基于主动遥感的冬小麦群体动态监测

摘要

茎蘖数的多少是冬小麦达到最佳产量的关键因素,也是氮肥调控的重要指标.然而为了获得一个代表性的茎蘖数,传统的人工数分蘖的方法费时费力.近年来,随着遥感,特别是主动遥感技术的长足发展,为解决这个问题提供了机遇.文章通过2008年-2009年在河北曲周县中国农业大学试验站的播期、播量和氮水平田间试验,采用主动光谱仪GreenSeeker对冬小麦群体动态进行实时监测,并建立模型来预测小麦茎蘖数.研究表明,以归一化植被指数(NDVI)和简单比值植被指数(RVI)这两个植被指数建立模型都能成功预测小麦茎蘖数,两个植被指数与茎蘖数有着显著的线性关系.NDVI与茎蘖数的决定系数(R2)介于0.25～0.64之间,RVI与茎蘖数的决定系数(R2)介于0.26～0.65之间.验证结果进一步证实了NDVI能更好地在生育前期预测冬小麦的茎蘖数,有高的决定系数(R2,0.54～0.64)、低的均方根误差(RMSE,260～350茎蘖数·m-2)和低的相对误差(RE,16.3%～23.0%).这些结果表明,主动光谱仪是一个很好的工具可以用来监测冬小麦的群体动态.%Tiller density plays an important role in attaining optimum grain yield and applying topdressing N in winter wheat.However, the traditional approach based on determining tiller density is time-consuming and labor-intensive. As technology advances, remote sensing might provide an opportunity in eliminating this7 problem. In the present paper, an N rate experiment and a variety-seeding and sowing dates experiment were conducted in Quzhou County, Hebei Province in 2008/2009 to develop the models to predict the amount of winter wheat tillers. Positive linear relationships between vegetation indices and tillers were observed across growth stages (R2, O. 25～0. 64 for NDVI; 0. 26～0. 65 for RVI). The validation results indicated that the prediction using NDVI had the higher coefficient of determination (R2, 0. 54～0. 64), the lower root mean square error (RMSE,260～350 tillers m-2) and relative error (RE, 16. 3％～23.0％) at early growth stages of winter wheat. We conclude that active GreenSeeker sensor is a promising tool for timely monitoring of winter wheat tiller density.