ELMFIRE / FARSITE Validation Pipeline

Automated end-to-end pipeline that matches historical MTBS burn perimeters to USFS ignition points, downloads all required geospatial inputs, runs wind and fuel-moisture models, executes ELMFIRE and FARSITE, and produces a multi-page validation PDF comparing simulated vs. observed burn areas.

Repository layout

Data/
├── pipelineConfig.py                  ← single source of truth for all settings
├── parallel_api.py                    ← shared utilities (Tee, logging, retries)
├── case_metadata.py                   ← read/write per-case JSON metadata
├── cleanPipelineOutputs.py            ← clean per-case simulation outputs

# Master entry point (replaces run_validation.py)
├── runWildfireAV                      ← single CLI: setup / run / pdf

# Setup scripts (run once, locally)
├── setup/
│   ├── setupPipeline.py               ← orchestrates setup steps 1-5
│   ├── processScarsAndPoints.py       ← step 1: match perimeters → ignition points
│   ├── separateScarsAndPointsToCases.py ← step 2: create numbered case folders
│   ├── getSatelliteEndTimes.py        ← step 3: compute satellite start/end times
│   └── eraseInvalidCases.py          ← step 4: remove short/invalid cases

# Per-case simulation pipeline (run in parallel)
├── pipeline/
│   ├── runPipelineParallel.py         ← orchestrates per-case steps (see below)
│   ├── runBatch.py                    ← parallel batch runner
│   ├── getLandfireProductsForFireSim.py ← step 1:  download LANDFIRE from LFPS
│   ├── splitLandfireTifBands.py       ← step 2:  split multi-band LANDFIRE.tif
│   ├── makePhiAndAdjFiles.py          ← step 3:  create adj/phi rasters
│   ├── downloadWeatherData.py         ← step 4:  fetch ERA5 weather (OpenMeteo)
│   ├── downloadAndRunWindninja_WXS.py ← step 5a: WindNinja (WINDNINJA_SOURCE=install)
│   ├── downloadAndRunWindninja_wxModel.py ← step 5b: WindNinja wx-model variant
│   ├── downloadAndRunWindninja.py     ← step 5c: WindNinja (legacy)
│   ├── wn_to_geotiff.py               ← helper: convert WindNinja ASCII → GeoTIFF
│   ├── applyNelsonModel.py            ← step 6:  compute dead-fuel moisture
│   ├── getBarrierFile.py              ← step 7:  rasterise road/water barriers
│   ├── createElmfireInputFiles.py     ← step 8:  write ELMFIRE namelist (.data)
│   ├── prepareFarsite.py              ← step 9:  create FARSITE inputs (LCP, etc.)
│   ├── runElmfireCase.py              ← step 10: execute ELMFIRE
│   ├── runFarsiteCase.py              ← step 11: execute FARSITE (Linux native)
│   └── farsiteWindToGeotiff.py       ← step 12: extract ws/wd from FARSITE winds
│                                         (only when WINDNINJA_SOURCE=farsite)

# Validation report
├── validation/
│   └── getValidationPDF.py            ← generate multi-page validation PDF

# Developer / debug utilities
├── tools/
│   ├── monitorBatch.py                ← real-time batch progress monitor
│   ├── prefetchLandfire.py            ← pre-fetch LANDFIRE for all cases
│   ├── debugBandCounts.py             ← debug raster band counts per case
│   ├── debugSatelliteEndTimes.py      ← inspect satellite coverage curves
│   ├── compareOutputs.py              ← multi-model PDF comparison (fast)
│   ├── visualiseSingleCase.py         ← plot observed vs simulated for one case
│   └── getWeatherHerbie.py            ← fetch weather via Herbie (dev use)

