October 1, 2025Scientific ReportsOpen Access

An autoencoder driven deep learning geospatial approach to flood vulnerability analysis in the upper and middle basin of river Damodar

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This study uses a CNN-led autoencoder to map flood vulnerability in regions with scarce data, highlighting significant risk factors.

The flood risk zonation map indicates that 92% of the study area is safe, while less than 8% is at moderate to very high flood risk.
Using eleven geospatial layers, the autoencoder's reconstruction performance metrics include MSE, precision, recall, and accuracy.
K-means clustering alongside CNN Autoencoder was employed to classify flood vulnerabilities and produce the risk map for the river basin.
The study emphasizes the critical role of Drainage Density on model outcomes, while factors like Aspect may introduce noise.

Thappitla et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark