In recent decades, the vegetation of the Sanjiangyuan region has undergone a series of changes under the influence of climate change, and ecological restoration projects have been implemented. In this paper, we analyze the spatiotemporal dynamics of vegetation in this region using the satellite-retrieved normalized difference vegetation index (NDVI) from the global inventory modeling and mapping studies (GIMMS) and moderate resolution imaging and spectroradiometer (MODIS) datasets during the past 34 years. Specifically, the characteristics of vegetation changes were analyzed according to the stage of implementation of different ecological engineering programs. The results are as follows. (1) The vegetation in 65.6% of the study area exhibited an upward trend, and in 53.0% of the area, it displayed a large increase, which was mainly distributed in the eastern part of the study area. (2) The vegetation NDVI increased to differing degrees during stages of ecological engineering. (3) The NDVI in the western part of the Sanjiangyuan region is mainly affected by temperature, while in the northeastern part, the NDVI is affected more by precipitation. In the southern part, however, vegetation growth is affected neither by temperature nor by precipitation. On the whole region, vegetation growing is more affected by temperature than by precipitation. (4) The impacts of human activities on vegetation change are both positive and negative. In recent years, ecological engineering projects have had a positive impact on vegetation growth. This study can help us to correctly understand the impact of climate change on vegetation growth, so as to provide a scientific basis for the evaluation of regional ecological engineering effectiveness and the formulation of ecological protection policies.

Mu Us Sandy Land is a typical semi-arid vulnerable ecological zone, characterized by vegetation degradation and severe desertification. Effectively identifying desertification changes has been a topical environmental issue in China. However, most previous studies have used a single method or remote sensing index to monitor desertification, and lacked an efficient and high-precision monitoring system. In this study, an optimal monitoring scheme that considers multiple indicators combination and different machine learning methods (Classification and Regression Tree-Decision Tree, CART-DT; Random Forest, RF; Convolutional Neural Networks, CNN) was developed and used to analyze the spatial–temporal patterns of desertification from 2000 to 2018 in Mu Us Sandy Land. The results showed that: (a) The random forest model performed best for monitoring desertification based on medium and low-resolution remote sensing images, and the four-index combination (Albedo, NDVI, LST and TGSI) obtained the highest classification accuracy (OA = 87.67%) in Mu Us Sandy Land. Surprisingly, the model accuracy of the three-index combination (NDVI, LST and TGSI) (OA = 85.74%) is comparable to the four-index combination. (b) The TGSI index used to characterize soil information performs well, while the LST is not conducive to the extraction of desertified land in several desertification monitoring indicators. (c) Since 2000, the area of extremely severe desertified land has shown a reversal trend; however, there is significant interannual fluctuation in the total and light desertification land area affected by extreme climate. This research provides a novel approach and a valuable reference for monitoring the evolution of desertification in regional studies, and the results improve the research system of desertification and provide a data basis for desertification cause analysis and prevention.

Biological soil crusts (BSCs) play an essential role in desert ecosystems. Knowledge of the distribution and disappearance of BSCs is vital for the management of ecosystems and for desertification researches. However, the major remote sensing approaches used to extract BSCs are multispectral indices, which lack accuracy, and hyperspectral indices, which have lower data availability and require a higher computational effort. This study employs random forest (RF) models to optimize the extraction of BSCs using band combinations similar to the two multispectral BSC indices (Crust Index-CI; Biological Soil Crust Index-BSCI), but covering all possible band combinations. Simulated multispectral datasets resampled from in-situ hyperspectral data were used to extract BSC information. Multispectral datasets (Landsat-8 and Sentinel-2 datasets) were then used to detect BSC coverage in Mu Us Sandy Land, located in northern China, where BSCs dominated by moss are widely distributed. The results show that (i) the spectral curves of moss-dominated BSCs are different from those of other typical land surfaces, (ii) the BSC coverage can be predicted using the simulated multispectral data (mean square error (MSE) < 0.01), (iii) Sentinel-2 satellite datasets with CI-based band combinations provided a reliable RF model for detecting moss-dominated BSCs (10-fold validation, R² = 0.947; ground validation, R² = 0.906). In conclusion, application of the RF algorithm to the Sentinel-2 dataset can precisely and effectively map BSCs dominated by moss. This new application can be used as a theoretical basis for detecting BSCs in other arid and semi-arid lands within desert ecosystems.

The aeolian desertification in Mu Us Sandy Land (MUSL) in northern China have been paid much attention, but the relative contributions of climate change and human activities to desertification dynamics are still not clear. Based on the Landsat MSS, TM, ETM+ and OLI images in 1975, 1990, 1995, 2000, 2005, 2010 and 2015, we developed a database of aeolian desertification land distribution, discussed the spatial and temporal variation of aeolian desertification, and discovered the relative contributions of climate change and human activities to desertification reversal, using the trends of the potential net primary productivity (NPP) and the human-influenced NPP with meteorological data and MODIS NPP products. The results indicated that aeolian desertification developed firstly from 1975 to 2000, with serious and severe aeolian desertification land continually increasing, and then changed into a reversal state from 2000 to 2015, as the serious aeolian desertification land decreased, although the severe, moderate and light aeolian desertification land lightly increased. Human activities were the dominant factor in desertification dynamics in MUSL and had different contributions to aeolian desertification reversal in different periods. This study will improve our understanding of the processes of aeolian desertification.

Drought monitoring is an important scientific basis for drought impact evaluation and the selection of mitigation strategies. Since the drivers of drought vary among regions, there is no universal drought index applicable to different regions. The Shiyang River Basin, an inland river basin, located in Gansu Province, China, has a closed water cycle system. Drought is a dominant nature disaster for the sustainable development of the region. Thus, this is an ideal area to explore the suitability of drought-monitoring indices. Here, we took the Shiyang River Basin as an example, in order to explore suitable indicators for agricultural drought monitoring in inland river basins. This study assessed the twelve different widely used drought indices used for monitoring the impact of drought on crop growth, represented by net primary production (NPP). The results showed that the vegetation status-based drought indices (VCI and NVSWI) had the highest significant correlation (0.6 ≤ |R| ≤ 1, p < 0.05) on NPP, and the integrated drought indices (DSI and ISDI) had the strong significant correlation (0.4 ≤ |R| < 0.6, p < 0.1). These four indices are good indicators for agricultural drought monitoring. Studies based on these four indices showed that agricultural drought has a tendency to slow down from 1982 to 2020. This is inconsistent with the monitoring of drought indices based on the meteorological variables that show a trend of increasing drought. This is mainly due to the increased efficiency of water management and its use in inland river basins. This indicates that other water resource information, such as runoff, should be included to construct an integrated agricultural drought-monitoring indices in management intensive regions, such as in an inland river basin.