Surface water body extraction using Landsat 8 images and different forms of physical models

Abstract EN

Floods are natural disasters or phenomena that often occur in different areas of Thailand.

During August 2017, Kalasin province, which is located in the central north-eastern area of Thailand, was affected by tropical storm Sonca, resulting in heavy rainfall and floods.

These floods had a tremendous impact on the way of life of the people in the area, as well as resulting in damage to agricultural areas.

The objectives of this research are to propose a method of surface water body extraction from Landsat 8 data, together with four different forms of physical models: 1) the Normalised Difference Vegetation Index, 2) the Normalised Difference Water Index, 3) the Modified Normalised Difference Water Index, and 4) the Water Ratio Index (WRI).

The result of the research found that the data analysis from satellite Landsat 8, together with the WRI physical model, was the most reliable method.

The overall classification accuracy was equivalent to 94.44%, and the kappa statistic was equivalent to 0.8154.

In addition, this method could also efficiently specify the water area from the land area.