Topic: Wildfire Applications with SAR
Goal: Create a fire perimeter and burn severity map using radar remote sensing data.
Introduction: The severity, frequency, and area affected by wildfires globally have increased over the past few decades. In the Western US, drier conditions, extreme temperatures, and variable precipitation have increased fire severity. Current wildfire monitoring relies on field survey and optical imagery, which can be deterred by weather conditions. Radar remote sensing is unaffected by these conditions, and thus can help improve existing, manually generated or in-situ based maps of fire perimeters and burn severity.
Datasets:
- Location: /home/jovyan/shared/NASA_Summer_School_2026/wildfires/materials/Data
- Eaton Fire BAER Soil Burn Severity - Burn severity data from post-fire analysis
- Eaton Fire NIFC Perimeter - Fire perimeter from National Interagency Fire Center
- Eaton Fire UAVSAR - Polarimetric SAR (synthetic aperture radar) data from before and after the fire
- Background and Resources - Supporting reference materials and documentation
Scripts and Notebooks:
- Location: /home/jovyan/shared/NASA_Summer_School_2026/wildfires/materials/uavsar-wildfire/notebook
- crop_image.ipynb - Crop raster image by fire perimeter
- preprocessing_segmentation.ipynb - Preprocess image and perform superpixel segmentation
- perimeter_generation.ipynb - Generate fire perimeter from segmented image
- burn_severity_generation.ipynb - Process SAR data and generate burn severity map
- gen_perim_sbs.ipynb - Additional perimeter generation tool
- weighted_inc_merge.ipynb - Merge and weight incidence angle files
Geographic Focus: Eaton Fire event (Southern California) or UAVSAR archive
Project Approach:
Feel free to choose your own wildfire of interest from the UAVSAR archive or use the provided Eaton Fire datasets. The provided UAVSAR HV GeoTIFFs have had RTC (radiometric terrain correction) applied and linear power units converted to decibels. You can begin by subtracting the pre-fire minus post-fire images (equivalent to dividing) and cropping the image to your selected fire(s) as there are several burn scars in this flight line. You may choose to explore generating a SAR-derived fire perimeter, burn severity map, or both. Using the Jupyter notebooks developed by previous students, you can compare other segmentation or clustering methods to improve upon the results.
Analysis Workflows:
Fire Perimeter:
- Use crop_image.ipynb to crop the raster image by the fire perimeter
- Use preprocessing_segmentation.ipynb to preprocess the cropped image and perform superpixel segmentation
- Use perimeter_generation.ipynb to generate the final fire perimeter
Burn Severity:
- Use crop_image.ipynb to crop the raster image by the fire perimeter
- Use burn_severity_generation.ipynb to process the SAR data and generate the burn severity map
Contact Scientist: Karen An (karen.an@jpl.nasa.gov)
Project Server (CMDA): https://hub.jpl-cmda.org