A Difference Wavelet Feature Index for Estimating Aerial N Uptake of Winter Wheat from In Situ Hyperspectral Remote Sensing

Abstract

The real-time and accurate assessment of crop aerial nitrogen (N) uptake is of significant importance for optimizing N fertilization. To develop a robust method for determining aerial N uptake in winter wheat, a field experiment with different N fertilizer levels was conducted over three successive years at two ecological sites in Henan, China. This research systematically compared the correlation between aerial N uptake and spectral parameters derived from various spectral transform methods: continuum removal (CR), standard normal variate transform method, first derivative reflectance (FDR), and continuous wavelet transforms (CWT). The findings revealed that CWT exhibited the highest efficacy among all the spectral transform methods, followed by FDR, with R² values of 0.777 for WF(4,770) and 0.764 for FDR₇₄₈. A new index, termed the difference wavelet feature index (DWF), is defined as DWF(4 560 770) = WF(4560) − WF(4770). This simple yet effective index significantly enhances the assessment of aerial N uptake, achieving an R² of 0.815. Validation with independent data showed that the RMSE for the DIDA, FDR₇₄₈, WF(4770), and DWF(4 560 770) under different cultivation factors were 3.578–4.361 g m^-2, 3.501–4.219 g m^-2, 3.472–4.309 g m^-2, 3.262–4.030 g m^-2, respectively. It was further verified that the newly DWF(4 560 770) index has excellent universality and stability. Therefore, the aforementioned studies indicated that the novel DWF(4 560 770) is more suitable for evaluating aerial N uptake at the heterogeneous field scale and also has significant potential for precise prediction of aerial N uptake using UAV remote sensing.