欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2026, Vol. 42 ›› Issue (17): 113-120.doi: 10.11924/j.issn.1000-6850.casb2026-0215

• 资源环境生态土壤气象 • 上一篇    下一篇

Sentinel-1反演内蒙古河套地区近10 a土壤水分的研究

李耀琛1(), 韩仙桃2,3(), 王盈2,3, 李璐4   

  1. 1 内蒙古自治区生态与农业气象中心， 呼和浩特 010051
    2 呼和浩特市气象局， 呼和浩特 010020
    3 呼和浩特国家气候观象台， 呼和浩特 010010
    4 呼和浩特市新城区气象局， 呼和浩特 010070
  • 收稿日期:2026-03-25 修回日期:2026-06-16 出版日期:2026-09-15 发布日期:2026-09-09
  • 通讯作者:
    韩仙桃，女，1973年出生，内蒙古呼和浩特人，正高级工程师，本科，主要从环境气象方面的研究。通信地址：010020 内蒙古呼和浩特市金桥开发区世纪西街金桥二路 呼和浩特市气象局，Tel：0471-4348050，E-mail：。
  • 作者简介:

    李耀琛，男，1992年出生，内蒙古呼和浩特人，中级工程师，本科，主要从事卫星遥感在生态、农业等领域的应用研究。通信地址：010051 内蒙古自治区呼和浩特市新城区海拉尔大街气象小区，E-mail：。

  • 基金资助:
    内蒙古自治区重点研发和成果转化计划项目“面向黄河流域生态系统的高温干旱复合灾害风险预警关键技术”(2026YFHH0129); 内蒙古自治区重点研发和成果转化计划项目“气候及水资源承载力约束下内蒙古黄河流域沙漠沙地生态服务功能响应及风险评估”(2025YFHH0219); 内蒙古自治区自然科学基金分析测试专项项目“基于轨道植被观测仪的半干旱区人工草地碳汇监测与诊断系统构建”(2026FX017); 内蒙古自治区气象局科技创新项目“大青山保护区碳储能力时空变化及与气象因素相关性分析”(nmqxkjcx202601); 内蒙古自治区气象局科技创新项目“基于多源数据融合的牧草图像识别方法应用”(nmqxkjcx202501); 中国气象局大气探测重点开放实验室资助科研项目“基于特定目标物的交叉极化隔离度检验方法研究”(2024KLAS07M); 内蒙古自治区气象局科技创新项目“WRF-Hydro Entension模型在洪水预报预警中的应用研究”(nmqxkjcx202403)

A Study on Soil Moisture Inversion in Hetao Region of Inner Mongolia Using Sentinel-1 Data over Past Decade

LI Yaochen1(), HAN Xiantao2,3(), WANG Ying2,3, LI Lu4   

  1. 1 Inner Mongolia Center of Ecology and Agrometeorology, Hohhot 010051
    2 Hohhot Meteorological Bureau, Hohhot 010020
    3 Hohhot National Climate Observatory, Hohhot 010010
    4 Hohhot Xincheng Meteorological Bureau, Hohhot 010070
  • Received:2026-03-25 Revised:2026-06-16 Published:2026-09-15 Online:2026-09-09

RichHTML

2

PDF (PC)

16

摘要：

为评估Sentinel-1合成孔径雷达(SAR)数据在典型干旱半干旱农区的土壤水分反演能力，本研究以内蒙古河套地区为研究区，基于2015—2024年获取的99景Sentinel-1 GRD影像，分别提取VV和VH两种极化方式的后向散射系数，与研究区内10个土壤水分站监测站点10 cm、20 cm深度的体积含水量、重量含水率、有效水分贮存量和相对湿度等实测数据进行相关性分析，并应用水云模型评估植被校正效果。结果表明：（1）VV极化方式的后向散射系数与10 cm深度各土壤水分指标均呈显著负相关，决定系数(R2)可达0.75以上，其反演效果显著优于VH极化；20 cm深度相关性显著下降，验证了C波段SAR在黏壤土条件下有限的穿透深度（约10 cm）。（2）分月验证结果显示，VV极化与10 cm土壤相对湿度的相关性呈现明显季节性特征：7月最高，8月次之，5月最低，此季节性变化与河套地区雨热同期的降水分布高度吻合，表明Sentinel-1 SAR的VV极化能有效反映汛期主降水时段的土壤水分动态。（3）不同地类对比分析表明，VV极化在作物覆盖区和草地覆盖区均表现良好，R2分别达0.80和0.83，但作物覆盖区的月际波动更为显著，受生育期蒸散和降水脉冲的共同影响。（4）采用默认经验参数的水云模型校正后，反演精度未见明显提升，表明模型参数需针对研究区的具体下垫面类型进行本地化率定。本研究证实Sentinel-1 SAR的VV极化方式可有效反演干旱半干旱区0~10 cm表层土壤水分，反演精度受季节降水和地类覆盖的显著调制，可为数据匮乏地区土壤水分的业务化监测及农业干旱预警提供技术支撑。

关键词: 土壤水分, Sentinel-1 SAR, 反演, 河套地区, 数据融合, 机器学习, 水云模型

Abstract:

This study evaluates the potential of Sentinel-1 synthetic aperture radar (SAR) data for retrieving surface soil moisture in the Hetao region of Inner Mongolia, a typical arid and semi-arid agricultural area, over a 10-year period from 2015 to 2024. A total of 99 Sentinel-1 GRD images were processed to obtain backscattering coefficients for both VV and VH polarizations. The backscattering coefficients were correlated with in-situ soil moisture measurements from 10 meteorological stations at depths of 10 cm and 20 cm, including volumetric water content, gravimetric water content, effective water storage, and relative humidity. The water cloud model (WCM) was applied to assess its effectiveness in vegetation correction. The results demonstrate that the VV-polarized backscattering coefficient exhibits a strong negative correlation with soil moisture at 10 cm depth, with coefficients of determination (R²) exceeding 0.75 across all moisture metrics, significantly outperforming VH polarization. In contrast, correlations with 20 cm soil moisture were weak, confirming the limited penetration depth of C-band SAR. Seasonal analysis reveals distinct temporal patterns: the highest correlations occur in July and August, while the lowest correlations appear in May during the crop emergence phase. Land cover comparison indicates robust performance across both cropland (R²=0.80) and grassland (R²=0.83) sites, though croplands exhibit stronger monthly variability due to crop phenological dynamics. The WCM showed negligible improvement in retrieval accuracy when using default empirical parameters, suggesting the need for locally calibrated vegetation parameters. These findings confirm that Sentinel-1 VV polarization provides reliable surface soil moisture estimates (0-10 cm) in semi-arid agricultural regions, with performance strongly modulated by seasonal precipitation patterns and land cover type. The methodology enables operational soil moisture monitoring to support irrigation scheduling and drought early warning in data-scarce regions.

Key words: soil moisture, Sentinel-1 SAR, retrieval, the Hetao region, data fusion, machine learning, WCM

中图分类号: