基于GEE的西安市土地裸露化指数构建与时空演变监测

Construction and Spatiotemporal Evolution Monitoring of Bare Patch Index in Xi'an Based on GEE

摘要:

土地裸露化是反映区域地表覆盖退化、裸土暴露和生态环境变化的重要表征。针对单一遥感指数难以全面表征土地裸露化的问题，本文提出一种基于Google Earth Engine平台，选取改进型土壤调节植被指数（MSAVI）、裸土指数（BSI）、盐分指数（SI）和湿度分量（WET）作为基础指标，从4个不同维度构建土地裸露化指数（BPI）的方法。在此基础上，采用主成分分析方法实现多指标综合集成，并结合分层随机样本、高分辨率影像人工解译和不同遥感指数对比，验证BPI对土地裸露化的表征有效性。结果表明，第一主成分贡献率较高，能够较好集成土地裸露化相关信息;BPI与真实土地裸露化具有较高一致性，其Pearson相关系数和R2分别为0.944和0.890，RMSE和MAE分别为0.102和0.065，均优于NDVI、MSAVI、BSI和SI等单一指数，表明BPI能够更稳定地反映土地裸露化空间差异。基于BPI的时空监测结果表明，西安市土地裸露化具有明显空间异质性，不同等级裸露区域在研究期内呈一定变化。综上，BPI能够从多维地表特征综合刻画土地裸露化，可为区域土地退化监测、农业资源利用和国土空间治理提供参考。

Abstract:

Bare soil exposure is an important indicator reflecting regional surface cover degradation, bare soil distribution, and ecological environmental change. To address the limitation that a single remote sensing index cannot comprehensively characterize bare soil exposure, a bare patch index (BPI) construction method for Xi'an was proposed based on the Google Earth Engine platform. The modified soil-adjusted vegetation index (MSAVI), bare soil index (BSI), salinity index (SI), and wetness component (WET) were selected as basic indicators to characterize bare soil exposure from four dimensions, including vegetation coverage, bare soil exposure, salinity or high-reflectance features, and surface moisture conditions. Principal component analysis was then used to integrate these multiple indicators. In addition, stratified random samples, manual interpretation of high-resolution imagery, and comparison with different remote sensing indices were used to validate the effectiveness of BPI in characterizing bare soil exposure. The results showed that the first principal component had a high contribution rate and could effectively integrate information related to bare soil exposure. BPI showed high consistency with the observed bare soil exposure values, with a Pearson correlation coefficient and R2 of 0.944 and 0.890, respectively, and RMSE and MAE values of 0.102 and 0.065, respectively. These results were superior to those of single indices such as NDVI, MSAVI, BSI, and SI, indicating that BPI can more stably reflect the spatial differences in bare soil exposure. The spatiotemporal monitoring results based on BPI showed that bare soil exposure in Xi'an exhibited obvious spatial heterogeneity, and bare soil exposure areas of different grades changed to varying degrees during the study period. The results indicated that BPI can comprehensively characterize bare soil exposure based on multidimensional surface features, providing a reference for regional land degradation monitoring, agricultural resource utilization, and territorial spatial governance.