Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p.
The rainfall-runoff relationship is the subject of many studies because of its importance in the implementation of many development projects. The scientific community, in order to cope with water problems such as floods and droughts, used different models. But these models are usually faced with the...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Physio-Géo
|
| Series: | Physio-Géo |
| Subjects: | |
| Online Access: | https://journals.openedition.org/physio-geo/940 |
| _version_ | 1797314380059115520 |
|---|---|
| author | Yao Blaise Koffi |
| author_facet | Yao Blaise Koffi |
| author_sort | Yao Blaise Koffi |
| collection | DOAJ |
| description | The rainfall-runoff relationship is the subject of many studies because of its importance in the implementation of many development projects. The scientific community, in order to cope with water problems such as floods and droughts, used different models. But these models are usually faced with the non-linearity of the rainfall-runoff relationship. In the case of the Bandama Blanc purpose of this study, this non-linearity is enhanced by the presence of several agro-pastoral dams located in the northern part of the study area and using the waters of this river. This thesis therefore deals with the modeling of flows of Bandama Blanc hydrometric stations (Bada Marabadiassa, Tortiya and Bou) using neural networks already experienced in the context of non-linear relationship. It plans to provide more robust tools to African hydrologist in the simulation and forecasting of river flows. To achieve this goal, two Multilayer Perceptrons trained with the backpropagation algorithm of error have been built. The first model was used only in simulation and the second in simulation and prediction. The conceptual model GR2M was used to validate the results obtained with neural networks. An extensive database climate (rainfall and temperature) and river monthly flow was used in this study. The results obtained are very satisfactory and well above those obtained with the overall conceptual model GR2M. Indeed, neural networks are able to explain more than 70% of the variation in rates, with Pearson correlation coefficients exceeding 0.80. However, these models have difficulty to simulate and predict extremes flow probably because of the reduced number of data at our disposal and separation of bases calibration and validation. |
| first_indexed | 2024-03-08T02:46:15Z |
| format | Article |
| id | doaj.art-66c068a713684322ac82c2f06def7152 |
| institution | Directory Open Access Journal |
| issn | 1958-573X |
| language | English |
| last_indexed | 2024-03-08T02:46:15Z |
| publisher | Physio-Géo |
| record_format | Article |
| series | Physio-Géo |
| spelling | doaj.art-66c068a713684322ac82c2f06def71522024-02-13T13:23:55ZengPhysio-GéoPhysio-Géo1958-573X31310.4000/physio-geo.940Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p.Yao Blaise KoffiThe rainfall-runoff relationship is the subject of many studies because of its importance in the implementation of many development projects. The scientific community, in order to cope with water problems such as floods and droughts, used different models. But these models are usually faced with the non-linearity of the rainfall-runoff relationship. In the case of the Bandama Blanc purpose of this study, this non-linearity is enhanced by the presence of several agro-pastoral dams located in the northern part of the study area and using the waters of this river. This thesis therefore deals with the modeling of flows of Bandama Blanc hydrometric stations (Bada Marabadiassa, Tortiya and Bou) using neural networks already experienced in the context of non-linear relationship. It plans to provide more robust tools to African hydrologist in the simulation and forecasting of river flows. To achieve this goal, two Multilayer Perceptrons trained with the backpropagation algorithm of error have been built. The first model was used only in simulation and the second in simulation and prediction. The conceptual model GR2M was used to validate the results obtained with neural networks. An extensive database climate (rainfall and temperature) and river monthly flow was used in this study. The results obtained are very satisfactory and well above those obtained with the overall conceptual model GR2M. Indeed, neural networks are able to explain more than 70% of the variation in rates, with Pearson correlation coefficients exceeding 0.80. However, these models have difficulty to simulate and predict extremes flow probably because of the reduced number of data at our disposal and separation of bases calibration and validation.https://journals.openedition.org/physio-geo/940Artificial Intelligence (AI)Multilayer Perceptron (PMC)global modelslearningvalidation |
| spellingShingle | Yao Blaise Koffi Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. Physio-Géo Artificial Intelligence (AI) Multilayer Perceptron (PMC) global models learning validation |
| title | Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. |
| title_full | Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. |
| title_fullStr | Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. |
| title_full_unstemmed | Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. |
| title_short | Modélisation pluie-débit en région tropicale humide : application des réseaux de neurones sur quatre stations hydrométriques du Bandama Blanc (Bada, Marabadiassa, Tortiya et Bou) situées au Nord de la Côte d'Ivoire. Thèse de l'Université de Cocody (Côte d'Ivoire), 2007, 219 p. |
| title_sort | modelisation pluie debit en region tropicale humide application des reseaux de neurones sur quatre stations hydrometriques du bandama blanc bada marabadiassa tortiya et bou situees au nord de la cote d ivoire these de l universite de cocody cote d ivoire 2007 219 p |
| topic | Artificial Intelligence (AI) Multilayer Perceptron (PMC) global models learning validation |
| url | https://journals.openedition.org/physio-geo/940 |
| work_keys_str_mv | AT yaoblaisekoffi modelisationpluiedebitenregiontropicalehumideapplicationdesreseauxdeneuronessurquatrestationshydrometriquesdubandamablancbadamarabadiassatortiyaetbousitueesaunorddelacotedivoirethesedeluniversitedecocodycotedivoire2007219p |