A nitrogen spectral response model and nitrogen estimation of summer maize during the entire growth period. Issue 5 (3rd March 2020)

• Record Type: Journal Article
• Authors: Xu, Xiaobin; Zhu, Hongchun; Li, Zhenhai; Wang, Jianwen; Yang, Guijun
• Is Part Of: International journal of remote sensing. Volume 41:Issue 5(2020)
• Page Start: 1867
• Page End: 1883
• Publication Date: 2020-03-03
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2019.1677967
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Timely and effective prediction of nitrogen content in summer maize could provide support data for precise fertilization. In this study, the feasibility and expansibility of predicting the nitrogen mechanism model of summer maize leaves through its entire growth period were investigated on the basis of the theory of leaf radiation transmission mechanism. A complete random test of data from two maize varieties and two nitrogen fertilizer applications in 2017 was conducted. Three versions of the leaf optical PROperties SPECTra (PROSPECT) model, namely, PROSPECT-4, PROSPECT-5, and PROSPECT-D were used to link the established leaf nitrogen density (LND) and chlorophyll- a + b (chl- a + b ) models, that is, chl- a + b -LND model. A nitrogen response transfer model (N-RTM) was established by linking the optimal PROSPECT and chl- a + b -LND models. Results were as follows. (1) chl- a + b estimation using the PROSPECT-D model yielded the highest accuracy (the coefficient of determination ( R 2 ) = 0.774, the normalized root mean squared error (nRMSE) = 13.19%) among the three PROSPECT models, it shows that the model considering more factors can better reflect the internal law of blade, and could be used as the basic model of N-RTM; (2) Established chl- a + b -LND models based on the dataset from each growth stage showed differences using the confidence interval method, and the R 2 values of the optimal regression model at V12, VT, and R3 were 0.794, 0.781, and 0.821, respectively. Based on the changes of chl- a + b and LND during the growth period, a piecewise model was constructed; (3) The R 2 and nRMSE values between the measured and estimated LNDs were 0.656% and 22.86%, respectively. The validation results are better than the traditional empirical model. The results showed that the segmented model, which considered the interaction of various factors within the leaves and the change of chl- a + b -LND during the growth period, had better performance in nitrogen monitoring. The constructed nitrogen model in this study preliminarily realized the remote sensing prediction of the nitrogen mechanism model and had a certain mechanism.