Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador
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, <i>No-alert</i...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-12-01
|
| Series: | Hydrology |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2306-5338/8/4/183 |
| _version_ | 1797504020351287296 |
|---|---|
| author | Paul Muñoz Johanna Orellana-Alvear Jörg Bendix Jan Feyen Rolando Célleri |
| author_facet | Paul Muñoz Johanna Orellana-Alvear Jörg Bendix Jan Feyen Rolando Célleri |
| author_sort | Paul Muñoz |
| collection | DOAJ |
| description | 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, <i>No-alert</i>, <i>Pre-alert</i> and <i>Alert</i> 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 <i>f</i>1-macro (<i>log-loss</i>) 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 <i>Pre-alert</i> and <i>Alert</i> 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. |
| first_indexed | 2024-03-10T03:58:40Z |
| format | Article |
| id | doaj.art-93e554804bf04e3ea641fed975e1d294 |
| institution | Directory Open Access Journal |
| issn | 2306-5338 |
| language | English |
| last_indexed | 2024-03-10T03:58:40Z |
| publishDate | 2021-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Hydrology |
| spelling | doaj.art-93e554804bf04e3ea641fed975e1d2942023-11-23T08:39:46ZengMDPI AGHydrology2306-53382021-12-018418310.3390/hydrology8040183Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of EcuadorPaul Muñoz0Johanna Orellana-Alvear1Jörg Bendix2Jan Feyen3Rolando Célleri4Departamento de Recursos Hídricos y Ciencias Ambientales, Universidad de Cuenca, Cuenca 010150, EcuadorDepartamento de Recursos Hídricos y Ciencias Ambientales, Universidad de Cuenca, Cuenca 010150, EcuadorLaboratory for Climatology and Remote Sensing, Faculty of Geography, University of Marburg, 35032 Marburg, GermanyFaculty of Bioscience Engineering, Catholic University of Leuven, 3001 Leuven, BelgiumDepartamento de Recursos Hídricos y Ciencias Ambientales, Universidad de Cuenca, Cuenca 010150, EcuadorWorldwide, 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, <i>No-alert</i>, <i>Pre-alert</i> and <i>Alert</i> 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 <i>f</i>1-macro (<i>log-loss</i>) 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 <i>Pre-alert</i> and <i>Alert</i> 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.https://www.mdpi.com/2306-5338/8/4/183flood early warningforecastinghydrological extremesmachine learningAndes |
| spellingShingle | Paul Muñoz Johanna Orellana-Alvear Jörg Bendix Jan Feyen Rolando Célleri Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador Hydrology flood early warning forecasting hydrological extremes machine learning Andes |
| title | Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador |
| title_full | Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador |
| title_fullStr | Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador |
| title_full_unstemmed | Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador |
| title_short | Flood Early Warning Systems Using Machine Learning Techniques: The Case of the Tomebamba Catchment at the Southern Andes of Ecuador |
| title_sort | flood early warning systems using machine learning techniques the case of the tomebamba catchment at the southern andes of ecuador |
| topic | flood early warning forecasting hydrological extremes machine learning Andes |
| url | https://www.mdpi.com/2306-5338/8/4/183 |
| work_keys_str_mv | AT paulmunoz floodearlywarningsystemsusingmachinelearningtechniquesthecaseofthetomebambacatchmentatthesouthernandesofecuador AT johannaorellanaalvear floodearlywarningsystemsusingmachinelearningtechniquesthecaseofthetomebambacatchmentatthesouthernandesofecuador AT jorgbendix floodearlywarningsystemsusingmachinelearningtechniquesthecaseofthetomebambacatchmentatthesouthernandesofecuador AT janfeyen floodearlywarningsystemsusingmachinelearningtechniquesthecaseofthetomebambacatchmentatthesouthernandesofecuador AT rolandocelleri floodearlywarningsystemsusingmachinelearningtechniquesthecaseofthetomebambacatchmentatthesouthernandesofecuador |