- To save more lives and prevent infrastructural damage during natural disasters, timely search and rescue missions are needed.
- In this commentary, scientists from the Indian Institute for Human Settlements have used freely available images from the satellite Sentinel-1, in combination with the open-source tool Google Earth Engine, to develop maps of the recent Assam and Bihar floods.
- Using this as an illustrative example, they call for using such technologies to aid search and rescue missions during disasters, an important element of disaster preparedness.
A few weeks ago, the states of Assam and Bihar were reeling under floods. Over 200 people were reported dead, with at least 10 million (one crore) of the states’ residents estimated to have been displaced. To save more lives and prevent further infrastructural damage, search and rescue missions during such disasters need to be effective, and more importantly, need to be rapid.
The answer to this may lie in space.
Open-source access to satellite images and new technologies to process these images have been a significant breakthrough to help document the true extent of flooding. Getting this information in time is key to plan and conduct evacuation missions, response operations and damage assessments.
The European Space Agency (ESA)’s Sentinel-1 mission and the web-based Google Earth Engine (GEE) platform are two recent developments that have helped timely capture and analysis of satellite information.
A research team from the Indian Institute for Human Settlements (IIHS) used this combination (Sentinel and GEE) to come up with an illustrative example of how such mapping can be used in the future to help in rescue missions, through accurate mapping of flood extents.
Sentinel-1 has two satellites in orbit and has a “revisit time” of six days. This means the data is available for monitoring of affected areas every six days; it is also available for processing for free within a day of being recorded.
These frequent revisits make this method especially ideal for flood monitoring. According to the ESA, over 75 percent of natural disasters that occur worldwide involve flooding.
Once GEE was launched in 2015, with a high capacity to store and process vast amounts of raw image data, processing of satellite images has become more accessible. The GEE platform has access to the complete repository of open data from the United States Geological Survey (USGS), ESA, and other such space agencies. It reduces the time taken to download, process, and analyse from weeks or months to hours. The platform is free and open to all; GEE is free for research, education and non-profit use, and uses commercial licenses for commercial applications.
Such maps are timely and can be more accurate and cost-effective than older ways of mapping flood information such as physical surveys, crowdsourcing information, or even using drones. All the other methods require either vast amounts of human resources, or are expensive, or more importantly, dependent on favourable weather conditions. For flooding, in particular, monsoon conditions such as cloud cover can be a challenge for capturing satellite images depending on the sensor and method used. Drones, for instance, are dependent on weather conditions.
The Sentinel-1 mission, which was launched in 2014 by ESA, helped to overcome such limitations. Sentinel-1 uses the microwave range of the electromagnetic spectrum, enabling it to sense Earth’s surface beneath clouds.
It also uses an active sensor — it emits its own radiation and doesn’t rely on sunlight, and can, therefore, be used day and night, independent of weather conditions. This is as opposed to optical satellites that have passive sensors that have to rely on the sun to be able to acquire data.
The combination of using Sentinel and GEE helps resolve some of the processing issues faced in earlier analysis tools. For example, the hiccup in using the Sentinel-1 is the size of the data and required computational power. A single GRD (ground range detected) scene is approximately one gigabyte (GB) in file size, and after processing, it generates an output of a couple of more GBs of data. The time required for downloading these large datasets and the time for image processing is dependent on the network and system configuration, which could be different across different kinds of users.
Once downloaded, multiple pre-processing steps are performed to convert the raw images to analysis-ready data, after which the processing can be done using ESA’s open-source application- Sentinel Application Platform (SNAP) software. In this context, a minimum of five to six scenes are required to cover the state of Assam and Bihar for one time period.
But in case of GEE, the engine pre-processes all the Sentinel-1 images, soon after they are available on the ESA’s portal. As GEE is a browser-based analysis platform, data can be processed easily with a few lines of code, and can be repeated when the new images are available, without having to download them individually.
To map the extent of flooding in Assam and Bihar, a research team from the Indian Institute for Human Settlements (IIHS) used GEE to process Sentinel imagery from June and July 2019. The flood extents were overlaid on Open Street Maps, an open mapping data set.
Read more: Wildlife and people work together during Assam’s annual tryst with floods
Geospatial tools such as these can help overlay different kinds of data sets together, such as height or terrain, and then extract only the information deemed useful by comparing these datasets.
For the Assam floods, Sentinel-1 images were analysed using GEE in two parts. Images from 8th – 15th June were used for mapping the pre-flood water extent, and the images from 4th to 14th July to map the water extent during the floods.
In the case of Bihar, images from 30th June to 8th July were used to map the pre-flood water extent, and from 12th and 17th July to map the water extent during the floods.
Each image from the satellite covers areas smaller than each state (about 171 x 252 square km); multiple images were combined to cover the entire extent of each state.
The following images show the flooded regions before and during the event.
Such images, when processed in real-time, can help visualise the impact of flooding on settlements and infrastructure, at scale, and can provide relief teams with accurate, up-to-date information.
In February 2019, when the coastal city of Townsville in Queensland, Australia, was flooded, a team of researchers from the University of New South Wales in Sydney used the ESA data and processed them to put together an accurate and comprehensive flood map in less than an hour. Besides this, they also made the data available on social media.
During the Bihar and Assam flooding, eight countries including the USA, China and Russia reportedly shared satellite data with India to improve rescue operations.
There is a need for more such open data from other radar imaging satellites, such as RISAT -2 by ISRO. If these are integrated with platforms like GEE and used more widely and in good time, they can become vital for improved disaster relief operations that can possibly mitigate the large extent of damages to life and property caused during disasters.
Other widely adapted applications of the Sentinel-1 data are land subsidence/ deformation and landslides mapping, maritime monitoring and agriculture health and yield predictions. Further advancement in the field of remote sensing for high resolution, free images should be encouraged.
Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., & Moore, R. (2017). Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031
Pratyush Tripathy is part of the Geospatial Lab at the Indian Institute for Human Settlements. His research focuses on understanding urban spatial inequalities using remote sensing techniques.
Teja Malladi leads the Geospatial Lab and is part of the Risk Lab at the Indian Institute for Human Settlements. He works in the fields of natural hazard and risk and vulnerability assessment using remote sensing and GIS.