氮素和小麦条锈病胁迫下小麦高光谱遥感估产模型研究

Study of Wheat Hyperspectral Remote Sensing Yield Estimation Model Under Stress of Nitrogen and Wheat Stripe Rust

为了构建不同施氮水平和条锈病发病条件下的小麦估产模型,设置了不同氮素水平及人工接种小麦条锈病,通过采用将多个关键生育期的光谱植被指数、一阶微分参数与小麦条锈病病情指数、叶片含氮量、产量构成因子、产量进行相关分析,植被指数、一阶微分参数与产量进行回归分析的方法,研究与产量相关性高的植被指数、微分参数,结果表明分别利用绿光红光比值植被指数(GR)和绿光波段一阶微分值总和(SDg)、蓝光波段一阶微分值总和(SDb)在灌浆期构建的估产模型预测效果较好,2010年预测准确率分别可以达到99.87%、99.98%,2011年预测准确率分别可以达到97.9%和95%。通过试验研究发现高光谱遥感技术在氮素和小麦条锈病双重胁迫下也可以较好的预测产量,这对研究多重胁迫、多种栽培措施下的小麦估产模型有重要意义。

Remote sensing （RS） is a good tool to estimate wheat yield. In order to get wheat yield estimation model under different nitrogen and wheat stripe rust levels, this paper analyzed the correlation relationship between vegetation index, the first derivative and the wheat leaf nitrogen content, wheat stripe rust disease index（D/）, yield components and wheat yield respectively, using regression statistics method to choose high correlation factors to estimate wheat yield. The results showed that the yield estimation model in filling stage used respectively the vegetation index of green red ratio （GR） and the sum of green band first derivative, the sum of blue band first derivative had a good estimation result. Yield models estimation accuracy in 2010 could reach 99.87% and 99.98%, in 2011 could reach 97.9% and 95%. This research found that the hyperspectral remote sensing technology had a good yield estimation result under the stress of nitrogen and wheat stripe rust, the research result discussed in this paper was of great significance to study wheat yield estimation models under more stress and cultural practices.