Estimating N status of winter wheat using a handheld spectrometer in the North China Plain |
| |
| Authors: | Fei Li Martin L Gnyp Liangliang Jia Yuxin Miao Zihui Yu Wolfgang Koppe Georg Bareth Xinping Chen Fusuo Zhang |
| |
| Institution: | 1. College of Resources & Environmental Sciences, China Agricultural University, 100094 Beijing, China;2. College of Ecology & Environmental Science, Inner Mongolia Agricultural University, 010019 Hohhot, China;3. Institute of Geography, University of Cologne, 50923 Köln, Germany;4. Institute of Agriculture Resource & Environment, Hebei Academy of Agricultural and Forestry Sciences, 050051 Shijiazhuang, China |
| |
| Abstract: | Excessive nitrogen (N) fertilizer application is very common in the North China Plain. Diagnosis of in-season N status in crops is critical for precision N management in this area. Remote sensing, as a timely and nondestructive tool, could be an alternative to traditional plant testing for diagnosing crop N status. The objectives of this study were to determine which vegetation indices could be used to estimate N status in winter wheat (Triticum aestivum L.) under high N input conditions, develop models to predict winter wheat N uptake using spectral vegetation indices and validate the models with data from farmers’ fields. An N rate experiment and a variety-N experiment were conducted in Huimin, Shandong Province from 2005/2006 to 2006/2007 to develop the models. Positive linear relationships between simple ratio vegetation indices (red vegetation index, RVI and green vegetation index, GVI) and N uptake were observed independent of growth stages and varieties (R2, 0.48–0.74). In contrast, the relationships between normalized difference vegetation indices (NDVI and GNDVI), red and green normalized difference vegetation index (RGNDI), and red and green ratio vegetation index (RGVI) were exponentially related to N uptake (R2, 0.43–0.79). Subsequently, 69 farmers’ fields in four different villages were selected as datasets to validate the developed models. The results indicated that the prediction using RVI had the highest coefficient of determination (R2, 0.60), the lowest root mean square error (RMSE, 39.7 kg N ha−1) and relative error (RE, 30.5%) across different years, varieties and growth stages. We conclude that RVI can be used to estimate nitrogen status for winter wheat in over-fertilized farmers’ fields before heading. |
| |
| Keywords: | NDVI red normalized difference vegetation index GNDVI green normalized difference vegetation index RGNDI red and green normalized difference vegetation index RVI red ratio vegetation index GVI green ratio vegetation index RGVI red and green ratio vegetation index Opt optimum N rate Con conventional N rate RMSE root mean square error RE relative error |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