# Data directories
# Input data (not version-controlled — see Input data requirements below)
├── inputs/
│   ├── mtbs_perimeters.gpkg           ← MTBS burn perimeters
│   ├── NIFC_FOD.gpkg                  ← NIFC fire occurrence points
│   ├── nasa_lance_allSatellites.gpkg  ← VIIRS/MODIS/Landsat hotspot detections
│   ├── osm_conus_roads.gpkg           ← OSM road network
│   ├── grwl.gpkg                      ← Global River Widths from Landsat waterways
│   └── osm_conus_rivers.gpkg          ← Backup river polygons
└── nelson_csharp/                     ← Nelson dead-fuel model (C# source + binary)

Prerequisites

Tool / Library	Where configured
Python ≥ 3.11	–
geopandas, rasterio, fiona, pyproj, shapely	conda/pip
pandas, numpy, requests, matplotlib	conda/pip
GDAL CLI (`gdal_translate`, `gdalbuildvrt`, `ogr2ogr`)	on `$PATH`
WindNinja CLI (`WindNinja_cli`)	conda env (see `WINDNINJA_CONDA_ENV`)
ELMFIRE executable (`elmfire`)	on `$PATH`
FARSITE Linux binary (`TestFARSITE`)	`pipelineConfig.FARSITE_FB_DIR`
Nelson C# model	`pipelineConfig.NELSON_EXE`
LFPS API access (USGS email)	`LFPS_EMAIL` environment variable
OpenMeteo ERA5 (free, no key)	`pipelineConfig.OPENMETEO_URL`

WSL setup

This pipeline is designed to run inside Windows Subsystem for Linux (WSL2) with Ubuntu 24.04. FARSITE runs as a natively compiled Linux executable so Wine is not required.

1. Enable WSL2

In an elevated PowerShell:

wsl --install -d Ubuntu-24.04
wsl --set-default-version 2

Restart when prompted, then open the Ubuntu 24.04 terminal and create your Linux user account.

2. Install Miniconda

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
# Follow the prompts; let the installer run conda init
source ~/.bashrc

3. Create the Python environment

conda create -n elmfire python=3.11 -y
conda activate elmfire

# Geospatial stack
conda install -c conda-forge \
    geopandas rasterio fiona pyproj shapely \
    pandas numpy requests matplotlib \
    gdal -y

All GDAL CLI tools (gdal_translate, gdalbuildvrt, ogr2ogr) come with the gdal conda package and are available on PATH inside the environment.

4. Install ELMFIRE

Follow the ELMFIRE build instructions. The compiled elmfire executable must be on PATH inside the elmfire conda environment. Set ELMFIRE_PATH_TO_GDAL in pipelineConfig.py to the bin/ directory of the same conda environment, e.g.:

ELMFIRE_PATH_TO_GDAL = "/home/<user>/miniconda3/envs/elmfire/bin/"

5. Install WindNinja (optional — only needed for `WINDNINJA_SOURCE="install"`)

Follow the instructions in the WindNinja Github. Note that the pipeline was setup in Ubuntu-24.04, when asked to run scripts for 22.04, run the provided files for 24.04 instead. No need to install the GUI version.

Set WINDNINJA_CONDA_ENV in pipelineConfig.py to the environment name that has WindNinja_cli on its PATH (default "base").

6. Build or install the native Linux FARSITE binary

The pipeline now runs FARSITE as a natively compiled Linux executable — Wine is no longer required. Obtain or build the TestFARSITE Linux binary and place it in any directory on the machine (e.g. /home/<user>/farsite/src/).

Set FARSITE_FB_DIR in pipelineConfig.py to that directory and FARSITE_EXE_NAME to the executable filename:

FARSITE_FB_DIR   = Path("/home/<user>/farsite/src")
FARSITE_EXE_NAME = "TestFARSITE"

Verify:

/home/<user>/farsite/src/TestFARSITE --help

7. Clone and configure

cd /home/<user>/elmfire_validation
git clone https://github.com/nick-cloudfire/autoValidate.git Data
cd Data

8. Install the Nelson dead-fuel moisture model

git clone https://github.com/nick-cloudfire/Nelson-Dead-Fuel-Moisture.git <nelson_csharp>

# Install .NET 8 SDK
sudo apt install -y dotnet-sdk-8.0

# Build the release binary
cd /home/<user>/elmfire_validation/Data/nelson_csharp
dotnet publish -c Release

The compiled binary path is set automatically by pipelineConfig.NELSON_EXE.

