格尔木河流域盐渍化时空分布特征及影响因素分析

Spatiotemporal characteristics and inf luence factors of salinization in the Golmud River Catchment

关键词:  遥感监测  土壤盐渍化  包气带岩性  地下水埋深  格尔木河流域

Key words:  remote sensing monitoring  soil salinization  lithology of unsaturated zone  depth to groundwater table  Golmud River Catchment

摘要:

旱区土壤盐渍化是威胁区域粮食安全的主要因素之一，研究其时空分布特征及影响因素对土地资源合理利用和盐渍化防控具有重要意义。利用2000—2021年MODIS数据，构建由归一化植被指数（NDVI）和盐分指数（SI）组成的遥感盐渍化监测模型，分析我国西北典型内陆河流域——格尔木河流域盐渍化的时空分布与动态变化，并结合土壤颗粒组成、气象资料和地下水埋深探讨其演变影响因素。结果表明：（1）监测指数与实测全盐量的相关系数R²=0.66，表明该指数可用于流域内盐渍化动态监测；（2）2000—2021年盐渍土总面积减少647 km²，盐渍化整体呈减轻趋势，但局部仍有加剧；（3）土地利用转移矩阵显示，中度盐渍土占比最高，其与重度盐渍土的动态变化主导了流域盐渍化特征；（4）盐渍化受多因素共同控制，细粒粉土促进盐分积聚，浅埋地下水提供盐分来源，强烈的蒸发作用则驱动了盐分上升。

Abstract:

Background, aim, and scope Soil salinization is one of the ten major threats to soil health. According to the latest statistics released by the FAO in 2022, about 8.7% of the world’s land area is affected by salinization. This issue is particularly critical in the vast arid and semi-arid regions of Northwest China. Salinization not only severely threatens the ecological environment but also poses significant challenges to food security. Therefore, investigating its spatiotemporal characteristics and influencing factors is of great significance for the rational use of land resources and the formulation of policies to mitigate salinization. The Golmud River Catchment, a typical inland river basin in Northwest China, faces severe salinization challenges due to the combined effects of natural processes and human activities. To address these issues, this study conducted a 22-year remote sensing analysis to clarify the spatiotemporal characteristics of salinization in the catchment and to identify its main influencing factors. Materials and methods Based on MODIS data from 2000 to 2021, a remote sensing salinization monitoring model combining the normalized difference vegetation index (NDVI) and the salinity index (SI) was applied to analyze the spatiotemporal dynamics of salinization in the catchment. Furthermore, multiple correlation analysis was conducted to examine the relationships among soil salinization, soil particle size, meteorological factors, and groundwater depth, in order to clarify the salinization characteristics of the catchment. Results The results show that: (1) using low-resolution but highly integrated MODIS data, the proposed remote sensing salinization monitoring model achieved R² of 0.66 with measured total salinity. Considering the regional hydrogeological conditions, the model is applicable for regional-scale salinization monitoring; (2) from 2000 to 2021, the total area of saline soil in the catchment only decreased by 647 km², indicating that salinization has generally slowed while local intensification persists. The overall dynamics of salinization are mainly driven by changes in moderately and strongly saline soils. Salinization is positively correlated with solar radiation (0.35) and surface temperature (0.30), and negatively with air pressure (−0.23); (3) the correlation between rainfall and salinization is weak (R²=0.02) due to spatial heterogeneity within the basin; (4) salinization in the catchment is jointly controlled by multiple factors: silt-grained soil promotes salt accumulation, shallow groundwater provides the salt source, and strong evaporation drives upward salt migration. Discussion Generally, the degree of salinization is determined by the grain size of the unsaturated zone and the depth of the groundwater table. In theory, salinization should not occur in the Gobi Desert areas of the study catchment. However, the results show a diametrically opposite pattern, which contradicts the expected hydrogeological conditions. The reasons for this phenomenon may involve two aspects. (1) The mathematical formulation of the model is not suitable for areas where NDVI values are less than 0, such as water bodies or zones with deep groundwater tables. In addition, for areas with dense vegetation where NDVI approaches 1, the formula degenerates into SI, whose applicability in such regions remains questionable. (2) The salinity index itself has limitations in distinguishing between dry, barren soil and saline soil. Precipitation contributes little to catchment salinization, mainly due to the unique physical and geographical features of the study area. Moreover, the influence of groundwater depth on salinization is not linear but characterized by a threshold effect. This may be attributed to the fact that some sampling sites are located in water-rich areas with high vegetation coverage, which are represented in the model as having low salinization. However, within the phreatic evaporation limit depth, as the groundwater table deepens, vegetation coverage gradually decreases, and salinization tends to intensify. Conclusions The model can effectively be used for dynamic monitoring of regional salinization when combined with hydrogeological condition analysis. The total area of saline soil in the Golmud River Catchment has decreased over the past 22 years. However, the data also reveal a complex pattern of overall mitigation and local intensification of salinization. The salinization process is jointly influenced by soil texture, meteorological factors, and groundwater depth. Recommendations and perspectives Salinization control in the Golmud River Catchment is complex, partly due to the inf luence of salt inputs from potassium salt production bases. Regular dynamic monitoring using remote sensing is therefore crucial for both ecological management and industrial development in the basin. However, the data used in this study have limitations, including low spatial resolution and the common susceptibility of multispectral data to atmospheric interference. Future studies could improve monitoring accuracy by integrating synthetic aperture radar (SAR) and unmanned aerial vehicle (UAV) data as complementary information sources.