Ninth International Conference on Permafrost ... - IARC Research

A Role of Description of Thaw/Freeze Processes in the Permafrost Zone forQuantifying Fire WeatherSergey VenevskySchool of Geography, University of Leeds, Leeds, UKMet Office – Hadley Centre for Climate Prediction and Research, Exeter, UKIntroductionWild fires happen frequently over northeastern Eurasia.The occurrence of these fires can be monitored from satellitedata and shows a large seasonal cycle, together with astrong interannual variation. Several indices are available inthe literature and aim at a quantification of fire risk basedon various meteorological and soil moisture parameters.In this paper, we evaluate the ability of three such indicesat quantifying the occurrence of fires over large areas ofSiberia (60°E–140°E, 48°N–72°N). We compare the abilityof Reg-FIRM fire weather danger index (Venevsky et.al.2002) at quantifying the occurrence of fires over large areasof Siberia (60°E–140°E, 48°N–72°N) and estimate a role ofthaw-freeze processes for correct description of fires withthis index.MethodsThe Reg-FIRM fire danger model was developed byVenevsky et al. (2002) for introducing fire processes in theLPJ and SEVER dynamic global vegetation models (Sitchet al. 2003, Venevsky & Maksyutov 2007). This index isevolved from the Nesterov index, according to Equation (3).The increase in fire risk caused by drier fuel loads is explicitlyaccounted for in the Reg-FIRM WFDI by an exponentially maxTdecreasing function of the soil moisture, S. The air drivinginput variables of the Reg-FIRM WFDI are the maximummaxminand minimum daily air temperature ( T airand T airin°C) and the daily soil moisture S in the upper soil layer,expressed in relative volumetric units. minWe used as input the weather parametersT airSfrom the ECMWF liqoperational data and the GLC-2000 vegetation max classification.TWe compare these indices to the number Sof airliqfires detected S liqbythe MODIS spaceborne instrument at 8-day time scales overmaxmina 4.5-year period. T airTS airThe Reg-FIRM WFDI would be unrealistic liqSover Siberia,in particular during the spring min season, without distinguishingTbetween the liquid and the airSfrozen components liqof the soilmoisture S. We replaced S in Equation (4) by the liquidfraction of soil moisture S liq, defined by S liqS ,where α is the frozen fraction of soil moisture diagnosedAlexey RubtsovSchool of Geography, University of Leeds, Leeds, UKeach month from a global run of the HadGEM1 AtmosphericSGeneral Circulation Model liqα )1 ×−=S (of the Hadley CenterThe frozen fraction of soil moisture, used in the Reg-FIRMWFDI was diagnosed from the output of the HadGEM1general circulation model (Johns et al. 2006) with a monthlytime step. However, the HadGEM1 has a “warm bias” 2°–7°C331in the Siberian region so that the predicted thaw is too earlyby 2–10 days compared to the observed values (Legates &Willmot 1990). This warm bias depends upon latitude, andit is caused by an underestimate of the low cloud fraction,which results in an overestimate of the downward shortwaveradiation at the surface. A rough-and-ready correctionof the thaw date is given by:ldd− )6 −=2(corroldmid(1)d−monthold mid−monthmid−montholdwhere dmid−monthis a mid-month thawing date in a givencorrgrid point, d is the mid-month thawing-date daymid −monthcorrected corr for the warm bias, and l is the latitude in degrees.Equationdmid(1) −monthwas empirically obtained and accounts fordelay in thaw and shift forward of freeze of roughly one dayper degree of increasing latitude north of 62°N.ResultsWith the frozen water correction, spring fires that aredetected by MODIS, are well captured by the Reg-FIRMWFDI. The improvement is quantified by both the averagedcorrelation coefficients (% of grid cells with positivecorrelation increased from 55 to 73) and their spatialdistributions. This finding confirms the importance of thawand freeze processes for predicting the occurrence of firesin boreal forests as suggested in previous studies (Venevsky2006).ReferencesJohns, T.C., Durman, C.F., Banks, H.T. et al. 2006. The newHadley Centre climate model HadGEM1: Evaluationα )1 ×−=S of coupled ( simulations. Journal of Climate 19(7):1327-1353.Legates, D.S. & Wilmott, C.J. 1990. Mean seasonalα )1 ×−= (and spatial variability in gauge-corrected, globalprecipitation. <strong>International</strong> Journal of Climatology10: 111-127.Sitch, S., Smith, B., Prentice, I.C. et al. 2003. Evaluation ofα )1 ×−= ( ecosystem dynamics, plant geography and terrestrialcarbon cycling in the LPJ dynamic global vegetationmodel. Global Change Biology 9(2): 161–185,doi:10.1046/j.1365-2486.2003.00569.x.Venevsky, S., Thonicke, K., Sitch, S., et al. 2002. Simulatingfire regimes in human dominated ecosystems: IberianPeninsula case study. Global change Biology 8: 984-998.