Desertification, primarily wind erosion desertification, is a significant environmental and socioeconomic problem affecting a quarter of the world's land area and nearly a billion people. China is among the most severely affected countries, with wind erosion desertification impacting 1.8 million km². The Mu Us Sandy Land, located in a crucial farming-pastoral ecotone of northern China, is a typical desertification area. Its desertification dynamics directly influence the region's economic development. Traditional desertification monitoring methods are time-consuming and inefficient. Remote sensing technology offers a cost-effective and objective solution for large-scale monitoring and data updating. Previous research on the Mu Us Sandy Land has utilized remote sensing, but long-term, comprehensive analyses of spatiotemporal changes and driving factors are still needed. This study aims to fill this gap by employing Landsat TM/OLI imagery (1990-2017) and meteorological data to analyze desertification dynamics and identify driving forces, providing a scientific basis for resource allocation and sustainable development in the region. The Mu Us Sandy Land's ecological importance as a key barrier in northern China, preventing further desertification and mitigating drought and flood risks, underscores the significance of this research.

Early research on the Mu Us Sandy Land, dating back to the late 1950s, identified anthropogenic factors as contributors to desertification. A 1983 publication assessed the natural environment's role in desertification and proposed a comprehensive control plan. In 1992, UNESCO recognized the Mu Us Sandy Land as one of the world's most environmentally sensitive areas. Several studies have used remote sensing to monitor land desertification in China, including dynamic monitoring research on the Mu Us Sandy Land. These studies have shown varying degrees of desertification change over time and have identified factors such as land use patterns, human activities, and climate as influencing desertification. However, a comprehensive analysis of long-term spatiotemporal dynamics and the interplay of natural and human factors is still lacking. The present study builds upon this existing literature by conducting a detailed analysis of desertification changes in the Mu Us Sandy Land from 1990 to 2017, using a comprehensive set of methods and data sources.

This study used six Landsat TM/OLI remote sensing images (1990, 1995, 2000, 2007, 2010, and 2017) obtained from the geospatial data cloud. Strict geometric and atmospheric corrections were performed. Images were interpreted using an ArcGIS 10.3 platform, with 350 verification points selected from 2017 imagery and verified through field investigation. The overall classification accuracy was 92.6%, with a Kappa coefficient of 91.0%. Desertified land was classified into five types: non-desertified, mildly, moderately, severely, and extremely desertified. The annual growth rate of desertified land was calculated using a formula. A transfer matrix was used to analyze the quantitative transformation and change structure of various land use types. The Aeolian Desertification Index (ADI) was calculated to provide a comprehensive analysis of desertification changes. A desertification land gravity transfer model was employed to describe the dynamic spatial and temporal changes of desertified lands. Meteorological data (precipitation, temperature, and wind speed) from the China meteorological science data sharing service and human factors data (population, livestock, and farmland) from national statistical yearbooks were also used. Statistical analyses, including regression modeling, were conducted to analyze the trends and relationships between different factors.

The analysis revealed a 1680 km² reduction in the total desertified land area over the 30-year period, with an average annual loss of 60 km². The study identified three stages of desertification change: a development stage (1990-2000) with increasing severely and extremely desertified land; a rapid reversal stage (2000-2010) showing a significant decrease in severely and extremely desertified land, and an increase in mildly desertified land; and a stable reversal stage (2010-2017) characterized by a continued decrease in severely and extremely desertified land, with a slower rate of change than in the previous stage. Spatially, the center of gravity for different desertification levels showed a west-to-east distribution, consistent with natural conditions. The Aeolian Desertification Index (ADI) confirmed the three-stage pattern of desertification change, peaking at 2.7 in 2000 and decreasing to 2.0 in 2017. Analysis of climatic factors showed a warm-dry trend from 1990 to 2010, followed by a warm-wet trend from 2010 to 2017. Decreasing wind speed played a significant role in reducing wind erosion and promoting desertification reversal. Human factors, including population growth, livestock production, and changes in cultivated land area, were also found to significantly influence desertification dynamics. The implementation of environmental protection policies, such as the returning farmland to forest (or grassland) program, contributed to the reversal of desertification.

The findings demonstrate a clear turning point in desertification dynamics in the Mu Us Sandy Land around the year 2000. The reversal of desertification is attributed to the combined effects of decreasing wind speed and the implementation of effective environmental policies and human interventions. The warm-dry climate from 1990 to 2010 initially promoted desertification, but the subsequent shift to a warm-wet climate, along with human efforts, significantly reversed this trend. The spatial distribution of desertification reflects the interplay of natural conditions and human activities, with higher desertification levels concentrated in the west and gradually decreasing towards the east. The study highlights the importance of integrating climate change considerations into desertification management strategies. The success of the reversal in the Mu Us Sandy Land provides valuable insights for desertification control in other arid and semi-arid regions.

This study provides a comprehensive analysis of desertification dynamics in the Mu Us Sandy Land over the past three decades. The three-stage pattern of development, rapid reversal, and stable reversal, is clearly documented. Decreasing wind speeds, coupled with effective environmental policies and human interventions, have been instrumental in reversing desertification. The findings underscore the importance of integrated approaches that consider both natural and human factors in desertification management. Future research should focus on further investigating the long-term effects of climate change and the effectiveness of various desertification control measures. Furthermore, studying the economic and social impacts of desertification and the benefits of restoration efforts would add valuable insights to this critical issue.

While this study provides a valuable long-term analysis of desertification dynamics in the Mu Us Sandy Land, it has certain limitations. The reliance on remote sensing data might limit the accuracy of finer-scale spatial changes or ground-level details. Furthermore, the model used for calculating the annual growth rate of desertified land could be further improved to address complexities of desertification processes. Future research could benefit from incorporating more detailed ground-truthing data, using higher-resolution remote sensing imagery, and exploring more sophisticated statistical models.