Computer model of Tsunami vulnerability using machine learning and multispectral satellite imagery

Sri Yulianto Joko Prasetyo, Wiwin Sulistyo, Prihanto Ngesti Basuki, Kristoko Dwi Hartomo, Bistok Hasiholan


This research aims to develop a tsunami vulnerability assessment model on land use and land cover using information on NDVI, NDWI, MDWI, MSAVI, and NDBI extracted from sentinel 2 A and ASTER satellite images. The optimization model using algorithms LASSO and linear regression. The validation test is MSE, ME, RMSE and MAE which show that the linear regression has a higher accuracy than the LASSO. The NDWI interpolation values are 0.00 - (-0.35) and MNDWI interpolation values are 0.00 - (-0.40) which are interpreted as the presence of water surfaces along a coast. MSAVI are values (-0.20) - (-0.35) which are interpreted as the presence of no vegetation. The NDBI interpolation values are values 0.15-0.20 which are interpreted as the presence of built-up lands with social and economic activities. While the NDVI interpolation values are 0.20-0.30 which are interpreted as the presence of vegetation densities, biomass growths from the photosynthesis process, and moderate to low levels of vegetation health. The digital elevation model ASTER analysis shows that all areas with high socioeconomic activities, low NDVI, high NDWI/MDWI, high MSAVI and high NDBI are in areas with low elevation (<10 meters) so they have a high vulnerability to tsunami waves.


Machine learning; Remote sensing; Spatial interpolation; Tsunami; Vegetation indices

Full Text:




  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Stats