Using artificial neural networks for predicting flood events in Artvin, Türkiye

Olgu Aydın; Nussaibah B. Raja

Makale detayları

Using artificial neural networks for predicting flood events in Artvin, Türkiye

Yazarlar:

Olgu Aydın | Nussaibah B. Raja

Yayıncı:

Murat YAKAR

Ana Başlıklar:

Geographical Information Systems (GIS) in Planning

Dergi:

Turkish Journal of Engineering Vol.9 Issue 2, pp.189-201

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001339335

pISSN:

2587-1366

Anahtar kelimeler:

Özet:

 Flood risks in Artvin, Turkey, have become a critical concern due to the long-term economic, social, cultural, and environmental damages caused by flood-related disasters. Given this, it is essential to utilize reliable methods for flood prediction. Artificial Neural Networks ANNs, adept at reacting to rapid changes in rainfall, are employed as a machine learning approach to provide valuable flood information for urban areas. This study aims to develop an accurate and timely flood prediction model for Artvin using daily average rainfall data from 58 weather stations between 2009 and 2016. Flow values for various locations in Artvin Ardanuç, Arhavi, Artvin, Borçka, Hopa, Murgul, Şavşat, Yusufeli are calculated using the rational method. ANNs were trained with daily rainfall data and consecutive rainfall inputs from 1 to 7 hours to predict flow values. The models performance, with 75 of the data used for training and 25 for validation, showed an error ratio of 0.225 and high prediction accuracy for flow values, exceeding 20 m³h in most locations except Hopa. The R² results for 17 hours indicated high performance 0.643-0.725, suggesting effective warning times of 35 hours for Artvin. The study also highlights the increasing necessity for flood management strategies in the Eastern Black Sea Region, particularly Artvin, which has experienced severe flash floods and significant flooding events since 2017. The regions frequent and intense rainfall, exacerbated by global climate change, underscores the urgent need for robust monitoring, early warning systems, and comprehensive flood mitigation plans to address drainage, land management, and safeguard infrastructure. Effective warning systems that provide real-time estimates of rainfall and flow are crucial for timely preventive measures.

 Flood risks in Artvin, Turkey, have become a critical concern due to the long-term economic, social, cultural, and environmental damages caused by flood-related disasters. Given this, it is essential to utilize reliable methods for flood prediction. Artificial Neural Networks ANNs, adept at reacting to rapid changes in rainfall, are employed as a machine learning approach to provide valuable flood information for urban areas. This study aims to develop an accurate and timely flood prediction model for Artvin using daily average rainfall data from 58 weather stations between 2009 and 2016. Flow values for various locations in Artvin Ardanuç, Arhavi, Artvin, Borçka, Hopa, Murgul, Şavşat, Yusufeli are calculated using the rational method. ANNs were trained with daily rainfall data and consecutive rainfall inputs from 1 to 7 hours to predict flow values. The models performance, with 75 of the data used for training and 25 for validation, showed an error ratio of 0.225 and high prediction accuracy for flow values, exceeding 20 m³h in most locations except Hopa. The R² results for 17 hours indicated high performance 0.643-0.725, suggesting effective warning times of 35 hours for Artvin. The study also highlights the increasing necessity for flood management strategies in the Eastern Black Sea Region, particularly Artvin, which has experienced severe flash floods and significant flooding events since 2017. The regions frequent and intense rainfall, exacerbated by global climate change, underscores the urgent need for robust monitoring, early warning systems, and comprehensive flood mitigation plans to address drainage, land management, and safeguard infrastructure. Effective warning systems that provide real-time estimates of rainfall and flow are crucial for timely preventive measures.

