基于多源遥感数据融合的灌溉面积监测技术与应用

Irrigation area monitoring technology and application based on the fusion of multi-source remote sensing data

    摘要
  • 摘要:
    目的 基于多源遥感数据融合监测技术，精确识别灌溉面积已成为确保粮食生产稳定性和提高水资源利用效率的关键。
    方法 针对我国西北部沙漠绿洲灌区，通过综合运用MODIS、Landsat遥感数据以及我国气象局陆面数据同化系统（CLDAS-V2.0）和SMAP（Soil moisture active passive）再分析数据，结合增强型自适应时空融合算法（Enhanced spatial and temporal adaptive reflectance fusion model, ESTARFM）和随机森林（Random forest, RF）回归模型，构建了逐日高分辨率地表数据集，并利用该数据集对表层土壤水分进行反演，以提取灌溉面积。
    结果 所构建的表层土壤水分反演模型验证集决定系数（R2）为0.77，平均绝对误差（MAE）和均方根误差（RMSE）分别为0.024和0.038 cm3/cm3。在反演所得生育期日尺度表层土壤水分基础上，对反演的逐日土壤水分时间序列进行突变分析，进而识别若羌河灌区生育期灌溉事件，与实地观测的灌溉数据对比，灌溉事件的识别总体精度达到95%。
    结论 结果验证了融合多源遥感数据与机器学习算法在构建高时空分辨率土壤水分数据集方面的有效性。该方法能够捕捉荒漠绿洲区农田水分的动态变化及灌溉行为，不仅为干旱区农业水资源的精细化管理和灌溉调度提供了可靠的技术支撑，也为区域水资源优化配置及保障粮食安全提供了重要的科学依据。

     

    Abstract:
    Objectives Detection technology based on multi-source remote sensing data fusion, accurately identifying irrigated areas is crucial for ensuring the stability of food production and enhancing water resource efficiency. This study addresses the desert oasis irrigation districts in Northwestern China.
    Methods In the context of desert oasis irrigation districts in northwest China, this study employs a comprehensive approach integrating MODIS and landsat remote sensing data, alongside the China meteorological administration's land data assimilation system (CLDAS-V2.0) and SMAP (Soil Moisture active passive) reanalysis data. By utilizing the enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM) and the random forest (RF) regression model, a daily high-resolution surface data set is constructed. This data set is then used to invert surface soil moisture, enabling the extraction of irrigation areas.
    Results The surface soil moisture inversion model achieved a coefficient of determination (R2) of 0.77, a mean absolute error (MAE) of 0.024 cm3/cm3, and a root mean square error (RMSE) of 0.038 cm3/cm3. Based on the inverted daily scale surface soil moisture during the growing season, a change-point analysis of the daily soil moisture time series was conducted. This allowed for the identification of irrigation events in the Ruoqiang River irrigation district during the growing period. Comparison with field-observed irrigation data showed that the overall accuracy of irrigation event identification reached 95%.
    Conclusions The results validated the effectiveness of integrating multi-source remote sensing data and machine learning algorithms in constructing a high spatiotemporal resolution soil moisture dataset. This method is capable of capturing the dynamic variations of farmland moisture and irrigation behaviors in desert oasis regions. It not only provides reliable technical support for the refined management of agricultural water resources and irrigation scheduling in arid areas, but also offers an important scientific basis for optimizing regional water resource allocation and ensuring food security.

     

    • HTML全文
    • 参考文献(29)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回