Project Inferno is a data processing and visualization pipeline that uses NASA’s publicly available satellite remote sensing datasets to analyze wildfire extent and severity. By leveraging MODIS and VIIRS sensor data from FIRMS and EarthData Search, the pipeline extracts spectral band information, applies quality masks, and computes indices such as NBR, dNBR, and RdNBR to generate detailed burn severity maps.

• Fire mapping with NASA satellite remote sensing data
• Severity analysis via algorithms calculating NBR, dNBR, and RdNBR
• Visualizations showcasing fire dynamics

Read the Article

Project Showcase

Technologies Used

• Python for data processing and analysis
• Streamlit for interactive web apps
• GDAL and Rasterio for geospatial data handling
• Geopandas for spatial data manipulation
• PyHDF for HDF file handling
• Plotly for interactive visualizations
• Numpy for numerical operations and vectorization with ND arrays
• Pandas for data manipulation and analysis
• autopep8 for code formatting and PEP8 compliance

Data Sources

• NASA FIRMS — fire mapping data
• NASA EarthData Search — HDF files

Instructions and Bootstrapping

Activate Virtual Environment using conda

conda init
conda create --name inferno
conda install -c conda-forge gdal libgdal-hdf python-kaleido
conda activate inferno

Installation using pip

pip install streamlit plotly numpy pyhdf
pip install GDAL osgeo gdal kaleido

Project File Outline

1. Streamlit Application

   • config.py: application configuration
   • app.py: main application logic
   • pages/*.py: individual UI components
   • firms.py: NASA FIRMS API integration

2. Data Processing Scripts

   • hdf_extraction.py: extract HDF metadata
   • MCD64A1_date_extraction.py: extract burn date & uncertainty
   • data_masking.py: apply bitmask to spectral bands
   • dnbr_rdnbr_calc.py: compute spectral indices & create char

Usage

chmod +x run_inferno_pipeline.sh
./run_inferno_pipeline.sh

WARNING!: this could take up to 30 minutes to run, depending on the specs of you system

Sensor Synopsis

MODIS

• Platforms: Terra and Aqua
• Spatial: 250 m, 500 m, 1000 m
• Spectral: 36 bands (0.4 µm – 14.4 µm)
• Temporal: 1–2 days

VIIRS

• Platforms: NOAA 20, NOAA 21, Suomi NPP
• Spatial: 375 m, 750 m
• Spectral: 22 bands (0.402 µm – 12.49 µm)
• Temporal: Daily

Data from Sensor Configurations

Shapefile Files (GEOJSON)

• Eaton Perimeter: datasets/Eaton_Perimeter_20250121.geojson
• Palisades Perimeter: datasets/Palisades_Perimeter_20250121.geojson

CSV Files

• MODIS: firms_data/MODIS_C61.csv — wide-area fire detection
• VIIRS J1: firms_data/J1_VIIRS_C2.csv — 375m, NOAA-20
• VIIRS J2: firms_data/J2_VIIRS_C2.csv — 375m, NOAA-21
• VIIRS Suomi: firms_data/SUOMI_VIIRS_C2.csv — 375m, Suomi NPP
• LANDSAT: firms_data/LANDSAT.csv — 30m, post-burn

HDF Files

• MCD64A1A: Burned Area Monthly L3 Global 500m
• MOD09A1: Terra Vegetation Indices 16-Day L3 Global 500m
• MYD09A1: Aqua Vegetation Indices 16-Day L3 Global 500m

Cleaned Bands

• MODIS: data_processing/MOD09A1/cleaned_bands/
• MYD09A1: data_processing/MYD09A1/cleaned_bands/

Visualizations

• data_processing/visualizations/dnbr_visualization.html: dNBR fire severity
• data_processing/visualizations/RdNBR_mod_over_time.html: RdNBR over time (MODIS)
• data_processing/visualizations/RdNBR_myd_over_time.html: RdNBR over time (MYD09A1)
• data_processing/visualizations/dnbr_mod_over_time.html: dNBR over time (MODIS)
• data_processing/visualizations/dnbr_myd_over_time.html: dNBR over time (MYD09A1)
• data_processing/visualizations/dnbr_aggregated.html: Aggregated dNBR

Future Features

• Add traceability to the pipeline (to keep track of data lineage): look into the trace library
• Add more visualizations for fire dynamics (esp for dNBR and RdNBR)