 Flood risks in Artvin, Turkey, have become a critical concern due to the long-term economic, social, cultural, and environmental damages caused by flood-related disasters. Given this, it is essential to utilize reliable methods for flood prediction. Artificial Neural Networks ANNs, adept at reacting to rapid changes in rainfall, are employed as a machine learning approach to provide valuable flood information for urban areas. This study aims to develop an accurate and timely flood prediction model for Artvin using daily average rainfall data from 58 weather stations between 2009 and 2016. Flow values for various locations in Artvin Ardanuç, Arhavi, Artvin, Borçka, Hopa, Murgul, Şavşat, Yusufeli are calculated using the rational method. ANNs were trained with daily rainfall data and consecutive rainfall inputs from 1 to 7 hours to predict flow values. The models performance, with 75 of the data used for training and 25 for validation, showed an error ratio of 0.225 and high prediction accuracy for flow values, exceeding 20 m³h in most locations except Hopa. The R² results for 17 hours indicated high performance 0.643-0.725, suggesting effective warning times of 35 hours for Artvin. The study also highlights the increasing necessity for flood management strategies in the Eastern Black Sea Region, particularly Artvin, which has experienced severe flash floods and significant flooding events since 2017. The regions frequent and intense rainfall, exacerbated by global climate change, underscores the urgent need for robust monitoring, early warning systems, and comprehensive flood mitigation plans to address drainage, land management, and safeguard infrastructure. Effective warning systems that provide real-time estimates of rainfall and flow are crucial for timely preventive measures.

 Flood risks in Artvin, Turkey, have become a critical concern due to the long-term economic, social, cultural, and environmental damages caused by flood-related disasters. Given this, it is essential to utilize reliable methods for flood prediction. Artificial Neural Networks ANNs, adept at reacting to rapid changes in rainfall, are employed as a machine learning approach to provide valuable flood information for urban areas. This study aims to develop an accurate and timely flood prediction model for Artvin using daily average rainfall data from 58 weather stations between 2009 and 2016. Flow values for various locations in Artvin Ardanuç, Arhavi, Artvin, Borçka, Hopa, Murgul, Şavşat, Yusufeli are calculated using the rational method. ANNs were trained with daily rainfall data and consecutive rainfall inputs from 1 to 7 hours to predict flow values. The models performance, with 75 of the data used for training and 25 for validation, showed an error ratio of 0.225 and high prediction accuracy for flow values, exceeding 20 m³h in most locations except Hopa. The R² results for 17 hours indicated high performance 0.643-0.725, suggesting effective warning times of 35 hours for Artvin. The study also highlights the increasing necessity for flood management strategies in the Eastern Black Sea Region, particularly Artvin, which has experienced severe flash floods and significant flooding events since 2017. The regions frequent and intense rainfall, exacerbated by global climate change, underscores the urgent need for robust monitoring, early warning systems, and comprehensive flood mitigation plans to address drainage, land management, and safeguard infrastructure. Effective warning systems that provide real-time estimates of rainfall and flow are crucial for timely preventive measures.

Makale detayları

Using artificial neural networks for predicting flood events in Artvin, Türkiye

Yazarlar:

Olgu Aydın | Nussaibah B. Raja

Yayıncı:

Murat YAKAR

Ana Başlıklar:

Geographical Information Systems (GIS) in Planning

Dergi:

Turkish Journal of Engineering Vol.9 Issue 2, pp.189-201

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001339335

pISSN:

2587-1366

Rainfall

Flood forecasting

Artificial Neural Network (ANN)

Early warning systems

Artvin

Sayı

Makale detayları

Using artificial neural networks for predicting flood events in Artvin, Türkiye

Yazarlar: Olgu Aydın | Nussaibah B. Raja

Yayıncı: Murat YAKAR

Ana Başlıklar: Geographical Information Systems (GIS) in Planning

Dergi: Turkish Journal of Engineering Vol.9 Issue 2, pp.189-201

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: İngilizce

Document ID (Org No.): 2001339335

pISSN: 2587-1366

Rainfall

Flood forecasting

Artificial Neural Network (ANN)

Early warning systems

Artvin

Sayı

Yazarlar:

Olgu Aydın | Nussaibah B. Raja

Yayıncı:

Murat YAKAR

Ana Başlıklar:

Geographical Information Systems (GIS) in Planning

Dergi:

Turkish Journal of Engineering Vol.9 Issue 2, pp.189-201

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001339335

pISSN:

2587-1366