9. Obtain the native Linux FARSITE binary

The FARSITE Linux binary is not included in this repository. Place the compiled TestFARSITE executable in any directory (e.g. /home/<user>/farsite/src/). No Wine SDK layout is required — only the single binary needs to be present at the path pointed to by FARSITE_FB_DIR.

Edit pipelineConfig.py — at minimum update:

BASE_VALIDATION  = Path("/home/<user>/elmfire/elmfire_validation/")
FARSITE_FB_DIR   = Path("/home/<user>/farsite/src")   # dir containing TestFARSITE
WINDNINJA_SOURCE = "install"                           # or "farsite"

LANDFIRE_EMAIL is read from the LFPS_EMAIL environment variable (not stored in the config file). Add the following to your ~/.bashrc (or SLURM job script):

export [email protected]   # must be registered at https://lfps.usgs.gov

Then reload your shell: source ~/.bashrc

Input data requirements

All input data lives under inputs/ and is not version-controlled (listed in .gitignore). Download these datasets once and place them at the paths below.

MTBS burn perimeters

Source: MTBS Data Access
Place at: inputs/mtbs_perimeters.gpkg

NIFC fire occurrence points

Source: NIFC Fire Occurrence Database (FOD) — download as GeoPackage
Place at: inputs/NIFC_FOD.gpkg
Required fields: IncidentName, FireDiscoveryDateTime, FireOutDateTime

VIIRS / MODIS satellite hotspot detections

Source: NASA FIRMS archive (VIIRS/MODIS active fire detections for the US) Login with your email to download the CONUS dataset.
Merge all downloaded shapefiles into a single GeoPackage with layer output. Required columns: ACQ_DATE (date), ACQ_TIME (HHMM string), plus geometry.
Place at: inputs/nasa_lance_allSatellites.gpkg

OSM road and waterway barriers

File	Contents	Config key
`inputs/osm_conus_roads.gpkg`	US road network (layer `lines`, field `highway`)	`ROADS_GPKG`
`inputs/grwl.gpkg`	Global River Widths from Landsat (layer `lines`, field `waterway`)	`WATER_GPKG`
`inputs/osm_conus_rivers.gpkg`	Backup river polygons for areas with poor GRWL coverage	`BACKUP_WATER_GPKG`

Road GeoPackages can be extracted from a US OSM .pbf file using osmium + ogr2ogr, or downloaded from GeoFabrik.

LANDFIRE (downloaded automatically)

LANDFIRE terrain and fuel rasters are fetched automatically per case via the USGS LFPS API (step 1 of the simulation pipeline). No manual download needed — only a valid email registered at USGS LFPS, set via the LFPS_EMAIL environment variable (see WSL setup step 9).

Required layout before first run

Data/
├── inputs/
│   ├── mtbs_perimeters.gpkg
│   ├── NIFC_FOD.gpkg
│   ├── nasa_lance_allSatellites.gpkg
│   ├── osm_conus_roads.gpkg
│   ├── grwl.gpkg
│   └── osm_conus_rivers.gpkg
└── nelson_csharp/                (built from source — see WSL setup above)

# FARSITE binary lives outside this repo (see WSL setup step 6 / 9):
/home/<user>/farsite/src/TestFARSITE

Quick-start

1. Configure `pipelineConfig.py`

Edit the User-modifiable parameters and Paths sections:

# Paths
BASE_VALIDATION      = Path("/home/<user>/elmfire/elmfire_validation/")
FIRE_ROOT            = Path("/scratch/yourname/FirePairs")
FARSITE_FB_DIR       = Path("/home/<user>/farsite/src")   # dir with TestFARSITE binary

# User parameters
MIN_FIRE_YEAR        = 2024
MAX_FIRE_YEAR        = 2025
MAX_PARALLEL_CASES   = 14

# Wind source: "install" = WindNinja, "farsite" = derive winds from FARSITE output
WINDNINJA_SOURCE     = "install"

# ELMFIRE
ELMFIRE_PATH_TO_GDAL = "/path/to/your/conda/envs/elmfire/bin/"

2. Run the full pipeline

runWildfireAV is the single entry point. The --start flag selects the phase to begin from; all later phases also run automatically.

