analyse_ml_fire_projections

Repository to process the netCDF fire projections made here and analyse them. Also includes processing and analysis scripts for processing and analysing the climate and fire weather variables.

Preliminary:

• Ensure you have a Google Earth Engine account linked to a project
• It is strongly recommended to have access to a Google Cloud Storage bucket
• Use of a supercomputer or HPC is encouraged for running the scripts

Order for running scripts:

1. Before beginning the analysis conduct netCDF_processing to convert the files to csv format for more convenient analysis.
   • Run split_netCDF_into_years first to reduce the netCDF file load - it converts the whole time series into single years for each scenario. You can also separate the whole file into North America/Eurasia (for example) as required.
   • Use Check and test shapefiles.ipynb to load and check the shapefiles used in the analysis. Land cover per region can also be found here.
   • Then you can choose to find the time series by ecoregion, geographical region or land_cover class, as required. The scripts all perform the same job: the burned area is found for a specific shapefile over the whole time series and each scenario and saved into individual csv files.
   • Combine the individual scenario csv files into one master csv file for each ecoregion using Process ecoregion CSVs.ipynb.
   • We conduct our analysis on an ecoregion level, so remaining code files can easily be adapted for land cover or geographical region analysis.
2. Conduct Climate_and_fire_weather_variable_processing to process the climate and fire weather indices before analysing.
   • Process_data in any order or simultaneously for both the historic and future periods.
   • Then Check_for_missing_data using these scripts. Sometimes when batch processing the data simultaneously it causes the Earth Engine system to be overloaded, and as such certain months can be missed.
   • After the processing use Shorten CSVs. When the processing scripts run, it will save all previous iterations in the CSV file. To remove previous iterations, use this Notebook to save just the current scenario.
   • Add December 2023 and all 2024 historic data for ERA5-Land and CEMS using Add 2024 E5l and CEMS data.ipynb.
3. Process the ecoregion_mean values for each variable into individual ecoregion csv files, before combining into a single master_summary csv file for all ecoregions.
4. Analyse the data on an individual or group level using the Jupyter Notebooks.
5. Make circumpolar plots across all ecoregions using Plots from master summary.ipynb. Bar plots of burned area can also be made here for grouped regions.