Eighth Symposium on Fire and Forest Meteorology


Assessing nationwide hourly wildfire risk in Italy using Limited Area Model forecasts and meteorological observations: the role of live fuels moisture dynamics

Francesco Gaetani, CIMA Research Foundation, Savona, Italy; and M. Cannac and M. D'andrea


Wildfires in Italy are one of the main causes of land degradation and a serious threat for local Communities and infrastructures. In Italy from 1970 through 2008 there have been 357254 wildland fires reported for an average of 9100 fires per year. The highest total in any given year was 18664 fires (1985) and the lowest total came in 1972 with 2358 fires reported. Basing on the same dataset (1970-2008) the average annual total burnt area is of 1100 km2, while, on average, the 90% of all fires in Italy burn less than 0.01 km2.

To cope with such a wildfire situation, since 2003, the Italian National Civil Protection Department (Dipartimento di Protezione Civile, DPC) makes use of a fully operational system for nationwide wildfire risk assessment, namely, RISICO (RISchio Incendi e COordinamento) (Fiorucci et al., 2008).

The algorithms of the system are based on dynamic models able to simulate in space and in time the effects that environment and vegetation physiology have on fuels and, in turn, on the potential behaviour of wildfires. The system automatically acquires from remote databases information and data both of in situ and spatial nature. Meteorological observations, RADAR data, Limited Area Model weather forecasts, EO data, and fuel data are gathered and managed by a unified Interface able to process and display many layers of information simultaneously.

RISICO system is continuously updated and new features are developed to reflect the feedback received from the operational users and the latest available knowledge in the field of vegetation modelling and fire behaviour simulation.

In this direction, recent efforts have been undertaken to characterise the time scales of water stress response of Mediterranean biota, since the hourly moisture content of such a vegetation highly discriminates risk dynamics of vast areas in Italy, especially during the summer season. In particular, basing on the use of HR weather forecasts and analyses, a new model has been recently developed and implemented in RISICO system, whose aim is to simulate the phenological stage of the different vegetation biota and, in turn, their response to the meteorological variables.

Methods and data

RISICO system is based on three main models; 1) a model for the evaluation of Fine Fuel Moisture Contents dynamics, whose design followed reasoning lines similar to those of the Fire Weather Index of the Canadian Forest Fire Danger Rating System (CFFDRS); 2) a potential fire spread model able to evaluate the potential behavior of the wildfire front; and, 3) a model for the prediction of plant phenology and, in turn, moisture contents dynamics of live fuels.

Meteorological data are the main dynamic information used by the system. In particular, RISICO has access to three different meteorological models, namely the deterministic 25 km cell side forecasts of the ECMWF (European Centre for Medium-Range Weather Forecasts) and the meteorological non-hydrostatic Limited Area Models (LAM) COSMO I7 and COSMO I3 (http://www.cosmo-model.org) whose horizontal resolution is of about 6 km and 3 km respectively. The data refer to the runs at 00:00 UTC and at 12:00 UTC and include five meteorological variables discretized in time steps of 3 hours over a time horizon of 72 hours. The considered meteorological fields are the 3-hour cumulated rainfall, the air temperature, the dew point temperature, and the wind speed and direction.

In addition to forecast weather data, RISICO system currently uses weather data from 1523 stations in Italy operated by local (Regional) Meteorological Services. Weather observations are collected from the Italian stations by the National Civil Protection Department and downlinked to a unified database for processing and storage.

3-hour daily raster maps (grids) of temperature, relative humidity, and precipitation are created by interpolating values between weather stations using suited interpolation techniques. The weather grids, with the same time and space granularities of the LAM's, are then used to initialize the runs of the fuel models of RISICO.

Dead fuel moisture dynamics is considered as a function of the meteorological variables, while live fuel dynamics models take into account also the phenological stage of the considered biota.

The system simulates the dynamics of (fine) dead and live moisture contents with a resolution of 3-hour time step and 100 metres grid cell. Basing on these values and on other parameters (topography and fuel load) the system calculates the potential rate of spread and the linear intensity of the cells, whose aggregation in time and space represents the expected risk.

The study analyzes the performances obtained by the system under different parameter values and dataset configurations. In particular, the paper presents the results obtained by the system in simulating the hourly moisture dynamics of fine live fuels using different meteorological information (forecasts and observations) characterized by different spatial and time scale.


Fiorucci P., Gaetani F., Minciardi R., 2008. Development and application of a system for dynamic wildfire risk assessment in Italy. Environmental Modelling & Software, 23, 6, pp. 690-702.

Poster Session 1, Formal Poster Reviewing with Icebreaker Reception
Tuesday, 13 October 2009, 5:30 PM-7:30 PM, Big Sky Ballroom

Previous paper  Next paper

Browse or search entire meeting

AMS Home Page