College of Geography and Environmental Science, Henan University, Kaifeng, 475004, People's Republic of China.
National Ecosystem Research Network of China, Henan Dabieshan National Field Observation & Research Station of Forest Ecosystems, Xinyang, 464000, People's Republic of China.
Environ Sci Pollut Res Int. 2022 Feb;29(9):12661-12679. doi: 10.1007/s11356-021-15387-z. Epub 2021 Sep 23.
As a key input variable to many global climates, land surfaces and crop models, cropping intensity (CI) accurately assesses and predicts crops' output, in view of the global decline in food production in recent years due to declining natural resources, urban expansion and declining quality of arable land. Hence, research on CI mapping can have a contribution to solve this problem. Unfortunately, existing remote sensing data for CI mapping research, including Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat images, are not adequate for obtaining CI information at higher spatial and temporal resolution. In this regard, we develop an algorithm to extract CI based on per-pixel physiognomy. To be specific, the algorithm is based on the Google Earth Engine (GEE) platform and constructs a high temporal (10 days) spatial (30 m) resolution dataset with the fusion of Landsat 7/8 and Sentinel-2 A/B image data and extracts CI information using a time series of peak discovery method, threshold method and phenological period feature extraction to obtain the 2018 Chinese Huai River Basin (HRB) CI map. Our results suggest that the overall accuracy (OA) of CI extraction in the HRB is 92.72%, with a kappa coefficient of 0.864. The single-season crop, double-season crop and three-season crop account for 41.6%, 57.7% and 0.7% of the total farmland area, respectively. Compared to existing CI identification and extraction methods, this approach achieves higher accuracy in the identification and extraction of CI information over a larger area.
作为许多全球气候、陆地表面和作物模型的关键输入变量，种植强度 (CI) 可以准确评估和预测作物的产量，鉴于近年来由于自然资源减少、城市扩张和耕地质量下降，全球粮食产量下降。因此，对 CI 制图的研究可以有助于解决这个问题。不幸的是，现有的用于 CI 制图研究的遥感数据，包括中分辨率成像光谱仪 (MODIS) 和陆地卫星图像，不足以获得更高时空分辨率的 CI 信息。在这方面，我们开发了一种基于逐像素地貌的 CI 提取算法。具体来说，该算法基于 Google Earth Engine (GEE) 平台，利用 Landsat 7/8 和 Sentinel-2 A/B 图像数据的融合构建了一个高时间（10 天）空间（30 m）分辨率的数据集，并使用峰值发现方法、阈值方法和物候期特征提取的时间序列来提取 CI 信息，以获得 2018 年中国淮河流域（HRB）的 CI 图。我们的结果表明，在 HRB 中提取 CI 的总体精度（OA）为 92.72%，kappa 系数为 0.864。单季作物、双季作物和三季作物分别占农田总面积的 41.6%、57.7%和 0.7%。与现有的 CI 识别和提取方法相比，该方法在更大的区域内实现了更高的 CI 信息识别和提取精度。
