A Monte Carlo Framework to Evaluate the Benefits of Flood Warnings in an Urban Flood-Prone Polder Area

Polders, situated in delta regions and enclosed by dykes to avert flooding (from rivers or tides), depend on pumping mechanisms to transfer water from internal artificial rivers to external ones, particularly during storms. Urban polders are highly susceptible to pluvial flooding if their drainage, storage, and pumping capacities are insufficient. This study introduces a Monte Carlo (MC) framework to assess the effectiveness of rainfall threshold-based flood warnings in mitigating pluvial flooding in an urban flood-prone polder area based on 24-hour forecasts. The framework calculates metrics including the potential duration of waterlogging, the maximum area inundated, and the costs of pump operation, taking into account a wide range of possible storm scenarios. The benefits of flood warnings are evaluated by comparing these metrics across different scenarios: scenarios with no warnings, perfect forecasts, deterministic forecasts, and probabilistic forecasts. Probabilistic forecasts incorporate the idea of 'predictive uncertainty' (PU). A specific polder region in Nanjing was selected for this case study. Findings indicate a balance between waterlogging duration and pumping costs, and demonstrate that probabilistic rainfall predictions can significantly improve these metrics. These insights are valuable for designing and assessing the advantages of a rainfall threshold-based flood early warning system (FEWS) in a polder area.