Abstract
Flooding is one of the most destructive natural disasters, posing significant risks to under-construction and existing structures. It can also compromise critical infrastructure, such as roads and railways, by weakening embankments. While infrastructure damage is severe, the foremost concern remains the population's safety, making technological advancements and timely information dissemination crucial. Flood forecasting is vital in preparing communities and enabling flood defense organizations to respond effectively. This study aimed to develop a reliable flood forecasting method for the downstream sections of the Marcal River, where population density is high, using real-time data.
Accurate flood forecasting relies on a comprehensive monitoring network and precise measurements that predict water flow and other hydrological conditions over several days. Real-time data during flood events is also essential for emergency response. Key hydrological and meteorological factors, including water levels, flow rates, and precipitation, are integral to this process. The study analyzed daily water flow data from 1960 to 2018, collected from stations along the Marcal River and its tributaries, combined with precipitation data, to forecast the river's flow at its outlet in Mórichida. Multi-level regression analysis, incorporating first- and second-order polynomials, was used to predict flood peaks at this outflow. The model employed flood wave peaks and simultaneous rising or receding flows from five additional river stations. Focusing on events with peak flows exceeding 20 m3/s, the researchers identified 68 cases, with 9-20 measurements per event. Confidence and prediction intervals confirmed the model's accuracy, predicting flood peaks within ±10 m3/s, offering a reliable, less complex alternative to traditional models.
