A Hybrid Deep Learning and Geoelectric Sensing Measurement over Bengkulu Flood

One of the most severe natural catastrophes, floods result in extensive devastation and a high death toll on a worldwide scale. Effective disaster management and disaster mitigation depend on early flood prediction and identification. In order to improve the accuracy of flood trigger prediction, this research introduces a unique method that combines the strength of Deep Convolutional Neural Networks and geoelectric data. Then model successfully developed with the architecture with 3 layers, dropout 0, 2×2 pooling size and 4 in fold size. The model use to investigate the land use change over two decades over Bengkulu area and show the pattern of change and integrated the geoelectric measurement to see the affect of the condition and showed that 705, 33Ha and 881,96Ha of the area already change into built-up and palm oil plantation respectively while forest and bare land show reduction value of -578, 17Ha and - 729, 65Ha respectively. The accuracy score obtained in this study was 0.922 and the kappa score was 0.990. Furthermore this research showed the geoelectrical measurement showed that the porosity over the area over the flooding disaster (Rawa Makmur district) have been reduced in the depth of 0 5m due to the change.