Water security is a strategic issue for the sustainable development of the Chinese nation. Urban stormwater and mine water serve not only as hazards but also as valuable unconventional water resources. Their deep geological storage presents dual benefits of strategic water resource reserves and disaster mitigation. Nevertheless, the large-scale application of this technology is hindered by the absence of standardized criteria for target layer selection, unclear coupling mechanisms of the thermal-stress-seepage-hydrochemical-microbial fields leading to uncontrollable permeability enhancement and capacity expansion, ambiguous evolutionary patterns of groundwater environments in target layers following storage, and a lack of methodologies for assessing geo-environmental effects. This study addresses the core scientific issue of geo-environmental effects of deep geological water storage by constructing a comprehensive scientific framework that encompasses refined hydrogeological evaluation, intelligent target layer optimization, geological environment evolution, fracturing-induced capacity expansion, microbial regulation, and multi-field coupling. Focusing on mine water from inland mining areas and urban stormwater in the Guangdong-Hong Kong-Macao Greater Bay area, the framework quantitatively evaluates the hydrogeological structure and host geological environment of deep target layers such as sandstone and limestone. By integrating machine learning with multi-source geological data, it establishes an environmentally friendly intelligent optimization method for selecting deep storage target layers. Through water-rock-microbe coupling simulations and field deep storage experiments, a coupled numerical model of the thermal-stress-seepage-hydrochemical-microbial fields is developed using AI high-resolution modeling and digital twin technologies. This model reveals the mechanisms of water-rock interactions during deep water storage, elucidates the role of microbial communities in regulating seepage channel opening and closure, and delineates the evolutionary patterns of the host geological environment. Furthermore, an assessment method for geo-environmental effects of deep storage is established, accompanied by differentiated risk identification, early warning, and prevention and control strategies. The expected results will provide a scientific basis for resolving technical bottlenecks in geological water storage and advancing the safe and efficient utilization of unconventional water resources.