Predicting Wildfires

This repository was created during my masters program with the goal of using sentinel-2 satalite imagry to create a risk map for a wildfire occuring at a 10m scale based on NDVI. We also attempted to predict a fire occuring near a weather station using historical weather and logistic regression. Although this was a group project, all of the code in this repository was written by me.

Databases:

Due to the sheer amount of data that needed to be processed I created SQL databases to keep track of the data.

• Database Builder L1C & L2A.ipynb
• Database Builder Weather.ipynb

Exploration:

Some notebooks exploring and visualizing the data.

• Exploration and Cleanup of Weather Data.ipynb
• Exploration of Sentinel-2.ipynb or try Exploration of Sentinel-2.ipynb at nbviewer.org
• Logistic Regression.ipynb

Processing and Deep Learning

General Processing

• noaa_weather.py
• process.py
• s2.py
• Atmospheric Correction

Models and model testing

• Training SSIM
• Training Binary Cross Entropy
• Model Evaluation
• Earth Engine Processing/ Prep

Bash Files

• ec2-user-data.sh
• Slurm CPP HPC
• Mount FSx for Lustre