FOREST FIRE POTENTIAL INDEX FOR NAVARRA AUTONOMIC COMMUNITY (SPAIN)

This study presents the development of a Forest Fire Potential Index at a regional scale for the Autonomic Community of Navarra at 500 meters spatial resolution. The index developed is based on the Fire Potential Index (FPI) applied by Sebastian (2001) at European scale and designed originally by Burgan (1998). The FPI is a dynamic Forest Fire Potential Index based on fuel characteristics and moisture status. The FPI uses the extinction moisture from fuel type map, the ten-hour timelag dead fuel moisture from meteorological data (temperature and relative moisture) and green vegetation percentage from Relative Greenness Vegetation Index. This research investigates the suitability of NDWI derived from MODIS satellite images to assess fire potential, and compares it with NDVI. The study period lasts from February 2000 to December 2005. The output of the model ranges between 1 and 100 and it is updated every eight days. The result of this study shows the usefulness of MODIS SWIR information for characterizing fire potential dynamics at a regional scale. In the bioclimatic Mediterranean region average values of both indexes (FPINDVI and FPINDWI) explain the unimodal behaviour of forest fires typical of this area. In addition, both show a good correlation between forest fire potential and fires occurrence. In the Atlantic bioclimatic region FPINDWI explains better the bi-modal behaviour of forest fires than FPINDVI. Thus, this indicates that NDWI is a useful vegetation index for estimating forest fire potential in the Atlantic region. Introduction Forests play an important role in the environment (Morgan et al., 2001). In Spain, forest fires are one of the main causes of destruction of natural resources, representing a threat for forest sustainability and for human life, therefore, forest fire potential estimation is one of the main concerns of the Spanish Environmental Administration. Fire danger indexes which are used to assess fire potential (Velez, 2000) take into account a wide range of factors like weather, fuel, and topography (Deeming et al., 1978). Water status of live vegetation is one of the main factors in affecting forest fire behaviour (Verbesselt et al., 2002), one of the reasons is that high moisture content increases the heat required to ignite a fuel (Maki et al., 2004) in addition, this is a particularly difficult parameter to estimate (Chuvieco et al., 2004). Scientists have studied and evaluated vegetation stress based on the Normalized Difference Vegetation Index (NDVI) (Chuvieco et al., 2002; Illera et al., 1996; Verbesselt et al., 2002) which is probably the index most frequently used for this purpose. The relationship between surface temperature and the NDVI is strongly correlated to surface moisture status (Verbesselt, et al., 2002; Alonso et al., 1996, Chuvieco et al., 1 ETSI Montes UPM. 2 Direccion General de la Conservacion de la Naturaleza. 3 Universidad de Almeria. Session No. 4—Fire Potencial Index Navarra—Huesca et al. 1999, Chuvieco et al., 2003). Thus, vegetation greenness provides a useful parameterization of the vegetation moisture content (Burgan et al., 1998). Also several studies indicate the relationship between vegetation water status and the information obtained from the shortwave-infrared (SWIR) (Khanna et al., 2007) domain due to the broad fundamental absorption band of water at 2.8 microns. Hence, it has been found a clear relationship between Normalized Vegetation Water Index (NDWI) and fuel moisture content (Ceccato et al., 2001; Zarco Tejada et al., 2003; Danson and Bowyer, 2004). Advanced Very High Resolution Radiometer (AVHRR) information has been used often in forest fire research because of the availability of thermal band and high acquisition frequency (Sannier et al., 2002: Gonzalez-Alonso et al., 1997; Aguado et al., 2003). The Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on the TERRA satellite improves the spatial and the spectral resolution provided by AVHRR. The availability of the SWIR spectral region allows estimating parameter related to moisture (Fensholt et al., 2003). The moisture content of small dead fuels is an essential parameter on forest fire ignition (Viney et al., 1990). Several studies have shown a high correlation between dead fuel moisture content with fire occurrence (Gomes et al., 2006; Pedrosa et al., 2006). Dead fuels are more dangerous than live vegetation because they are drier and more atmospheric dependent, (Verbesselt et al., 2006) so that they respond to atmospheric moisture faster than live vegetation, whose moisture content is also controlled by physiological activity (Sun et al., 2006). The Fire Potential Index (FPI) (Burgan et al., 1998) combines meteorological and remote sensing data integrating satellite and surface observations (Burgan et al., 1998). This index has showed a high correlation with fire occurrence