Several icy moons in our solar system (e.g., Europa and Enceladus) are among the most promising places to search for life beyond Earth, owing to evidence for vast liquid water oceans beneath their frozen surfaces. A remarkable yet poorly understood phenomenon associated with these moons is the eruption of water plumes: jets of water vapor, ice, salts, and organic compounds shooting into space, detected by the Hubble Space Telescope and sampled by NASA's Cassini spacecraft. These plumes offer a rare opportunity to study what lies beneath the ice without landing or drilling, but a fundamental question remains unanswered: where do they originate? Two possible sources exist: (1) the deep subsurface ocean and/or (2) shallow liquid water pockets trapped within the ice crust. This distinction is critical as ocean-sourced water is far more likely to contain the ingredients needed to support life. In contrast, shallow reservoir water is exposed to intense radiation and lacks the necessary conditions for habitability. This four-year PhD project, funded by the Swedish Research Council (Vetenskapsrådet, VR) from 2025 and hosted at Umeå University, addresses this question through a novel computational approach. For the first time, we combine two simulation tools (both developed by our team) into a single self-consistent framework: a hybrid-kinetic plasma model that tracks individual charged-particle trajectories in three dimensions, and a neutral Monte Carlo model simulating gas and dust in the plumes and moon atmospheres. Together, they allow us to simulate, with unprecedented physical accuracy, how Europa's interior ocean, atmosphere, and plumes interact with the surrounding magnetic and plasma environment. In the first project phase, we quantify how Europa's subsurface ocean leaves a detectable electromagnetic imprint on the surrounding plasma, essential groundwork before introducing active plume dynamics. The results will directly support the science objectives of NASA's Europa Clipper and ESA's JUICE missions, both currently en route to Jupiter's icy moons. In the proposed HPC project, simulations are performed using Amitis, a GPU-accelerated plasma model and the only kinetic model capable of performing these calculations at the required resolution and scale. Each production run involves ~25 billion particles tracked across a large three-dimensional grid, requires 16 NVIDIA GH200 GPUs in parallel with a total of 1.5 TB of GPU high-bandwidth memory (~92GB per GPU), and takes <7 days to complete. No general-purpose HPC system currently available in Sweden can provide this level of GPU memory and computation per simulation, making access to Arrhenius essential. We request 22110 GPU-hours per month and 50000 GB of flash storage. Within two weeks of each run, simulation output is transferred to our servers at Umeå University and HPC2N, processed through a well-tested Python-based processing pipeline that analyzes, selects, and compresses data to ~10% of its original volume, and archived for long-term use. Data associated with peer-reviewed publications will be made openly available through Zenodo, in compliance with FAIR and VR open-access requirements.