Monitoring changes in groundwater is challenging due to the hidden nature of aquifers and the limited availability of observational infrastructure, such as monitoring bores. Inconsistent national reporting practices and sparse in situ groundwater monitoring networks affect the ability to reliably estimate groundwater abstraction and withdrawal, complicating efforts to assess aquifer status and inform groundwater management. These data and knowledge gaps also constrain our understanding of key drivers of groundwater variability, including groundwater pumping, rainfall variability, and land use change. The invisibility of groundwater and lack of suitable monitoring capability continue to pose governance challenges, making satellite tools vital for sustainable groundwater management. As demonstrated in this chapter, satellite-based monitoring offers a scalable solution, enabling integration into national water databases to support freshwater accounting and resource management. Satellite-based groundwater, as estimated from the Gravity Recovery and Climate Experiment (GRACE) mission, is essential to advancing understanding of groundwater hydrologyGroundwater hydrology at different spatial scales. This chapter discusses the capabilities of satellite-based techniques and other traditional and modelling approaches (e.g. GRACE, radar remote sensing, data fusion, modelling, and assimilation) for groundwater assessment. Using a case study over the Murray Darling basin, this chapter demonstrates a machine learning approach to reveal soil moisture uncertainties in GRACE groundwater assessment. It emphasizes the need to understand and quantify uncertainty in auxiliary soil moisture data products usually used to isolate groundwater signals from GRACE-derived changes in terrestrial water storage. The chapter demonstrates how satellite gravity methods such as GRACE can track groundwater variability and reveal responses to climate and human pressures. Although GRACE has revolutionized global monitoring of freshwater changes, its coarse spatial resolution limits its usefulness for local and regional water management. To overcome this limitation, high-resolution GRACE data are now being produced through various machine learningMachine learning downscaling techniques. This chapter reviews recent downscaling approaches and highlights the emergence of novel machine learning methods that enhance the spatial resolution of GRACE-based groundwater estimates.