Runoff modeling for a small catchment using artificial neural network
Rainfall-runoff modeling is one of the most studied topics in hydrology. Various types of intelligent computing tools are developed and proven to be efficient in modeling complex hydrological model, such as Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), and Storm Wat...
| Glavni autor: | |
|---|---|
| Daljnji autori: | |
| Format: | Final Year Project (FYP) |
| Jezik: | English |
| Izdano: |
2009
|
| Teme: | |
| Online pristup: | http://hdl.handle.net/10356/15903 |
| _version_ | 1826118011428274176 |
|---|---|
| author | Tan, Wei Chuong. |
| author2 | Chua Hock Chye Lloyd |
| author_facet | Chua Hock Chye Lloyd Tan, Wei Chuong. |
| author_sort | Tan, Wei Chuong. |
| collection | NTU |
| description | Rainfall-runoff modeling is one of the most studied topics in hydrology. Various types of intelligent computing tools are developed and proven to be efficient in modeling complex hydrological model, such as Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), and Storm Water Modeling Management (SWMM). This study demonstrates the application of ANN in modeling the runoff for small sub-catchment CP1, Kranji Singapore. Sixty-six rainfall events were used to calibrate the network model and then determine the performance of ANN. In this project, Feed Forward Neural Network with Back-propagation training algorithm was chosen as modeling tool. Performance of ANN was investigated using different set of events as input and divided into four types. These four ANN rainfall-runoff models are also compared favorably with ANFIS and SWMM models. It is observed that Type 3 ANN model with less input data can reduce the size of model structure and thus Type 3 has better predictive capability and outperforms other 3 models. Comparison between ANN models and ANFIS with SWMM was conducted and the result shows that SWMM gives the best performance among them and ANFIS is comparable to SWMM. Analysis was done and the importance of the events used in training phase was noticed. The shape and type of rainfall-runoff hydrograph indirectly affect the performance of network model. There are some events containing the discontinuous hydrograph are difficult to be calibrated even by SWMM. It is suggested that those events can be excluded from the modeling process. |
| first_indexed | 2024-10-01T04:36:46Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/15903 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T04:36:46Z |
| publishDate | 2009 |
| record_format | dspace |
| spelling | ntu-10356/159032023-03-03T16:58:13Z Runoff modeling for a small catchment using artificial neural network Tan, Wei Chuong. Chua Hock Chye Lloyd School of Civil and Environmental Engineering DRNTU::Engineering::Civil engineering::Water resources Rainfall-runoff modeling is one of the most studied topics in hydrology. Various types of intelligent computing tools are developed and proven to be efficient in modeling complex hydrological model, such as Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), and Storm Water Modeling Management (SWMM). This study demonstrates the application of ANN in modeling the runoff for small sub-catchment CP1, Kranji Singapore. Sixty-six rainfall events were used to calibrate the network model and then determine the performance of ANN. In this project, Feed Forward Neural Network with Back-propagation training algorithm was chosen as modeling tool. Performance of ANN was investigated using different set of events as input and divided into four types. These four ANN rainfall-runoff models are also compared favorably with ANFIS and SWMM models. It is observed that Type 3 ANN model with less input data can reduce the size of model structure and thus Type 3 has better predictive capability and outperforms other 3 models. Comparison between ANN models and ANFIS with SWMM was conducted and the result shows that SWMM gives the best performance among them and ANFIS is comparable to SWMM. Analysis was done and the importance of the events used in training phase was noticed. The shape and type of rainfall-runoff hydrograph indirectly affect the performance of network model. There are some events containing the discontinuous hydrograph are difficult to be calibrated even by SWMM. It is suggested that those events can be excluded from the modeling process. Bachelor of Engineering (Environmental Engineering) 2009-05-19T01:01:00Z 2009-05-19T01:01:00Z 2009 2009 Final Year Project (FYP) http://hdl.handle.net/10356/15903 en Nanyang Technological University 55 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Civil engineering::Water resources Tan, Wei Chuong. Runoff modeling for a small catchment using artificial neural network |
| title | Runoff modeling for a small catchment using artificial neural network |
| title_full | Runoff modeling for a small catchment using artificial neural network |
| title_fullStr | Runoff modeling for a small catchment using artificial neural network |
| title_full_unstemmed | Runoff modeling for a small catchment using artificial neural network |
| title_short | Runoff modeling for a small catchment using artificial neural network |
| title_sort | runoff modeling for a small catchment using artificial neural network |
| topic | DRNTU::Engineering::Civil engineering::Water resources |
| url | http://hdl.handle.net/10356/15903 |
| work_keys_str_mv | AT tanweichuong runoffmodelingforasmallcatchmentusingartificialneuralnetwork |