Remotely assessing leaf N uptake in winter wheat based on canopy hyperspectral red-edge absorption

摘要

Remote sensing is a rapid, non-destructive method for assessing crop nitrogen (N) status. In this research, we investigated the quantitative relationship between leaf N uptake and ground-based canopy hyperspectral reflectance in winter wheat (Triticum aestivum L.). We conducted field experiments over four years at different sites (Xinyang, Zhengzhou and Shangshui) in Henan, China using different N application rates, growth stages and wheat cultivars and developed a novel spectral index with improved predictive capacity for leaf N uptake estimation. Sixteen vegetation indices in the publications were examined for their reliability in monitoring leaf N uptake in winter wheat. Linear regression was integrated with optimized common indices DIDA and SDr/SDb to investigate the dynamic nature of leaf N uptake, which resulted in coefficients of determination (R-2) of 0.816 and 0.807 and root mean square error (RMSE) of 1.707 and 1.767, respectively. Our novel area index, designated shifting red-edge absorption area (sREA), was constructed according to analysis of the red-edge characteristics and area-based algorithm with the formula:sREA = 1/2 x (R680+Delta lambda - R-680) x Delta lambda, Delta lambda= 320xD(725)+140xD(756)-140xD(680)/7xD(700)+4xD(725). This index is highly correlated with leaf N uptake (highest R-2 = 0.831; lowest RMSE = 1.556). On the whole, calculation of R2 and RMSE confirmed that sREA prediction models were better than optimized common indices for 16 out of 17 datasets across growing seasons, sites, N rates, cultivars and stages. Fitting independent data to the equations resulted in RE values of 19.6%, 18.8%, 17.6% and 16.2% between measured and estimated leaf N uptake values for RSI(D-740, D-522), SDr/SDb, DIDA and sREA, respectively, further confirming the superior test performance of sREA. These models can therefore be used to accurately predict leaf N uptake in winter wheat. The novel index sREA is superior for evaluating leaf N status on a regional scale in heterogeneous fields under variable climatic conditions. (C) 2016 Elsevier B.V. All rights reserved.