# Full pipeline from scratch (clean cases → setup → run → pdf):
./runWildfireAV --start setup

# Clean simulation outputs, re-run all cases, regenerate pdf:
./runWildfireAV --start run

# Regenerate PDF only (no simulation):
./runWildfireAV --start pdf

# Parallel run with 8 workers, skip completed cases:
./runWildfireAV --start run --workers 8 --skip-done

# Process specific cases only:
./runWildfireAV --start run --cases 00001 00005 00008

See ./runWildfireAV --help for all options.

Pipeline phases

Phase: setup (`--start setup`)

Runs the one-time case-preparation pipeline. Deletes all existing numbered case folders first (full reset), then rebuilds them from the raw MTBS and USFS input data.

Step	Script	Output
1	`setup/processScarsAndPoints.py`	`perimeters_ignitions.gpkg`, `all_ignitions.gpkg`
2	`setup/separateScarsAndPointsToCases.py`	`00001/`, `00002/`, …, `fire_pairs_summary.csv`
3	`setup/getSatelliteEndTimes.py`	`fire_pairs_summary_with_satellite.csv`
4	`setup/eraseInvalidCases.py`	removes cases shorter than `MIN_HOURS_DURATION` h
5	`write_metadata_from_summary`	`case_metadata.json` in each folder

Phase: run (`--start run`)

Cleans per-case simulation outputs (from landfire_bands onwards), then executes the per-case simulation pipeline in parallel using ProcessPoolExecutor. Step order depends on WINDNINJA_SOURCE:

`WINDNINJA_SOURCE = "install"` (default)

Step 1   getLandfireProductsForFireSim  →  LANDFIRE.tif
Step 2   splitLandfireTifBands          →  inputs/{dem,slp,asp,fbfm40,cc,ch,cbh,cbd}.tif
Step 3   makePhiAndAdjFiles             →  inputs/{adj,phi}.tif
Step 4   downloadWeatherData            →  inputs/weather.wxs
Step 5   downloadAndRunWindninja        →  inputs/{ws,wd}.tif  (WindNinja)
Step 6   applyNelsonModel               →  inputs/{m1,m10,m100}.tif
Step 7   getBarrierFile                 →  inputs/barrier.tif
Step 8   createElmfireInputFiles        →  <case>.data
Step 9   prepareFarsite                 →  farsite/{landscape.lcp, farsite.input, …}
Step 10  runElmfireCase                 →  outputs/time_of_arrival_*.tif
Step 11  runFarsiteCase                 →  farsite/outputs/farsite_ArrivalTime.asc

`WINDNINJA_SOURCE = "farsite"`

Step 1   getLandfireProductsForFireSim  →  LANDFIRE.tif
Step 2   splitLandfireTifBands          →  inputs/{dem,slp,asp,fbfm40,cc,ch,cbh,cbd}.tif
Step 3   makePhiAndAdjFiles             →  inputs/{adj,phi}.tif
Step 4   downloadWeatherData            →  inputs/weather.wxs
Step 5   (WindNinja skipped)
Step 6   applyNelsonModel               →  inputs/{m1,m10,m100}.tif
Step 7   getBarrierFile                 →  inputs/barrier.tif
Step 8   createElmfireInputFiles        →  <case>.data
Step 9   prepareFarsite                 →  farsite/{landscape.lcp, farsite.input, …}
Step 10  runFarsiteCase                 →  farsite/outputs/farsite_ArrivalTime.asc
                                            farsite/outputs/farsite_WindGrids.tif
Step 11  farsiteWindToGeotiff           →  inputs/{ws,wd}.tif  (from FARSITE wind grids)
                                            deletes farsite_WindGrids.tif after extraction
Step 12  runElmfireCase                 →  outputs/time_of_arrival_*.tif

