Using cloud computing for better flood inundation mapping
The Kerala rains of July-August 2018 caused substantial loss of lives and property and left major cities flooded for days.
Maps showing where flooding may occur or flood inundation maps can help in better flood risk preparedness. Using openly accessible satellite data and a cloud computing platform, an international team has now developed a powerful tool for a near real-time mapping of flood extent. The paper published in PLOS ONE notes the new flood inundation maps showed an accuracy of over 94%. Space-based sensors known as synthetic aperture radar (SAR) have been used widely for monitoring and mapping of flood-water inundation. SAR is capable of acquiring data in all-weather condition, making it useful for mapping and monitoring flood inundation areas.
These sensors operate on the constellation of two SAR satellites belonging to the Copernicus Programme launched by the European Space Agency.
The data from the satellites was utilised on a cloud-based platform known as Google Earth Engine (GEE) for the rapid processing of big data. The GEE also has publicly made available numerous satellite image collections and has functions for image processing and analysis.
The team studied water inundation maps from 2015 and their analysis was clearly able to show the areas submerged underwater in 2018. “Once you have the data, it just takes a few minutes as you can apply machine learning and computer vision techniques to quickly generate the water inundation maps. This can help swiftly deploying the rescue team and rescue operations can be started immediately,” explains Varun Tiwari, Remote Sensing and Geoinformation Analyst from the International Centre for Integrated Mountain Development, an intergovernmental organisation based in Kathmandu, Nepal. He is the first and corresponding author of the work.
The team also analysed the rainfall data from 1981 to 2018 and were able to predict the major reasons behind this flood. “The monsoon season of Kerala has seen an increasing rainfall trend and this has played a major role. This also depicts that more floods are likely to happen in the near future,” adds Tiwari. “Other studies have also pointed out that the flooding event would have not taken place if the capacity of the major six reservoirs would have been 34% more. The 2018 Kerala rains have taught us many lessons and now with these real-time mapping we can be better prepared during the time of the disaster.”
He explains that cloud computing platforms and satellite data are also being used by researchers for landslide prediction, drought forecasting, crop monitoring and avalanche forecasting. “With the availability of ample amount of open-access satellite data and cloud computing platforms we are now enabled to solve real-world problems,” adds Mr. Tiwari.