不同施氮量下小麦遥感估产模型构建

Yield estimation model of wheat based on remote sensing data under different nitrogen supply conditions

为了构建不同施氮量下的小麦遥感估产模型,评估模型品种间的适用性,以不同施氮量处理的小麦(2010年1个品种,2011年3个品种)为试验材料,分别利用2011年单一品种和3个品种小麦冠层一阶微分参数、植被指数、施氮量和产量数据进行产量模型构建,共构建了6个模型,利用2010、2011年的相关数据进行模型检验。结果显示,建模和预测数据均来自同一品种时,其预测效果在6个模型中均较稳定且较好(R2=0.419);建模数据来自3个品种时,其预测效果在6个模型中也较好(R2=0.405);建模和预测效果检验来自不同品种时拟合和预测效果差异较大,表明所建模型品种间适用性较差,但是采用多品种数据建模可以提高估产效果;6个模型均没有氮素因子,可能是由于氮素对模型贡献不显著。

In order to build wheat yield estimation models under different nitrogen levels, and to evaluate the appli?cability of the models between varieties, canopy first?order differential parameters, vegetation index, nitrogen supply and yield data in 2011 from three varieties were used. Totally six yield estimation models were established. The data in 2010 were used for forecast test, and the data in 2011 were used for fitting test. When the modelling and forecast data were from the same variety, the forecast was stable ( R2=0?419) . while the modelling data was from three varieties, the forecast was good as well ( R2=0?405) . When the modelling and forecast data came from different varieties, big difference was presen?ted between fitting and forecast results, suggestive of the poor applicability of the model among varieties. Due to the insignif?icant contribution to the yield estimation model, nitrogen was not included in the six models.