Vectorized dataset of check dams on the Chinese Loess Plateau using object-based classification method from Google Earth images

作者：Yi Zeng, Tongge Jing, Baodong Xu, Xiankun Yang, Jinshi Jian, Renjie Zong, Bing Wang, Wei Dai, Lei Deng, Nufang Fang, Zhihua Shi · 年份：2023 · DOI：10.5194/essd-2023-120 · 被引用次数：2 · 研究领域：Soil erosion and sediment transport、Hydrology and Sediment Transport Processes、Hydrology and Watershed Management Studies

Abstract. The Chinese government has invested tens of billions of dollars and about 60 years to implement a large-scale check dam project on the Chinese Loess Plateau (CLP) to control severe soil erosion. These check dams have trapped billions of tons of eroded sediment over the past few decades, significantly reducing the sediment load of the Yellow River, which was once the river with the largest sediment load in the world. However, there is still great uncertainty about how much sediment is trapped by check dams and what roles they play in the flow and sediment variability in the Yellow River, because the number and spatial distribution of check dams are still unclear. In this study, we produced the first vectorized dataset of check dam on the CLP, combining high-resolution and easily accessible Google Earth images with object-based classification methods. We first investigated and analysed the key characteristics of check dams, and obtained the 0.3–1.0 m resolution Google Earth image of the best extraction period. Then we preliminarily obtained the rough check dam layer through multi-scale segmentation, threshold classification, and river network superposition. Finally, a self-developed human-computer interaction program combined with auxiliary data, visual interpretation, and expert knowledge is used to improve the classification accuracy of check dams. The accuracy of the dataset is verified by 1947 collected test samples, and the producer’s accuracy and user’s accuracy...