in California and Nevada; it has been tested also in Europe showing good results in the Mediterranean region (Sebastian et al., 2001). The main goal of this research is to apply the actual FPI index at a regional scale in order to explain the forest fires behaviour in the three bioclimatic regions where the Iberian Peninsula is included. In addition, this study evaluates the potentiality of the Normalized Vegetation Water Index (NDWI) in forest fire potential determination. MODIS spectral range covers the short wave infrared (SWIR), necessary for NDWI calculation (Barbosa et al., 2001). In addition, MODIS spectral bandwidths are finer and avoid the water absorption regions in the NIR (Huete et al., 1996). Thus, this instrument seems to be appropriate to study forest fire potential. Study Area The study region is the Navarra Autonomic Community, situated in the NorthWest part of the Iberian Peninsula and with a surface of 10.420 Km. This region can be divided in three bioclimatic areas: Mediterranean, Atlantic and Alpine. Climatic conditions within each region are similar enough to dictate similar characteristics of soil and potential climax vegetation; hence, they show distinct forest fire behaviour. The Atlantic and Alpine regions are located in the Northern area. This is a mountainous area with high slopes and 2000 meter average elevation and where precipitation can be as high as 1600 mm per year. The Atlantic region is characterized mainly by a warm marine climate, strongly influenced by the sea, with abundant rains, fog and drizzles and without extreme temperatures. In the alpine region two sub regions can be distinguished: one characterized by a continental Session No. 4—Fire Potencial Index Navarra—Huesca et al. climate and other sub region close to Mediterranean area which represent a transition zone between cold and warm Mediterranean climate. The Mediterranean area is located in the southern part of Navarra, average elevation is 300 meters and precipitation can be less than 400 mm. The climate is Mediterranean, with a clear Atlantic influence in its Western part and a greater influence of continental climate towards the East. Deciduous forest predominates in Atlantic, coniferous in Alpine and sclerophyllous oak forest in Mediterranean region. In the Atlantic region fires are frequent and generally small. Occurrence is characterized by a bi-modal pattern with two maximums one at the beginning of the spring, and another one in autumn. Intermediate fire frequency with a higher relative incidence of medium and large fires is common in Mediterranean region. In this area the forest fire patterns show an absolute maximum in summer. In the Alpine region the forest fires behaviour is characterized by a low fire frequency and a strongly seasonal and annual variability. Methodology This study deals with forest fire potential which can be defined as a measure, scaled from 0 to 100, of the fuel sources available for burning (Chuvieco et al., 1989). The inputs of the model are: extinction moisture, ten-hour timelag dead fuel moisture and vegetation content percentage. The output of the model is a regional forest fire potential index at 500 meters of spatial resolution. The index is updated every eight days. Model definition FPI estimates vegetation susceptibility to ignition, however does not take into account the probability of an ignition source. The FPI is defined in the equation 1 (Burgan et al., 1998; Sebastian and San Miguel-Ayanz, 2001): ( ) ( ) VC HR FMC FPI FRAC − × − × = 1 10 1 100 , [Eq.1] where FMC10HRFRAC [%] is the ratio between ten-hour-timelag dead fine fuel moisture (FMC10HR) [%] (Forsberg and Deeming 1971) and the extinction moisture content (H.EXT) [%]. VC [%] is the vegetation content percentage, which depends on the maximum percentage of live vegetation (VCMAX) [%] and the relative greenness (RG) [%]. Ratio between dead fuel moisture content and extinction moisture The dead fine fuel takes ten hours to lose 63% of the difference in moisture between its initial content and the equilibrium moisture with the atmosphere, supposed constant the temperature and the atmospheric moisture (Chuvieco et al, 2004). This fuel corresponds with small branches of diameter between 0.6 and 2.5 cm (Anderson, 1985). The dead fine fuel moisture depends on atmospheric moisture. According to Forberg (1927) the moisture of the dead fine fuel constantly tends to reach the value of the atmospheric equilibrium moisture, which is changing continuously. Thus, the humidity of the fine and dead fuel is calculated according to the equation 2. EMC HR FMC × = 28 . 1 10 [Eq.2] Session No. 4—Fire Potencial Index Navarra—Huesca et al. Where EMC [%] is the equilibrium moisture content. The equilibrium moisture is unique for each combination of temperature and relative moisture. The algorithms used are the ones develop by Fosberg et al. (1971) (Eqs. 3, 4 and 5). T H EMC × − × + = 014784 .