Troubleshooting #
When the GUI reports Process failed!, the console that started AFEIM shows the actual error.
Common problems #
| Symptom | Likely cause | Fix |
|---|---|---|
Time serise file not found., or the run stops partway | A date in the period has no file, or the file name doesn’t contain the date | Keep Start/End Date inside the data’s coverage. Check the file naming. |
operands could not be broadcast together with shapes … | Inputs on different grids | Align the data to the base map: use Preprocess, the Base Map option in Daily Fuel Moisture, or AFEIM.Tools.snap(). |
Missing P_h_YYYY.tif, P_b_YYYY.tif or similar | Wrong Constant/Yearly choice in Frequency Probability | Match the time step to the data. Yearly needs one file per year in its own folder. |
Burned area doesn’t change after editing u_max or m_e | The existing a/ folder was reused | Delete a/ or use a new Output Folder. |
| Historical results overwritten | Same Output Folder used for several runs | Use one Output Folder per scenario. |
| P near zero everywhere | Fuel moisture given in % or as the FFMC index | Provide moisture as a fraction (0.12, not 12). |
| Burned area labels far too small | Fire perimeter shapefile in a geographic CRS | Reproject the shapefile to a projected CRS in metres. |
The input file/folder does not exist. (Python API) | Wrong path assigned to m.input.* | Check the path. Relative paths start from the notebook’s folder. |
ImportError for _my_funcs | Not Windows, or Python outside 3.9–3.11 | Use a supported setup. |
| GDAL fails to install | pip build of GDAL | Install GDAL from conda-forge before installing AFEIM. |
| Configuration is empty | New or different Python environment | The configuration is stored per installation. Re-enter it, or copy AFEIM/data/IIASA_IM_Config.ini over. |
Known issues in 1.4.5+pub #
Model.calibration()raisesTypeError: calc_q() got an unexpected keyword argument 'model'. To calibrate, use Calculate Burned Area in the GUI.- In Spread Simulation, the Slope input has no effect.
- Use existing data for q fails if the q folder also contains the
*_ct.tiffiles. Copy only the q files to a separate folder. - In Create Labeling Data, fires whose centroids fall in the same pixel and period aren’t added up.
- G4M Interpret Yield Table needs the full build.
Run times #
These are indicative times for a 14-year daily run (2001–2014) on a regional grid:
| Step | Workstation | Laptop |
|---|---|---|
| Daily Fuel Moisture | ~1 h | — |
Frequency Probability: F_supp, P_b, P_h | minutes | minutes |
Frequency Probability: P_m | ~2 h | ~6 h |
Frequency Probability: total P | ~2.3 h | ~6 h |
| Graph, 16 years of daily data | ~5 min | — |
Tips for long runs:
- The GUI steps run on a single core. Open several GUI instances to run independent steps in parallel. Each step slows down somewhat when they share the machine.
- The Python API uses all cores and is much faster for long or large runs.
- Keep input and output data on a local SSD.