A forensic engineering framework for flood management of cascade reservoir systems

Extreme floods are increasing in frequency and severity, exposing critical gaps in reservoir operation policies and post-event accountability. This study proposes a forensic engineering framework to evaluate and optimize flood management in cascade reservoir systems. The framework is applied to Iran's Karkheh Basin, Iran, and combines scenario-based analysis, a dynamic Standard Operating Policy (SOP) extended to gated spillways, and real-time optimization using inflow forecasts. Five scenarios were analyzed for the April 2019 flood: (1) actual historical operation, (2-3) SOP-based rule curves driven by observed and forecasted inflows, (4) ideal hindsight operation, and (5) rolling-horizon real-time optimization with SVR (monthly) and LSTM (daily) inflow forecasts embedded in a Genetic Algorithm. Results show that, without the proposed Mashoureh Reservoir, relative downstream flood damage was reduced by 18%, 8%, 59%, and 39% under Scenarios 2-5, respectively, compared to Scenario 1. Inclusion of the Mashoureh reservoir reduces damage by 34%, 37%, 100%, and 55% according to scenarios 2-5. However, the impact of the Mashoureh reservoir was limited because 71% of the flood volume originated downstream this reservoir. These findings underscore the importance of combining structural and non-structural measures in the analysis of reservoir operation for flood-control purpose, and demonstrate the value of forensic scenario evaluation for flood preparedness, real-time operations, and planning of future infrastructure.