Subgrid sampling (SGS) has been increasingly integrated into hydraulic modelling to refine terrain representation without excessively increasing computational demands. This study investigates the effects of SGS and nature-based solutions (NbS), focusing on flow velocities, storage volumes and mass balance stability. Using a rain-on-grid TUFLOW model, we simulate a 1-in-30-year flood event across different NbS interventions, including leaky dams, floodplain planting and bunds with floodable depressions. Results demonstrate that higher SGS frequencies of 11 significantly enhance terrain resolution, improving the representation of flow paths and floodplain activation. However, a lower SGS frequency of three introduces notable numerical artefacts, leading to persistent negative cumulative volume errors and misrepresenting water retention within the system. NbS interventions generally minimise peak flow rates, with bunds and depressions (Option 3) proving the most effective, achieving a storage volume increase of up to 162.83% compared to the baseline scenario. Mass balance assessments show that models without SGS overestimate flood storage, potentially misguiding NbS implementation and compliance with flood storage regulations such as the UK Reservoirs Act (1975).