Tutorial 1: Flood Submergence Mapping using SAR imagery on UP42 Platform

Synthetic Aperture Radar (SAR) is an active radar sensing system that can produce high resolution radar images of the surface of the Earth in all weather conditions and during night time. For example, in the event of heavy rains, floods, fog etc., optical satellite sensors are unable to obtain images due to insufficient light and obstruction from clouds. We at Vasundharaa Geo Technologies have developed an algorithm to map the extent of flooded regions, “Flood Mapping using SAR” which is hosted on our partner platform UP42.

The objective of this tutorial is to provide users with step-by-step instructions on working with SAR images on the UP42 platform. In this tutorial, we study the extent of submergence that hit the Erbo Delta located on the east coast of Spain. On 21 January 2020, storm Gloria hit the delta engulfing 3000 hectares of fields.

Figure 1: Animation of the flood. Image Source: Sentinel Hub.

Data and Processing Blocks

For this tutorial we will use the Sentinel-1 GRD AOI clipped datasets available on the platform and the Flood Mapping processing block developed by Vasundharaa Geo Technologies Pvt. Ltd.

Note: It is required to create an account on the UP42 platform for which each user gets 10000 free credits.

Create a Project

Click on the “Start Project” to begin a new project. Input the name and project description and click on “Save”.

Figure 2: Dialog box of creating a project

Figure 3: Project Description dialog box

Create a Workflow

Next, we have to create a workflow to do the analysis. To do this select the project and click on “Create Workflow”.

Figure 4: Dialog box of creating a workflow

Add Blocks for Data and Processing

Once the necessary information in the workflow is added, the next step we have to add the data and processing blocks. For this click on “Add Block” under the data tab and select “Sentinel-1 L1C GRD AOI Clipped” and click on “Add Block”.

Similarly, under the processing block select “Flood Mapping using SAR” and click on “Add Block”.

Figure 5: Add data and processing blocks

Figure 6: Select the “Sentinel-1 L1C GRD AOI clipped” data (left) and “Flood Mapping using SAR” processing (right) blocks

Once both the blocks are added click on “Next”. Now we can configure the workflow by clicking on “Configure Job”.

Figure 7: Configure Job

Selecting Area of Interest and Dates

Now we select the area of interest (AOI) by drawing a polygon on the map. The map can be changed from street view to satellite view.

Figure 8: Input of Area of Interest (AOI) and interval of the event

Since this is a time series analysis, the flood mapping algorithm uses 2 images (pre- and post-event). Next we provide the dates for the analysis. In the “time” tab provide null and in the “time series” provide the dates for which imagery is required. The green box (Figure 8) indicates the interval of the first image (pre-event) and the blue box indicates the interval of the second image (post-event). Set “limit” to 1 which indicates that 1 image is selected for each given interval.

Once the details are input, we can run the “test query”. This is done to make sure there are no errors.

Note: The UP42 platform has a limit of 100 sq.km on selecting the AOI.

Real Run

On successful completion of the dry run (test query), we can do the analysis by selecting “Run as real job”.

Figure 9: Processing as a real job

Results

Once the job is done, the results can be downloaded by clicking on “Download”.

Figure 10: Downloading the results

Figure 11: The output is a binary raster

The result is completely automated and hence is both cost-effective and time efficient as compared to conventional SAR processing methods. This can be further analysed by fusing with geo-spatial data such as land-use, population density etc. We at Vasundharaa did a detailed analysis of the event and mapped the extent of submergence which is as shown in the figure below.

Figure 12: Flood submergence map in the Erbo Delta region on the eastern coast of Spain. The deep red color indicates flooded areas.

The above submergence map when overlayed with the population density of the region gives an estimate of the number of people affected by flood, which in this case was approximately 27826. The affected population is depicted in the Figure 13.

Figure 13: Population density overlayed on the submergence map that gives an estimate of the affected population.

Our other blocks on the platform are:

1. Burnt Area Extraction
2. Urban Estimation
3. Deforestation Mapping
4. Normalized Difference Vegetation Index
5. Enhanced Vegetation Index
6. Moisture Stress Index
7. Normalized Burnt Index
8. Normalized Difference Water Index
9. Soil Adjusted Vegetation Index

For more information and the work we do please visit our website.