Phase: pdf (`--start pdf`)

Generates validation.pdf in FIRE_ROOT. The report contains one page per case plus four summary pages:

Page	Content
Per-case	Burn-mask map, area-evolution curves, metrics table
Summary 1	Similarity scores vs. average wind speed
Summary 2	Similarity scores vs. burn area and fire duration (log x)
Summary 3	Area estimation bias per case (asymmetric log₂ bar chart)
Summary 4	Similarity score distribution histograms

Cleaning outputs

Cleaning is integrated into the entry point:

--start setup deletes all numbered case folders before re-running setup.
--start run deletes per-case simulation outputs (from landfire_bands onwards, including farsite/ directories) before re-running simulations. LANDFIRE.tif, case_metadata.json, and fire geometry files are preserved.

For fine-grained control use cleanPipelineOutputs.py directly:

python cleanPipelineOutputs.py                          # dry-run by default
python cleanPipelineOutputs.py --execute                # actually delete
python cleanPipelineOutputs.py --from elmfire_outputs   # only ELMFIRE outputs
python cleanPipelineOutputs.py --include-wind-tifs      # also delete ws/wd.tif

FARSITE execution (native Linux)

FARSITE runs as a natively compiled Linux executable (TestFARSITE). Wine is not required. runFarsiteCase.py writes a plain-text command file (farsite_linux.txt) with absolute Linux paths and invokes the binary directly. The completion sentinel is farsite_ArrivalTime.asc; the script verifies this file exists after the run.

After a successful FARSITE run, outputs are cleaned automatically:

WINDNINJA_SOURCE = "install": keeps only farsite_ArrivalTime.asc; deletes all other FARSITE outputs including farsite_WindGrids.tif.
WINDNINJA_SOURCE = "farsite": keeps farsite_ArrivalTime.asc + farsite_WindGrids.tif until farsiteWindToGeotiff.py has extracted ws.tif/wd.tif, then deletes farsite_WindGrids.tif too.

Case folder structure

<case_dir>/                              e.g. FirePairs/00001/
├── case_metadata.json                   all case attributes (times, name, area, …)
├── firescar.gpkg                        MTBS burn polygon
├── ignition_point.gpkg                  USFS ignition point
├── satellite_points.gpkg                hotspot points inside the burn polygon
├── LANDFIRE.tif                         raw multi-band download
├── 00001.data                           ELMFIRE namelist
├── pipeline.log                         stdout/stderr from the pipeline run
├── inputs/
│   ├── dem.tif                          elevation (m)
│   ├── slp.tif                          slope (degrees)
│   ├── asp.tif                          aspect (degrees)
│   ├── fbfm40.tif                       40-class fuel model
│   ├── cc.tif                           canopy cover (%)
│   ├── ch.tif                           canopy height (m)
│   ├── cbh.tif                          canopy base height (m)
│   ├── cbd.tif                          canopy bulk density (kg/m³)
│   ├── adj.tif                          adjacency (all 1s)
│   ├── phi.tif                          phi (all 1s)
│   ├── barrier.tif                      road/water width raster (m)
│   ├── weather.wxs                      RAWS-format weather file
│   ├── ws.tif                           wind speed  (N bands, one per hour)
│   ├── wd.tif                           wind direction (N bands, one per hour)
│   ├── m1.tif                           1-hr fuel moisture
│   ├── m10.tif                          10-hr fuel moisture
│   ├── m100.tif                         100-hr fuel moisture
│   └── windninja/                       WindNinja workspace (deleted after ws/wd built)
├── farsite/
│   ├── landscape.lcp                    FARSITE landscape file (converted from LANDFIRE)
│   ├── farsite.input                    FARSITE control file
│   ├── farsite_linux.txt                command-line args file (auto-generated)
│   ├── ignition.shp                     reprojected ignition point
│   ├── barrier.shp                      merged barrier polygons (optional)
│   └── outputs/
│       └── farsite_ArrivalTime.asc      FARSITE time-of-arrival output (sentinel)
├── outputs/
│   └── time_of_arrival_<HH>.tif         ELMFIRE TOA raster
└── scratch/                             ELMFIRE working directory

Configuration reference

All settings are documented inline in pipelineConfig.py.

