Title: Analysis of flood severity using intelligent deep networks and sentinel image for the Kerala region

Authors: Supriya Kamoji; Mukesh Kalla

Addresses: Department of Computer Science and Engineering, Sir Padampat Singhania University, Bhatewar, Udaipur, India ' Department of Computer Science and Engineering, Sir Padampat Singhania University, Bhatewar, Udaipur, India

Abstract: Accurate flood prediction and the classification of severity levels are vital for assessing the impact of floods and ensuring the safety of affected populations. Despite introducing numerous systems, many existing models require significant time to generate prediction results. Intelligent techniques like neural networks have been proposed to enhance prediction accuracy to address this. However, the complexity of data sources, such as satellite or sentinel data, poses challenges due to their unstructured nature and noisy attributes. This study introduces a pioneering approach known as the virtual bee-based recurrent model (VBRM) for flood prediction and severity classification. Specifically, the model categorises floods into low, medium, and high severity levels. The initial step involves gathering sentinel data related to the floods in Ernakulum, Kerala. This data is then used to train the VBRM, followed by a pre-processing stage that effectively filters out irrelevant features and noise. Subsequently, the refined data is fed into the classification layer, where the model extracts pertinent features and determines the severity level of floods. The ultimate objective is to achieve high prediction accuracy with minimal errors. Various performance metrics are employed and a comparative analysis is conducted against other existing models to evaluate the model's performance.

Keywords: deep networks; feature extraction; flood severity classification; optimisation; sentinel data; virtual bee-based recurrent model; VBRM.

DOI: 10.1504/IJHST.2025.143098

International Journal of Hydrology Science and Technology, 2025 Vol.19 No.1, pp.1 - 23

Received: 27 Mar 2023
Accepted: 30 Aug 2023
Published online: 03 Dec 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article