基于叶片尺度的东北粳稻产量估测

摘要

及时准确地估测水稻产量是服务现代农业的重要内容,对制定科学的粮食政策具有重要的现实意义.本研究以东北粳稻为例,利用试验区粳稻叶片植被指数归一化差值植被指数(NDVI)和光化学植被指数(PRI)估测粳稻产量.基于2015年粳稻生长关键期6-9月的叶片NDVI和PRI,结合试验小区产量数据,建立了基于试验区叶片NDVI和PRI的粳稻产量估算模型.单月NDVI与产量一元线性模型的R2范围为0.455～0.581,平均估产精度为96.36％.单月PRI与产量一元线性模型的R2范围为0.396～ 0.709,平均估产精度为96.68％.单月NDVI和PRI复合估产二元线性模型的R2范围为0.655 ～0.784,平均估产精度为97.26％.利用不同月份组合的NDVI累积和与PRI累积和建立的粳稻产量模型R2范围为0.765～0.949,估产精度均在97.48％以上.所建参数模型中拟合效果最好的是6月、8月、9月NDVI累积和与PRI累积和复合的估产模型,R2为0.949,估产精度高达98.82％,此模型可作为粳稻估产的一种参考模型.%Estimating rice yield timely and accurately is an important part of modem agriculture research and is also significant in food policy formulation.The study used normalized difference vegetation index and photochemical reflectance index of japonica rice leaf in test area to estimate rice yield.Yield estimation models were made,which based on NDVI and PRI data from June to September,2015.The range of the correlation coefficients (R2) of one-variable linear model between yield and NDVI,PRI was 0.455-0.581 and 0.369-0.709.The estimation accuracy monthly was 96.36％ and 96.68％,respectively.The range of the correlation coefficients (R2) of two-variable linear model between yield and NDVI,PRI was 0.655-0.784 and the estimation accuracy was 97.26％.The correlation coefficients of yield model between accumulated NDVI and PRI of different months and japonica rice yield arranged from 0.765 to 0.949 and the accuracy was all above 97.48％.The best model was the composite yield estimation model using the NDVI and PRIcumulative sum(CUSUM) of three months,June,August and September,its R2was 0.949 and the estimation accuracy was 98.82％,which could be used as a reference model of japonica rice yield estimation.