Section	Key settings
1. User-modifiable	`MIN/MAX_FIRE_YEAR`, `MTBS_AREA_THRESHOLD_ACRES`, `MAX_PARALLEL_CASES`, `WINDNINJA_SOURCE`
2. Paths	`FIRE_ROOT`, `BASE_VALIDATION`, `FARSITE_FB_DIR`
3. File/dir names	CSV names, shapefile names, subfolder names
4. Column names	master CSV column keys shared across scripts
5. MTBS / USFS	field names in raw input datasets
6. LANDFIRE	band order, product names, raster naming
7. Satellite	coverage fraction, gap tolerance, buffer size
8. Weather	OpenMeteo URL and model
9. WindNinja	exe, model type, chunk size, output height, mesh resolution factor
10. ELMFIRE	exe, GDAL path, simulation parameters, moisture content
11. FARSITE	`FARSITE_FB_DIR` (dir with Linux binary), `FARSITE_EXE_NAME`
12. Barrier	road/water widths, OSM field names
13. LFPS API	base URL, product list, `LFPS_POLL_SLEEP_S`, `LFPS_POLL_HEARTBEAT_S`
14. Nelson	path to C# executable

Troubleshooting

LANDFIRE job is stuck in queue

The LFPS queue can be very long when many users are submitting jobs simultaneously. The pipeline polls indefinitely — it will wait however long the queue takes.
A "still waiting…" heartbeat is logged every LFPS_POLL_HEARTBEAT_S seconds (default 1800 s / 30 min) so you can confirm the job is still live.
Do not kill and resubmit the job — that adds another entry to the queue and makes the wait longer for everyone.
Retries (up to MAX_RETRIES = 3) are reserved for genuine failures (job status Failed or network errors), not queue waits.
Check that LFPS_EMAIL is set in your shell (echo $LFPS_EMAIL) and that the address is registered at https://lfps.usgs.gov.

OpenMeteo weather download fails with 502 / 503 / 504

These are transient gateway errors from the OpenMeteo server. The pipeline automatically retries up to 6 times with exponential back-off (5 s → 10 s → 20 s → 40 s → 60 s → 60 s). No action needed unless all 6 attempts fail.
A permanent failure (any other 4xx status) is not retried and will appear in pipeline.log with the URL and response body.

WindNinja fails

Confirm WINDNINJA_CONDA_ENV matches the conda environment with WindNinja_cli on PATH.
Check inputs/windninja/chunk_000/windninja_cli.log for the error message.
Ensure the DEM covers the full fire domain.

FARSITE fails

Confirm FARSITE_FB_DIR points to the directory containing TestFARSITE and that the binary is executable: ls -l $FARSITE_FB_DIR/TestFARSITE.
Check farsite/pipeline.log for error output.
Run the binary interactively with the generated farsite_linux.txt to test: <FARSITE_FB_DIR>/TestFARSITE farsite/farsite_linux.txt.

ELMFIRE fails

Confirm ELMFIRE_EXE is on PATH: which elmfire.
Check ELMFIRE_PATH_TO_GDAL points to a directory containing gdal_translate.
The outputs/ and scratch/ directories are created automatically; existing contents are overwritten on re-run.

Setup step 1 reports "no valid cases found"

The four filter stages (area, year, name+date match, spatial) each raise a RuntimeError with a specific message and the relevant pipelineConfig.py knobs to relax:
- No perimeters: lower MTBS_AREA_THRESHOLD_ACRES or widen MIN/MAX_FIRE_YEAR.
- No points: widen MIN/MAX_FIRE_YEAR.
- No name+date matches: increase DAY_TOLERANCE_DAYS.
- No spatial matches: increase DAY_TOLERANCE_DAYS or lower MTBS_AREA_THRESHOLD_ACRES.

