Inundations in the Sri Lanka: monitoring and analysis from MODIS [Moderate Resolution Imaging Spectroradiometer] and ALOS [Advanced Land Observing Satellite] instrument

Sri Lanka is facing severe flood events during monsoon rainfall in each year all over the country. The rapid development of remote sensing and widely available satellite images can be used effectively to map the flood inundation in past years. This study is focused on the mapping of flood inundation together with flood recurrent based on both optical (MODIS) and microwave (ALOS/PALSAR) satellite images. In the first stage MODIS images with spatial resolution of 500m and temporal interval of eight day was used to map flood recurrent areas for risk assessment using images from 2000 to 2011. In the second state 16 satellite images from ALOS PALSAR images between 2006 and 2011 was analyzed by using pixel threshold value to map the flooded and non-flooded areas. The flood recurrent products from both MODIS and PALSAR images were generated to represent the repetition of flood inundated areas. The analysis of the results indicated that the PALSAR image based flood inundation mapping is much accurate and useful in the context of spatial variability than the temporal variability. The accurate land-cover map is also important to assess the flood damages and evaluate the future development and the cultivation planning. But there is no such an accurate and detailed land-cove map available for Sri Lanka to assess the flood damages. Thus, this study was focused on the preparation of land-cover map with GIS and RS approach. The land-cover classification was carried out by image fusion of optical (LANDSAT) and microwave (ALOS/PALSAR) under High Pass Filtering (HPF) technique. Unsupervised image classification method was used to classify the fused image in to different land-cover classes. Accuracy assessment of land-cover classification was conducted using existing ground truth information and Google Earth with as resulted in the overall accuracy as 71.16% and the Kappa statistics as 62.83%.