Hou Dawei, Chen Jing, Dong Jinwei, Ji Chao, Feng Jingxuan, Du Guoming, Yang Lixiao
School of Public Administration & Law, Northeast Agricultural University, Haerbin, 150030, China.
The Center of land remote-sensing engineering technology for Heilongjiang Province, Northeast Agricultural University, Haerbin, 150030, China.
Sci Data. 2025 Aug 5;12(1):1355. doi: 10.1038/s41597-025-05715-0.
As the cornerstone of China's food security, Northeastern China contributes nearly 20% of national rice production. However, we are still lacking of high-resolution rice maps with detailed and long time-series in this region, impeding crop management decisions for food security. Here we generated an annual 30 m resolution rice distribution dataset for Northeastern China since the 21st century (NECAR) using the Google Earth Engine platform and random forest classification. The workflow involved (1) hierarchical screening principle to select ground samples, (2) the linear interpolation and Whittaker smoothing Landsat5/7/8 time series data and (3) enhanced spectral-feature sets. The resultant annual maps have high overall accuracy (OA) ranging from 0.93 to 0.99, and the satellite estimates corresponded well with statistics for most cities (R ≥ 0.7, p < 0.01), with higher accuracy than that of similar crops mapping datasets. This is the first attempt in Northeastern China to reconstruct paddy rice patterns at a 30-m resolution over a detailed and extended time series, enabling in-depth analysis of potential environmental and economic impacts.
作为中国粮食安全的基石，中国东北地区的水稻产量占全国近20%。然而，该地区仍缺乏具有详细和长时间序列的高分辨率水稻分布图，这阻碍了为保障粮食安全而进行的作物管理决策。在此，我们利用谷歌地球引擎平台和随机森林分类法，生成了自21世纪以来中国东北地区的年度30米分辨率水稻分布数据集（NECAR）。工作流程包括：（1）采用分层筛选原则选择地面样本；（2）对Landsat5/7/8时间序列数据进行线性插值和维特克平滑处理；（3）增强光谱特征集。生成的年度地图总体精度较高（OA为0.93至0.99），卫星估算值与大多数城市的统计数据吻合良好（R≥0.7，p<0.01），精度高于类似作物测绘数据集。这是中国东北地区首次尝试在详细且扩展的时间序列上重建30米分辨率的水稻种植模式，从而能够深入分析潜在的环境和经济影响。
