Use the model

You can use this model in the Detect Objects Using Deep Learning tool available in the Image Analyst toolbox in ArcGIS Pro.

Recommended imagery configuration

The recommended imagery configuration is as follows:

• Resolution—10 meters.
• Dynamic range—8 bit.
• Bands—Three-band Sentinel-1 C band SAR GRD VV polarization band raster.

Extract oil spills

To extract oil spills from the imagery, complete the following steps:

1. Open ArcGIS Pro and create an ArcGIS Pro project.
2. Browse to Tools on the Analysis tab.

   Note:

   You can download the imagery from Copernicus Open Access Hub, Sentinel Hub, or ASF portal.

   

3. Click the Toolboxes tab in the Geoprocessing pane, select Image Analyst Tools, and browse to the Despeckle tool under Synthetic Aperture Radar.

   

   1. Click the Browse button to select the input in the Input Radar Data parameter.

      

      Go the directory containing radar data, select the manifest.safe file and click OK.

      

   2. Select VV band in the Polarization Bands parameter. Click Run.

      The output layer is added to the map.

4. Click the Toolboxes tab in the Geoprocessing pane, select Image Analyst Tools, and browse to the Log10 tool under Math.

   

   1. Add the output from the Despeckle geoprocessing tool as an input file into the Input raster or constant value option and click Run.

      

      The output layer is added to the map.

5. To remove the land area from the tile, in the Raster Functions window, search for Clip.

   

   1. In the Clip Properties pane, add the output from the Log10 function in the Raster parameter and click Run.

      

      The output layer is added to the map.

6. In the Raster Functions window, search for and click Composite Bands.

   

   1. On the Parameters tab of the function, in the Rasters parameter, add the output of the Clip function three times to create a three-band raster.

      

   2. Click the General tab of the function, in the Output Pixel Type parameter, choose 8 Bit Unsigned, and click Create new layer.

      

      The output layer is added to the map.

7. Zoom to an area of interest.

   

8. Click the Toolboxes tab in the Geoprocessing pane, select Image Analyst Tools, and browse to Detect Objects Using Deep Learning under Deep Learning.

   

   On the Parameters tab, set the variables as follows:

   1. For Input Raster, select Composite Bands.
   2. For Output Detected Objects, select the feature layer representing oil spills.
   3. Optionally, for Model Definition, select the pretrained or fine-tuned model .dlpk file.
   4. Optionally, for Arguments, change the values of the arguments if required.

      

   Set the variables on the Environments tab as follows:

   1. For Processing Extent, select Current Display Extent or any other option from the drop-down menu.
   2. For Cell Size, change if required.

      Note:

      The expected SAR image resolution is 10 meters.

   3. For Processor Type, select CPU or GPU as needed. Select GPU, if available, and set GPU ID to the GPU to be used.

      

9. Click Run.

   

   The output layer is added to the map.

10. Convert multiple overlapping polygons into a single polygon in cases where multiple polygons must be generated over a single oil spill. Use the Pairwise Dissolve tool to convert them.

    

11. Click the Toolboxes tab in the Geoprocessing pane, expand Analysis Tools, and browse to the Pairwise Dissolve tool.

    

    Set the variables on the Parameters tab as follows:
    1. For Input Features, select the detected oil spill feature layer.
    2. For Output Feature Class, set the output feature class that contains the dissolved oil spills.

       

    3. Click Run.

       The output layer is added to the map.