National-scale 10 M Annual Maize Maps for China and the Contiguous United States Using a Robust Index from Sentinel-2 Time Series
Yingze Huang,Bingwen Qiu,Peng Yang,Wenbin Wu,Xuehong Chen,Xiaolin Zhu,Shuai Xu,Laigang Wang,Zhanjie Dong,Jianyang Zhang,Joe Berry,Zhenghong Tang,Jieyang Tan,Dingding Duan,Yufeng Peng,Duoduo Lin,Feifei Cheng,Juanzhu Liang,Hongyu Huang,Chongcheng Chen
DOI: https://doi.org/10.1016/j.compag.2024.109018
IF: 8.3
2024-01-01
Computers and Electronics in Agriculture
Abstract:Accurate and continuous maps of maize distribution are essential for food security and sustainable agricultural development. However, there are no continuous national-scale and fine-resolution maize maps and explicit updated information on the spatiotemporal dynamics of maize for most countries. Maize mapping at the national scale is challenging due to the spectral heterogeneity caused by crop growth conditions, cropping patterns, and inter-annual variations. To this end, this study developed a novel crop index-based algorithm for national-scale maize mapping. Compared to other crops, maize is characterized by large-leaf-dominated canopies and high photosynthetic efficiency. Maize shows significant changes in chlorophyll and anthocyanin content. Therefore, a robust maize index was established by exploring the temporal Variation of the Vegetation-Pigment index (VVP) during the growing period. A simple decision rule was coded on the Google Earth Engine (GEE) platform, which was used for maize mapping based on the Sentinel-2 time series in China and the contiguous United States (US) from 2018 to 2022. The national-scale 10 m annual maize maps for China and the contiguous US were developed and in good agreement with the corresponding agricultural statistics data for many years (R2 > 0.94) and 9,412 reference points (overall accuracy of 90.09 %). Compared with simply applying the vegetation index, the VVP index took account of spectral heterogeneity caused by variations in crop growth conditions, cropping patterns, and inter-annual, and the omission error of maize was reduced by over 20 %. Moreover, the VVP index can significantly improve the spatial transferability of the Random Forest (RF) classifier. The first 10 m annual maize maps for China revealed that the planted area trend decreased and then increased from 2018 to 2022. The year 2020 was the turning point. The maize planted area consisted of 68 % single maize and 32 % double cropping with maize in 2020, with the northern boundary for double cropping with maize in the Yanshan Mountains. The maize planted area in China consistently decreased by 39,352 km2 (about 9 %) from 2018 to 2020. This is mainly due to the adjustment of the maize-planted structure in the "Sickle Bend" region of China (the "Sickle Bend" policy). However, the maize planted area gradually recovered from 2020 to 2022, primarily concentrated in regions with ever-planted. This study will provide essential information for cropping structure adjustment and related agricultural policy formulation and contribute to sustainable agricultural development by mapping maize from a national to a global scale.