Water management is a problem of matching supply and demand whilst sustaining environmental conditions for a range of sectors and ecosystems, potentially under changing conditions of climate or demand. In dryland regions, this is particularly difficult given low available water supply and high climate variability, often with lack of data for operations, planning and design. Addressing these challenges at national scale requires whole-system approaches to incorporate the range of relevant sectors and their interactions, and multi-scale approaches to capture the large-scale drivers of water availability and the fine-scale variability of supply and demand within catchments, irrigation districts or urban areas. In the context of the AFRI-SMART project “EO-Africa multi-scale smart agricultural water management” funded under the ESA EO Africa - National Incubators EXPRO+ programme, we have developed a multi-scale, multi-model approach to help address water management challenges in Morocco. On-going drought conditions in the country for the past 5 years have left reservoirs without water for irrigation, which must be prioritized for public water supply, impacting on food production, agricultural exports and farmer incomes. More accurate information on water resources distribution in space and time across scales and sectors is needed to address sustainable agriculture, to help guarantee food and water security, and increase resilience to hydro-meteorological extremes. At national scale multiple sources of information from ground observations, satellite remote sensing, and climate and hydrological models are integrated to provide the best estimate of hydroclimate and drought indices to characterize the large-scale variability in water supply. This feeds into basin scale hydrological modeling, focused on the Oum Er-rbia basin using the HydroBlocks modelling framework, which combines a 1-D land surface model with a cluster-based landscape representation, allowing large-domain simulations at 10s metres resolution. HydroBlocks is coupled to the RAPID stream flow routing scheme to provide high resolution stream flow estimates. Predicted stream flow is routed to the main reservoirs in the basin which are simulated using a simple mass balance approach. Withdrawals from the reservoirs are supplied to one of the basin’s irrigation districts of Doukkala. Actual and optimized irrigation water needs for specific crops, at fine resolution (daily, 10 m) are predicted using the energy-crop-water balance model FEST-EWB-SAFY driven by Landsat LST and Sentinel2 vegetation indices. The system is used to provide historic reconstructions of water availability and analyzed to identify times of supply risks. The system is also implemented in monitoring and seasonal forecast mode as a tool to understand upcoming risks to water supply and potential interventions to reduce risks, such as provision of early warning of risks, options for adjusted reservoir management, or altered/optimized irrigation scheduling. An open online decision support tool has been developed to provide intuitive near real-time visualization of information from the satellites/models and explore forecasts and future scenarios. We also discuss the collaboration with end user groups in helping to define the management problem and identification of critical decisions in water management across scales.