Prediction of Wheat Grain Protein by Coupling Multisource Remote Sensing Imagery and ECMWF Data

• Main Authors: Xiaobin Xu, Cong Teng, Yu Zhao, Ying Du, Chunqi Zhao, Guijun Yang, Xiuliang Jin, Xiaoyu Song, Xiaohe Gu, Raffaele Casa, Liping Chen, Zhenhai Li
• Format: Article
• Language: English
• Published: MDPI AG 2020-04-01
• Series: Remote Sensing
• Subjects: Multi-source Remote Sensing Imagery; European Center for Medium-range Weather Forecasts (ECMWF) meteorological data; grain protein content; hierarchical linear model
• Online Access: https://www.mdpi.com/2072-4292/12/8/1349

Industrialization production with high quality and effect on winter is an important measure for accelerating the shift from increasing agricultural production to improving quality in terms of grain protein content (GPC). Remote sensing technology achieved the GPC prediction. However, large deviations in interannual expansion and regional transfer still exist. The present experiment was carried out in wheat producing areas of Beijing (BJ), Renqiu (RQ), Quzhou, and Jinzhou in Hebei Province. First, the spectral consistency of Landsat 8 Operational Land Imager (LS8) and RapidEye (RE) was compared with Sentinel-2 (S2) satellites at the same ground point in the same period. The GPC prediction model was constructed by coupling the vegetation index with the meteorological data obtained by the European Center for Medium-range Weather Forecasts using hierarchical linear model (HLM) method. The prediction and spatial expansion of regional GPC were validated. Results were as follows: (1) Spectral information calculated from S2 imagery were highly consistent with LS8 (R² = 1.00) and RE (R² = 0.99) imagery, which could be jointly used for GPC modeling. (2) The predicted GPC by using the HLM method (R² = 0.524) demonstrated higher accuracy than the empirical linear model (R² = 0.286) and showed higher improvements across inter-annual and regional scales. (3) The GPC prediction results of the verification samples in RQ, BJ, Xiaotangshan (XTS) in 2018, and XTS in 2019 were ideal with root mean square errors of 0.61%, 1.13%, 0.91%, and 0.38%, and relative root mean square error of 4.11%, 6.83%, 6.41%, and 2.58%, respectively. This study has great application potential for regional and inter-annual quality prediction.