sensing fitting models, the optimal model was selected for remote sensing assessment of rice.With Hunan Province as the research area, based on the remote sensing estimation of rice yield and the identification of possible planting areas in rice , With county as a unit, the non-zone-based and zoning-based remote sensing fitting model was established based on the statistical output from 2000 to 2007 and MODIS EVI. The relative error, RMSE, and the fitting results were compared with the statistics to get the scatter plot The optimal remote sensing fitting model was selected and used to predict the total rice production in Hunan Province in 2008. The results show that the total remote sensing fitting model of rice based on zoning is better than the non-zoning-based model, and the optimal model is Quadratic nonlinear model and stepwise regression model, and the growth period mainly concentrated in the booting stage to the milk-ripening stage.Compared with the statistical results, the provincial-level relative fitting error of the total yield of rice was less than 5%, and the fitting result The error is generally smaller than the error of the prediction results.Zoning based total output of rice remote sensing yield model effectively improves the precision of rice remote sensing yield.