Abstract:

As a crucial component of the agro-pastoral ecotone, the single cropping and double cropping system transition zone in northern China is mainly located in the central section of the Mid-Spine Belt of Beautiful China, and it is very significant to monitor the spatiotemporal dynamics of cultivated land in this region by using remote sensing technology to effectively protect the cultivated land and sensitive agro-ecological environments, rationally develop the land resources, and make scientific regional development strategies and planning regarding its sustainable development goal. Based on the platform of Google Earth Engine (GEE) with the advantages of remote sensing data processing and analysis, in this study we used its multi-source and multi-resolution remote sensing satellite images and built a recognition feature set of cultivated land, and then, a long-time series of cultivated land information over the study region was obtained totaling 11 periods from 2000 to 2020 by employing random forest classifiers, and those spatiotemporal dynamics characteristics were analyzed. The results show: (1) the cultivated land remote sensing identification products for 11 periods from 2000 to 2020, obtained from GEE, are well reliable and applicable, which show an overall accuracy higher than 90% with a Kappa coefficient over 0.8, based on sampled cross-validation approach, and meanwhile most accuracies ranged between 65% and 80% that were validated against the official statistical data on cultivated land area; and (2) as for spatial distribution, there existed an obvious dominant geomorphic feature in terms of the cultivated land resources over the study region, mainly distributed in the plain and hilly areas with an altitude of less than 3500m, a slope of less than 15 ° and a topographic potential index of less than 1.24, a gravity distribution center of which was located in Lishi County, Shanxi Province while it had shifted 12.88 kilometers towards the southwest over the past nearly 20 years; and (3) for temporal change of the total area of cultivated land resources in this region, there was a slight decrease with 188.17×10⁴ hm² from 2000 to 2020 (especially on the high-quality cultivated land with the most significant decrease, which is mostly located in the low-altitude plain areas), and except for Shaanxi Province, a decreasing trend had been shown in terms of all the other provinces and municipalities during the past nearly 20 years. Hence, this study demonstrates that it is necessary to synthetically use multi-source remote sensing data and other diverse datasets and collaboratively take advantage of scientific and technological means and approaches to enable large-scale, long-term, and/or high-precision cultivated land monitoring and analyses so as to serve reasonably formulating scientifically sound development plans and, further, carrying out differentiated cultivated land protection measures and incentive-penalty mechanisms and optimizing land use structures, etc., and those would become a critical path to mitigate the continuous loss of cultivated land resources in northern China's transitional farming zones and prevent the unreasonable encroachment and/or occupation on farmland (peculiarly upon permanent basic farmland).

Key words: cultivated land monitoring, temporal and spatial characteristics, cropping system transition zone, coping strategy