Finland: Renewing the System for Forest Fire Risk Assessment at the Finnish Meteorological Institute (IFFN No. 18)
Renewing the System for Forest Fire Risk Assessment
at the Finnish Meteorological Institute
(IFFN No. 18 – January 1998, p. 65-67)
Boreal forests, characterised by the dominance of conifer trees (spruce and pine), form a major economically important natural resource for countries in northern Europe such as Finland (60° -70° N). Forests cover nearly 78% of the total land area of Finland, i.e. about 26 million ha out of which 20 million ha are managed. Forest fires in Finland cause losses in forest yield and potentially endangers public safety. Forest fire warnings have been issued and an effective survey for the early detection of forest fires has already been practised in Finland for many decades. Recently risk monitoring services have also been used to find a suitable timing for the prescribed burning of the forest floor (used as a means of forest regeneration), and to limit the use of machinery at peat milling sites under very dry and windy conditions.
In Finland a fire risk warning is issued under dry weather conditions when a fire index, specifically developed for this purpose, has reached a given threshold value. The fire risk index is also used to guide fire survey flights over the risk areas. These flights are organized by government officials in co-operation with private flying clubs. Adoption of the surveying flights in the early 1970s resulted in a significant reduction of the area burned annually (Fig.1). There are pressures to minimise the amount of flying hours due to the high cost of this surveying method. This can be achieved by providing high spatial resolution, timely and accurate information on the fire risk, thereby directing the surveying activity over those areas with the highest risk of forest fire.
Until very recently, the fire index calculated by the Finnish Meteorological Institute has been based on a statistical relationship between a number of weather variables and the occurrence of fires. Problems with the statistically based index, e.g. the difficulty to verify the index values by direct measurements, led to a development of a new physically-based index which was recently adopted for use at FMI.
Fig.1. Number of forest fires and the total area burned area in Finland during the period 1952-1992.
The new index is based on estimates of the volumetric moisture content of the (assumed) most typical fuel in the boreal forest, i.e. the top organic soil layer (including fallen litter and small branches). The driest forest environments are clearings, thus the influence of trees on soil moisture could be ignored when developing the algorithm. Surface soil moisture is calculated with a simple physically-based model that removes water from the surface organic matter by evaporation and adds it in proportion to precipitation. Evaporation from the surface organic matter is calculated by making use of weather station data and the well known ‘Penman-Monteith’-type formula for actual evaporation. The surface moisture model was calibrated and tested against measured field data during two summers under natural conditions. The organic surface soil layer is described with only two parameters: depth of the layer and soil density. Input variables required for the every three hour time interval by the model are solar radiation, air temperature and humidity, wind speed and precipitation. Except for solar radiation, all variables are reported every three hour at the standard synoptic weather stations. Solar radiation incident at the surface is not normally measured with sufficient spatial resolution, but can be calculated from cloud observations or sun shine duration data. Also satellite data on cloud characteristics can be used to estimate solar radiation, this attractive alternative is currently being investigated at FMI with very promising initial results.
The main problems in the use of soil moisture as a forest fire index are related to the poor spatial resolution of observations and the scaling of soil moisture with the realistic correspondence to fire risk. Spatial resolution of a fire index depends largely on the density of weather stations. With a sparse network of stations, local climate features near lakes and coasts, and on hilly terrain are poorly described. Even without terrain heterogeneity, daily precipitation during summer can vary significantly within a few kilometres distance. For instance, a shower can occur at a weather station while the surroundings remain dry, or vice versa. Weather radar networks, such as NORDRAD, covering most of Denmark, Sweden and Finland, can potentially provide good spatial coverage of summer rainfall and are increasingly being used for quantitative precipitation estimates. Use of radar networks can thus significantly improve the spatial resolution of a fire index.
A convenient way to transfer soil moisture into information of fire risk is to scale the volumetric soil moisture into an index that increases with increasing risk of forest fire, i.e. with decreasing soil moisture. FMI have introduced a scale between 1 and 6, where 1 indicates very wet and 6 very dry (Tab.1). Experience has shown that an index with this type of simple scaling is well adopted by public users. The threshold for fire warnings can be set to, e.g., the mid-point of the scale: when the index reaches value 4 a fire warning is issued, and when it drops below 4, a warning being in force will be removed.
Tab. 1. Scaling of the volumetric moisture fraction into classes of surface wetness. Forest fire warnings for the public were issued/withdrawn when the index had increased above/decreased below a value of 4.0
|
Index |
Vol. moisture |
Moisture status |
|
6.0 |
0.10 |
Very dry |
|
5.0 – 5.9 |
0.11-0.14 |
Dry |
|
4.0 – 4.9 |
0.15 – 0.19 |
Moderately dry |
|
3.0 – 3.9 |
0.20 – 0.25 |
Moderately wet |
|
2.0 – 2.9 |
0.26 – 0.32 |
Wet |
|
1.0 – 1.9 |
0.33 |
Very wet |
What volumetric soil moisture should the index value 4 correspond to? This can be determined on a national level, based on statistics of forest fire occurrence and long term climatic data; the policy was adopted in Finland that for the peak month of June, having the highest frequency of fires, forest fire warnings would be issued during 15 days out of 30 on an average year. The index was given a scale of variation such that for very dry months (less than 10% probability) fire warnings would cover the whole month, but on very wet months no fire warnings would be issued.
Public reporting of the calculated index is made via radio broadcasting. The decision of fire warnings is made by duty meteorologists based on the calculated index and the prevailing weather conditions. A spatial analysis of the index produced on a geographical map helps in deciding which administrative areas will be warned of a fire risk (Fig.2). Information of the forest fire index is also available in real time on the internet for the direction of the fire survey flights over the driest areas.
The prescribed scheme can be relatively easily calibrated for different layers of surface organic soil and litter, as may become necessary for specific purposes. For instance, during early spring, when green vegetation is still absent, the risk of grass fires may develop faster than the risk of extensive forest fires. Also fires on peat production sites form a special case for which a dedicated fire risk service could be developed.
Fig.2. Forest fire index mapped across Finland based on spatial interpolation of the station data. In this situation, forest fire warning would be issued to most of the southern and western parts of the country.
From: Martti Heikinheimo
Address:
Finnish Meteorological Institute
P.O.Box 503
FIN – 00101 Helsinki
e-mail: martti.heikinheimo@fmi.fi
