.. _wildfire_fwi_technical_documentation:

Technical Documentation
=======================

This page provides technical details on the Wildfire FWI application,
including workflow integration, data processing steps, algorithms, and
implementation within the Climate Digital Twin (Climate DT) framework.

Overview
---------

The Wildfire FWI application computes fire weather indices based on the
Canadian Forest Fire Weather Index (FWI) system. It operates within the
Climate DT workflow and processes meteorological data generated by
climate model simulations.

The application is executed in streaming mode, meaning that indicators
are computed continuously as simulation data becomes available.


Workflow integration
---------------------

The Wildfire FWI application is integrated into the Climate DT workflow
through a sequence of processing steps:

1. **Climate model output**

   - High-resolution simulations (kilometre-scale)  
   - Hourly or sub-daily atmospheric variables  
   - Key inputs include temperature, humidity, wind, and precipitation  

2. **Generic State Vector (GSV)**

   - Standardises model output across different climate models  
   - Provides a unified data structure  
   - Enables spatial subsetting and regridding  

3. **Preprocessing**

   - Conversion of variables to required units  
   - Derivation of additional variables (e.g. relative humidity, wind speed)  
   - Calculation of daily input values (e.g. 24-hour precipitation)  

4. **FWI computation**

   - Calculation of fuel moisture codes (FFMC, DMC, DC)  
   - Derivation of fire behaviour indices (ISI, BUI)  
   - Computation of the Fire Weather Index (FWI)  

5. **Output generation**

   - Daily FWI and related indices stored in NetCDF format  
   - Aggregated statistics computed where applicable  



Algorithms and methods
-----------------------

The FWI system is based on a sequence of physically motivated empirical relationships:

**Fuel moisture codes**

- **Fine Fuel Moisture Code (FFMC)** estimates the moisture content of fine fuels and litter  
- **Duff Moisture Code (DMC)** represents moisture conditions in intermediate soil layers  
- **Drought Code (DC)** reflects long-term drying of deep organic layers  

**Fire behaviour indices**

- **Initial Spread Index (ISI)** estimates the potential rate of fire spread  
- **Buildup Index (BUI)** represents the amount of fuel available for combustion  

**Fire Weather Index (FWI)**

- Combines ISI and BUI  
- Represents potential fire intensity under given meteorological conditions  

The system uses daily meteorological input variables:

- Temperature  
- Precipitation (daily accumulation)  
- Wind speed  
- Humidity (derived or directly provided)  


Adaptation for global applications
----------------------------------

The original FWI system was developed for Canadian forests. For use in
global climate simulations, several adaptations are applied:

- Adjustment of drying factors based on latitude  
- Seasonal adjustment of parameters  
- Simplifications to ensure numerical stability in different climates  

These adaptations enable the application of FWI across diverse climatic regions.


HPC and execution environment
------------------------------

The Wildfire FWI application is designed for high-performance computing (HPC) environments and runs as part of the Climate DT infrastructure.

Key characteristics include:

- Execution within containerised environments  
- Integration with streaming workflows  
- Capability to process large volumes of climate simulation data  
- Parallel processing where applicable  

The application operates alongside climate models, enabling near real-time computation of wildfire indicators.


Input and output data
---------------------

**Input data**

- Climate DT model outputs  
- Variables:
  
  - air temperature  
  - precipitation  
  - wind components  
  - humidity or dew point temperature  

**Output data**

- Daily FWI-related variables in NetCDF format  
- Gridded fields with latitude, longitude, and time dimensions  
- Optional aggregated statistics (e.g. percentiles, exceedances)  


Limitations and assumptions
---------------------------

The main assumptions and interpretation caveats are described in the :ref:`wildfire_fwi_data_description`, including the meteorological input variables, sensitivity to input-data biases, and the absence of explicit vegetation, fuel, ignition, and human-influence information. These caveats should be considered alongside the technical workflow described here.


Further resources
-----------------

For additional information, see:

- :ref:`wildfire_fwi_data_description`, for the variables, inputs, assumptions, and interpretation caveats used by the Wildfire FWI application
- `Climate DT information on the DestinE Platform <https://platform.destine.eu/climate-dt/>`_, for context on Climate DT simulations, data access, and impact-sector applications
- `DestinE Data Lake documentation <https://destine-data-lake-docs.data.destination-earth.eu/>`_, for data discovery and access services
- `EFFIS fire danger forecast technical background <https://forest-fire.emergency.copernicus.eu/about-effis/technical-background/fire-danger-forecast>`_, for the operational fire danger class definitions used by EFFIS/CEMS