Spatial Variability of Forest Fires in the Polish Voivodeships in the Period of 2009–2018

The article deals with the issue of forest fire risk and regional differentiation of this phenomenon in Poland. A set of indicators characterizing forest fires in voivodeships (equivalent of provinces) was developed, including: area indicators, variability indicators, 10-year indicators, normalized indicators and a synthetic indicator. The study is based on the data for the years 2009-2018 provided by the Local Data Bank. Assessment of the severity of forest fires phenomena in the analysed period was presented. The results of the assessment were compared to theoretically designated fire risk categories. Areas where the actual number of fires, average area and the burned area were adequate to the forecasted degree of fire risk were indicated. Additionally, the obtained results allowed to identify the voivodeships where the characteristics of forest fires for the analysed period differed from the forest fire risk degree found in the literature. This indicates the complex nature of the phenomenon resulting from, inter alia, large variability of factors that determine the occurrence of forest fires. An example of such a situation is the Lubuskie Voivodeship which is assigned to the medium fire risk category, while the actual intensity was found to be low. Another example is the Podkarpackie Voivodeship with a medium and low fire risk, where over the past 10 years the intensity of the phenomenon was high.


INTRODUCTION
related to the climate and weather conditions. It is also associated with such anthropogenic factors as land use, location of built-up areas, population density, as well as the agricultural use of neighbouring areas and cultural factors.
Expanding the knowledge on location and timing of forest fires occurrence, including ignition cause, lead into more efficient prevention measures that can be oriented to limit the specific factors favouring ignition occurrence. Such activities include public education, legislative instruments governing land use planning and land management, as well as law enforcement and vigilance in case of intentional arson [Gonzalez-Olabarria et al. 2015].
Poland, in relation to other European countries, is rated at an average level in terms of the number and frequency of forest fires. Medium and high fire risk concerns about 83% of forest area [Szczygieł 2012]. In the national level, however, there is visible spatial differentiation in the level of intensity of the phenomenon, which is caused by several reasons. For the purpose of determining the fire risk categories according to similar levels of susceptibility to fire, the following factors were identified: frequency of fires in the last 10 years, forest stand features (age, forest site types), climatic conditions (average relative air humidity and percentage of days with litter humidity below 15%) and anthropogenic conditions (number of inhabitants in relation to the forest area). The criteria listed above were published in the Regulation of the Minister of the Environment of 22 March 2006 on detailed rules of forest fire protection (Dz. U. 2006 No. 58, Item 405, as amended). In accordance with the Regulation, three categories of forest fire risk were designated: category I -high risk, category II -medium risk, category III -small risk. Classification of areas into a given category may take place on both forest districts and national parks levels, as well as in relation to administrative units [Szczygieł 2012].
The results of this type of categorization are applied in the recommendations for forest management plans, plans for the protection of national or landscape parks, as well as the objectives and development directions set up in planning documents of administrative units at various levels. Consequently, public administrative authorities use them as the guideline for decision making on various types of actions, directly and indirectly related to fire protection. One of the priorities of spatial policy is the protection of forest land from the pressures of increasing urbanisation, and land changes to non-forest purposes occur to a small extent [Pawłat-Zawrzykraj and Podawca 2019, Podawca and Pawłat-Zawrzykraj 2019].
According to the fire risk assessment of poviats (county level units) and voivodeships (equivalent of provinces) developed by Szczygieł et al. [2009] for 2008 data in accordance with the current criteria, poviats were classified as follows: • forest fire risk category I -197 poviats (mainly in the centre and in the West of the country), • forest fire risk category II -153 poviats were included • forest fire risk category III -30 poviats were included.
The assessment carried out by Szczygiel was reviewed by Pokojska in 2015. The results of the study showed that the share of poviats in categories I and III of fire risk (high and small risk) decreased slightly, while in category II (medium risk) it increased. In both studies, the greatest risk level was assigned to the areas located in the Śląsko-Łużycka Lowland (Lubuskie and Dolnośląskie Voivodships) and the eastern part of the Mazowiecka Lowland (the Mazowieckie Voivodship).

The goal of the study
The main purpose of the analysis was to compare and verify the theoretically determined classification of forest fire hazards in Poland, based on the methodology proposed by Szczygieł [2009] and Pokojska [2015] with the actual state of this phenomenon, as it occurred in the years 2009-2018.
Additional goals were to present the diversity of the voivodeships in terms of the number of forest fires, the actual burnt forest areas, and the size of a single fire incident.
The study covers not only a quantitative assessment but also an analysis in temporal context. The obtained results enable to observe the trends occurring in the yearly reference periods for the individual voivodeships in relation to particular parameters describing the forest fire hazard.

METHODS
In the literature concerning forest fires, many publications have sort of data reporting character, limited to the number of forest fires and the area burnt as a result of these events. This appears to be a too narrow approach in terms of presenting the fire threat degree, and certainly is insufficient in the context of spatial and comparative diversification of this phenomenon. The study proposes an analysis of statistical data based on a set of additional indicators, divided into area, variability and synthetic categories.

