Spatiotemporal Dynamic Changes of Cropland in Rich and Coarse Sediment Areas of Middle Yellow River Basin Based on Multi-source Data Fusion Products

[Objective] The rich and coarse sediment areas of middle Yellow River basin is a typical ecologically fragile area in China. Accurately quantifying the spatiotemporal dynamic changes of cropland in this region is crucial for evaluating regional food security, ecological restoration benefits, soil erosion conditions, and the downstream tranquility of the Yellow River. [Methods] Based on five high-resolution (30 m) land use/cover dataset products (CNLUCC, GLC_FCS30, CLCD, AGLC-2000-2015, GlobeLand30), this study employed methods such as cropland dynamic degree, deviation coefficient, transfer matrix, and spatial consistency analysis to comparatively analyze the spatiotemporal characteristics of cropland in the rich and coarse sediment areas. A high-precision fused dataset with a 30 m resolution was formed and validated. Based on this fused product, an analysis of the quantity, distribution, and structural characteristics of cropland in rich and coarse sediment areas of middle Yellow River basin in 1990—2020 years was conducted. [Results] There are significant differences in cropland characteristics among existing multi-source dataset products, with the fused products demonstrating higher accuracy compared to using any of the five existing products individually. Analysis based on the fused products reveals a trend of initially increasing followed by fluctuating decline in cropland area in 1990—2020, with the implementation of the Grain for Green project serving as a turning point. In 1990—2020, there has been a net decrease of 3 170.59 km2 in cropland area, representing a reduction of 17.63%. Furthermore, the proportion of cropland with slopes greater than 15° has been decreasing annually. The main types of dynamic transfer of cropland are into grassland, followed by artificial surfaces. The implementation of ecological projects such as the Grain for Green program, along with urbanization construction, is the primary cause of the decrease in cropland. [Conclusion] Researchers are encouraged to comprehensively analyze the strengths and weaknesses of multi-source products based on their research objectives. Exploring effective methods for fusing and interpreting multi-source data, with a focus on target land classes, can lead to a more in-depth understanding of the land characteristics in a specific region and, consequently, more precise research conclusions.