Remotely sensed data can be valuable for effective groundwater resource supply and demand monitoring

This study illustrated how remotely sensed data can be valuable for effective groundwater resource supply and demand monitoring, specifically for intensively irrigated areas reliant on groundwater, without the need for extensive skills and expensive in situ data.
The datasets used to create the charts for each research chapter are included. In Chapter 4, in situ precipitation observations are compared to the remotely sensed precipitation product selected for use, namely, CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data) at the Secondary Catchment A2 spatial resolution, and monthly, seasonal and annual temporal resolutions. In Chapter 5, the remotely sensed groundwater storage anomalies were downscaled to a higher spatial resolution using the Random Forest Machine Learning classifier, and CHIPRS precipitation and MODIS actual evapotranspiration as independent variables. In Chapter 6, active cultivation was identified from monthly Sentinel-2 composites and divided into monthly rainfed and irrigated areas using an existing product. Crop and irrigation water use were estimated using remotely sensed data and compared to the downscaled GWS changes from Chapter 6. Chapter 7 explored the implications and outcomes of incorporating a range of different specific yield (Sy) values into the downscaled GWS product created.

No field data was collected, and all products generated relied on open source, published datasets. Google Earth Engine codes are included, and made available with the publications.

本研究阐释了遥感数据在有效监测地下水资源供需方面所具有的宝贵价值，特别是在对地下水依赖程度较高的重度灌溉区域，即便无需具备广泛的专业技能和昂贵的现场数据。用于创建各研究章节图表的数据集已包含在内。在第四章中，对现场降水量观测与选用的遥感降水量产品进行了比较，即采用次级流域A2空间分辨率的CHIRPS（气候灾害小组红外降水量与站点数据）产品，并按月度、季节性和年度时间分辨率进行。第五章中，利用随机森林机器学习分类器将遥感地下水储存异常值下尺度至更高的空间分辨率，并以CHIRPS降水和MODIS实际蒸散量为独立变量。第六章中，通过月度Sentinel-2复合影像识别了活跃耕作，并利用现有产品将其划分为月度雨养和灌溉区域。利用遥感数据估计作物和灌溉用水量，并将其与第六章中下尺度化的GWS变化进行了比较。第七章探讨了将一系列不同的具体产量（Sy）值纳入下尺度化的GWS产品中所产生的意义和结果。未收集任何现场数据，所有生成产品均依赖于开源、已发布的数据库。包含Google Earth Engine代码，并随出版物一同提供。