The area indicators
Three indicators were used to assess the forest fire situation that occurred in the individual voivodeships: 1) The incidence density indicator W d-in , according to the formula (1):  The indicator is intended to illustrate the frequency of fires in relation to the total forest area in a voivodeship and show whether the number of fires is positively correlated with the size of forest area.
2) The burnt forest area indicator W bf , expressed by the formula (2): ] · 1000 − [ where: The indicator shows the percentage of burnt forest areas calculated per 1000 ha.
3) The indicator of average forest fire size W af , calculated by the formula (3): ] − [ The indicator aims to determine the degree of a single fire threat and at the same time, the potential duration of a firefighting operation. It was assumed that the smaller the area of a forest fire, the more efficiently it was extinguished and therefore, its spread was limited. It may be a measure of the effectiveness of the forest fire protection system.

Variability indicators
In order to assess the improvement or deterioration of the forest fire situation in the voivodeships, it was decided to show the trend of changes over last 10 years. The following three trend indicators were implemented: 1. The indicator of changes in the number of fires, expressed by the formula (4): 2. The indicator of changes in the destruction degree caused by forest fires WT bf , calculated by the formula (5): (20 ) The indicator shows a decrease or increase in the share of burnt forest area.
3. The indicator of changes in the average fire area WT af , expressed by the formula (6): (20 ) The indicator presents changes in the degree of forest fire risk, i.e. whether they cover larger or smaller areas in relation to a single incidence.

The synthetic indicator
Synthetic assessment of the diversity of the fire situation in the years 2009-2018 was carried out in three stages. In step one, 10-year indicators were calculated for individual area indicators and for each voivodeship, according to the formula (7): In the next step, in order to allow comparison, ] · 1000 − [ In the formula above, appropriate weights were used to reflect the importance of each criterion. It was assumed that from the point of view of environmental protection and sustainable development, the least destruction of forest areas by fires is paramount (weight 0.5). The second most important factor (weight 0.3) is a quick response of the fire services to extinguish the fire, which is reflected in the smallest possible area of a single fire. The lowest weight (0.2) was given to the number of incidents, assuming that the forest fire would be extinguished efficiently and would not lead to the destruction of a large area of forest.

Characteristics of voivodeships in terms of forest fires
The data presented in Table 1 were taken from the Local Data Bank [BDL] in the Agriculture, Forestry and Hunting category for the years 2009-2018. The information relating to forest areas came from the group Private and commune forests and subgroup Forest land area. The information concerning the number of forest fires and their area came from the group Threats and forest environment protection and subgroup The Forest fires by causes of occurrence. In addition, the source data were obtained from the annual reports entitled "State Forests in numbers" and from the National Forest Fire Information System.

RESULTS
The values the indicators calculated for annual reporting periods from 2009 to 2018 for each voivodeship are presented in Figures 1-3. According to the obtained results, the larger number of forest fires in relation to 1000 ha took place in the Mazowieckie, Łódzkie, Świętokrzyskie and Śląskie voivodeships, the least in the northern part (Zachodniopomorskie, Pomorskie, Warmińsko-Mazurskie, Podlaskie) and in the southern Poland (Podkarpackie, Małopolskie). It can be also stated that the highest dynamics of changes in the number of fires was observed in  In order to perform synthetic evaluation of the situation in terms of forest fire occurrences in the years 2009-2018, 10-year indicators for each administrative unit were implemented. The obtained indicators were normalized. Finally, the synthetic indicator W synt , that takes into account weighting of each 10-year surface factor, was calculated. The results are presented in Table 2 and in Figure 4.
The values of the normalized indicators as well as the proposed weight values allowed dividing voivodeships into 3 groups according to the level of the negative influence of forest fires on

DISCUSSION
The study results show that assessment of the forest fire situation varies depending to the selection of parameters that were taken into account. The obtained synthetic picture of the forest fires situation in 2009-2018 was compared with the fire risk categorization carried out by Szczygieł [2009] and Pokojska [2015] that takes into account the forest stand features, climate and anthropogenic conditions. In comparison with the classification made by Szczygieł [2009] The most diverse situation with respect to both forest risk classifications and calculated indicators describing the forest fire situation in the last 10 years can be observed in the Podlaskie Voivodship. The assessed fire risk is high [Szczygieł 2009] or medium [Pokojska 2015], whereas the data describing the fire occurrences in 2009-2018 showed a good situation due to the low number of fire incidents, the medium in terms of the burned forests area and bad considering the average fire area.

CONCLUSIONS
Summarising the presented analysis, it should first be noted that the actual forest fire hazard in the individual voivodeships is not associated only with environmental and anthropogenic conditions. This statement is best illustrated by the case of the Lubuskie Voivodeship, which is classified in category I of forest fire risk (high risk). The data describing the actual state of forest fire situation occurring in that part of the country, do not confirm the risk. This is probably due to good monitoring, rapid and efficient response of the fire service and increasing public awareness. An example of the opposite situation is the Podkarpackie Voivodeship, where the theoretically assigned forest fire risk is low but the actual state in 2009-2018 was quite unfavourable.
The analysis of the preliminary indicators for the individual years, clearly show fluctuations in the trends regardless the voivodeship location. This proves an obvious correlation between the climatic conditions and the forest fire situation. Therefore, high values of the indicators for the years 2012 and 2015 and definitely low ones for 2010 and 2017 can be observed in the analysis.
The problem of forest fires is a global threat. However, the available ways for improving or controlling the situation can be discussed at various levels, including regional ones. One may wonder whether, apart from the aspects of the natural and anthropogenic climate changes leading to long periods of drought, high temperatures and strong winds, the human beings themselves are not the most dangerous cause of forest fires. This hypothesis is confirmed by the statistics for the analysed period, which indicate that almost 75% of forest fires were caused by arson, negligence and accidents.