Abstract
Worldwide, machine learning (ML) is increasingly being used for developing flood early warning systems (FEWSs). However, previous studies have not focused on establishing a methodology for determining the most efficient ML technique. We assessed FEWSs with three river states, No-alert, Pre-alert and Alert for flooding, for lead times between 1 to 12 h using the most common ML techniques, such as multi-layer perceptron (MLP), logistic regression (LR), K-nearest neighbors (KNN), naive Bayes (NB), and random forest (RF). The Tomebamba catchment in the tropical Andes of Ecuador was selected as a case study. For all lead times, MLP models achieve the highest performance followed by LR, with f1-macro (log-loss) scores of 0.82 (0.09) and 0.46 (0.20) for the 1 h and 12 h cases, respectively. The ranking was highly variable for the remaining ML techniques. According to the g-mean, LR models correctly forecast and show more stability at all states, while the MLP models perform better in the Pre-alert and Alert states. The proposed methodology for selecting the optimal ML technique for a FEWS can be extrapolated to other case studies. Future efforts are recommended to enhance the input data representation and develop communication applications to boost the awareness of society of floods.
Highlights
Flooding is the most common natural hazard and results worldwide in the most damaging disasters [1,2,3,4]
We present the results of the flood forecasting models developed with the logistic regression (LR), K-nearest neighbors (KNN), random forest (RF), naive Bayes (NB), and multi-layer perceptron (MLP) techniques, and for lead times of 1, 4, 6, 8, and 12 h
We developed evaluated five conditions different flood early warning systems (FEWSs) relying on the most since these magnitudes can beand related to normal of common
Summary
Flooding is the most common natural hazard and results worldwide in the most damaging disasters [1,2,3,4]. Recent studies associate the increasing frequency and severity of flood events with a change in land use (e.g., deforestation and urbanization) and climate [2,5,6,7]. This holds for the tropical Andes region, where complex hydro-meteorological conditions result in the occurrence of intense and patchy rainfall events [8,9,10]. A key for building resilience to short-rain floods is to anticipate in a timely way the event, in order to gain time for better preparedness. In this study special attention is given to flash-floods, which are floods that develop less than 6 h after a heavy rainfall with little or no forecast lead time [14]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have