Satellite end time not found

Some fires have sparse satellite coverage. Cases where coverage never reaches the threshold are skipped gracefully.
Adjust COVERAGE_FRACTION or SAT_CHAIN_MAX_GAP_DAYS in pipelineConfig to be more lenient.

Ignition snapping fails

Raised as a RuntimeError if no valid FBFM40 pixel (≥ 101) exists within 2000 cells of the ignition point. Usually means the burn perimeter is outside the LANDFIRE domain or the fuel download failed.

Disk usage is excessive

WindNinja creates many step_NNN/ subdirectories. These are deleted automatically by wn_to_geotiff.clean_windninja_outputs() after ws.tif and wd.tif are built. If a run was interrupted mid-step, re-run from the windninja step: python cleanPipelineOutputs.py --from windninja --execute then ./runWildfireAV --start run.
All GeoTIFF outputs use LZW compression + tiling.

Development notes

pipelineConfig.py is the single source of truth. Never hard-code paths, filenames, or tunable numbers inside individual scripts.
Each step script is independently runnable. Pass a case_dir argument to main() for single-case testing, or omit it to process all cases under FIRE_ROOT.
parallel_api.py provides Tee, make_logger, retry_call, and run_subprocess. Import from here instead of reimplementing locally.
case_metadata.py handles all JSON serialisation of per-case metadata, including datetime normalisation.
Scripts in tools/ are standalone dev/debug utilities and do not participate in the main pipeline.

Name		Name	Last commit message	Last commit date
Latest commit History 38 Commits
docs		docs
pipeline		pipeline
setup		setup
tools		tools
validation		validation
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
case_metadata.py		case_metadata.py
cleanPipelineOutputs.py		cleanPipelineOutputs.py
parallel_api.py		parallel_api.py
pipelineConfig.py		pipelineConfig.py
runWildfireAV		runWildfireAV

Folders and files

Latest commit

History

Repository files navigation

ELMFIRE / FARSITE Validation Pipeline

Repository layout

Prerequisites

WSL setup

1. Enable WSL2

2. Install Miniconda

3. Create the Python environment

4. Install ELMFIRE

5. Install WindNinja (optional — only needed for WINDNINJA_SOURCE="install")

6. Build or install the native Linux FARSITE binary

7. Clone and configure

8. Install the Nelson dead-fuel moisture model

9. Obtain the native Linux FARSITE binary

Input data requirements

MTBS burn perimeters

NIFC fire occurrence points

VIIRS / MODIS satellite hotspot detections

OSM road and waterway barriers

LANDFIRE (downloaded automatically)

Required layout before first run

Quick-start

1. Configure pipelineConfig.py

2. Run the full pipeline

Pipeline phases

Phase: setup (--start setup)

Phase: run (--start run)

WINDNINJA_SOURCE = "install" (default)

WINDNINJA_SOURCE = "farsite"

Phase: pdf (--start pdf)

Cleaning outputs

FARSITE execution (native Linux)

Case folder structure

Configuration reference

Troubleshooting

LANDFIRE job is stuck in queue

OpenMeteo weather download fails with 502 / 503 / 504

WindNinja fails

FARSITE fails

ELMFIRE fails

Setup step 1 reports "no valid cases found"

Satellite end time not found

Ignition snapping fails

Disk usage is excessive

Development notes

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

5. Install WindNinja (optional — only needed for `WINDNINJA_SOURCE="install"`)

1. Configure `pipelineConfig.py`

Phase: setup (`--start setup`)

Phase: run (`--start run`)

`WINDNINJA_SOURCE = "install"` (default)

`WINDNINJA_SOURCE = "farsite"`

Phase: pdf (`--start pdf`)

Packages