Determining carbon emissions from Siberian fires using the Canadian and Russian Fire Danger Rating Systems

Russian boreal forests contain about 25-30% of the global terrestrial biomass reserve. Fire is an important disturbance process in these forests, with 12-15 million hectares annually burned in typical years. It is important to understand the interactions between climate variability and changing climate and fire behaviour and severity in these forests. Predictions, under a 2 x CO2 scenario, are for a drastic increase in wildland fire danger that has potential to lead to substantial increases in future fire activity across Siberia. Unlike North American boreal fires, where crown fires account for much of the area burned annually, up to 80% of all Siberian fires in a typical year burn as surface fires. Many of these fires burn at relatively low intensity and cause little tree mortality, while occasional high intensity surface and crown fires kill all or most of the mature trees. These fire characteristics cause a high variability in fire severity, with associated three to four-fold variability in fuel consumption. The burn signature on the landscape can be difficult to detect or disappear quite rapidly if fires are of low intensity. The latter is especially true where grasses are present, as they can re-establish quickly within a month of a fire. As a result, it is quite difficult to accurately estimate emissions from low to moderate resolution remote-sensing data. The Russian FIRE BEAR (Fire in the Boreal Eurasia Region) Project employs a multi-dimensional research approach (satellite, wildfire aerial monitoring, and on-ground experimental fires) to help validate and extend the accuracy and capability of remote-sensing products to quantify and monitor the role of wildland fires on carbon cycling in Siberian forests. As one component of this large study, we have been working to understand how to better classify fire severity remotely on an annual basis for Russian fires to account for carbon emissions. Analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) imagery has shown that when images saturate in the thermal channel this is a direct indicator of crown fire activity. However, the area of active fire pixels does not necessarily indicate the true area of crowning due to the coarse 1 x 1-km resolution of this imagery. A second approach in understanding carbon emissions is the use of the Russian Moisture Index (MI) and the Canadian Forest Fire Weather Index (FWI) System, which are being calculated on a daily basis for Russia. Unlike the MI, which is a one-index system that is an indicator only of fuel dryness, the FWI System has 3 fuel moisture codes and 3 fire behavior indices that can be used to help understand fire behavior characteristics for different fuel types. The FIRE BEAR Project has completed a replicated experimental burning program to produce regression equations for predicting carbon emissions from fires burning in dry Scots pine (Pinus sylvestris) forests for different burning conditions using the MI and the FWI System's Buildup Index (BUI). Both systems have been found to work quite well for estimating fuel consumption on dry sites (e.g., lichen sites). However, the MI appears to be less reliable on damper sites based on our initial experimental burn results in a larch (Larix siberica) dark conifer mixed forest. The FIRE BEAR Project has also been quantifying carbon emissions from our Scots pine experimental fires in Siberia. Samples were taken from a helicopter hovering 50 m over the fires, as the flame front passed underneath. This sampling represented a combined emission product for both flaming and smoldering combustion. The ranges of emission factors (EF) for the major carbon emission products for these fires were correlated with the FWI for each fire. Higher values for EF for CO2 and modified combustion efficiency (MCE) occurred with increasing FWI values. At the same time, EF values for CO and CH4 were observed to decrease. In an associated project, we have been developing a historical fire danger database for Russia that calculates the codes and indices of the FWI System from 1953 to the present. Contouring of this data has allowed us to produce maps, so that daily values can be estimated for any part of Russia. If the location of a burned area can be determined from historical records or from satellite imagery, we can determine the BUI for that fire and use it to determine fuel consumption, which can be easily converted to estimated carbon emission outputs. This two-step process can provide a realistic estimate of carbon emissions. However, obtaining reliable weather data to calculate fire danger has been a major problem, especially over the large remote portions of Russia. Weather stations that have complete weather (e.g., 1977-present) have varied from as many as 1493 stations in 1979 to only 576 stations in 2000 (post-Soviet era). High FWI values have been found to correlate well with the location of fire hot spots found by satellites. Since official Russia wildfire records prior to 1995 are highly inaccurate, sometimes underestimating annual burned area by up to an order of magnitude, a historical fire database for Russia is being currently constructed from National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) satellite imagery that has been archived since 1979 when the first polar-orbiting satellites were launched. When these burn areas are analyzed and related to the values of the FWI System that prevailed for each fire, we will be able to reconstruct the historic burning conditions and the corresponding annual emission outputs for these fires across Siberia.