Abstract
Currently, indexes from the Fire Weather Index System (FWI) are used to predict the daily fire hazard, but there is no reliable index available in the Mediterranean region to be compared with paleofire records and check for their long-term reliability. In order to assess the past fire hazard and the fire-season length, based on data availability and requirements for fire index computation, we first chose and tested the efficiency of the Drought Code (DC) in Corsica (the main French Mediterranean fire-prone region) over the current period (1979–2016). We then used DC as a benchmark to assess the efficiency of the Monthly Drought Code (MDC) and used it to assess the Fire-Season Length (FSL), which were both used to characterize the fire hazard. Finally, we computed the Holocene MDC and FSL based on the HadCM3B-M1 climate model (three dimensional, fully dynamic, coupled atmosphere-ocean global climate model without flux adjustment) datasets and compared both index trends with those from proxies of paleofire, vegetation, and land use retrieved from sedimentary records in three Corsican lakes (Bastani, Nino, and Creno). Our strategy was to (i) assess fire hazard without the constraint of the daily weather-data requirement, (ii) reconstruct Holocene fire hazard from a climate perspective, and (iii) discuss the role of climate and human fire drivers based on the MDC-Paleofire proxy comparisons. Using both the Prométhée fire database and the ERA-Interim climate database over Corsica for the current period, we showed that DC values higher than 405 units efficiently discriminated fire-days from no-fire-days. The equivalent threshold value from MDC was set at 300 units. MDC and FSL indexes calculated for each of the past 11 millennia Before Present (11 ka BP) showed high values before 7 ka BP (above 300 units for MDC) and then lower values for the mid- to late Holocene (below 300 units for MDC). Climate appeared as a key driver to predict fire occurrences, promoting fires between 11 and 8 ka BP when summers were warmer than the current ones and reducing fire hazard after 7–6 ka BP due to wetter conditions. Since 5 ka BP, humans have taken control of the fire regime through agro-pastoralism, favoring large and/or frequent events despite less fire-prone climate conditions. The current fire hazard and fire-season length computed over the last few decades (1979–2016) both reported values that were respectively higher and longer than those assessed for the previous six millennia at least and comparable for those before 7 ka BP. For the next decades, due to climate warming associated with land abandonment (fuel accumulation) and the increase in human-related sources of ignition, we can expect an increase in fire hazard and larger fire events.
Highlights
Fire is an integral part of ecosystems all around the world [1,2,3,4], and the average conditions in terms of fire seasonality, frequency, area burned, severity, and intensity define the given fire regime for each type of ecosystem [5]
Several studies used linear or non-linear models to relate meteorological variables to those of fires, and a large number of them used the indexes of the Fire Weather Index system (FWI) [23], which employs daily meteorological conditions to compute a variety of indexes aiming to estimate fuel moisture content, potential fire speed, and intensity, and an overall fire hazard index
This study aims to test whether the Monthly Drought Code (MDC) [28], originally developed for the boreal forest, could be used for the Mediterranean region
Summary
Fire is an integral part of ecosystems all around the world [1,2,3,4], and the average conditions in terms of fire seasonality, frequency, area burned, severity, and intensity define the given fire regime for each type of ecosystem [5]. The Mediterranean climate is characterized by a seasonal warm and dry climate with a marked summer drought [6,7], making it a fire-prone region [8,9]. Climate is probably one of the superordinate drivers of fires at regional scales [14] by controlling fire weather [15] (corresponding to the important factors determining fire probability of occurrence and fire behavior [16]), lightning-induced ignition [17], and the amount and distribution of flammable biomass [18,19]. Humans are a superimposed driver by their footprints on the vegetation composition (with crops, pasture, and deforestation) [20] and their organization in the landscapes, or directly by igniting fires (accidents, negligence, or intentional ignitions) [21,22]. Several studies used linear or non-linear models to relate meteorological variables to those of fires, and a large number of them used the indexes of the Fire Weather Index system (FWI) [23], which employs daily meteorological conditions to compute a variety of indexes aiming to estimate fuel moisture content, potential fire speed, and intensity, and an overall fire hazard index
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
