Estimates of carbon released from combustion of vegetation and organic soil during wildfires have improved dramatically over the past decade.  Biomass inventory, fire effects and fire severity studies have contributed more accurate data to improve these models. (See Ottmar 2007, Brendan Rogers webinar 2015)  However, figuring out the net effect of all the various effects of fire, the recovery phase and warming climate on the carbon stored in Alaska’s forests and tundra is a lot more challenging!  You’d have to consider changes in burn extent and/or severity, increases in plant productivity in recovering burns, changes in species composition and what that means for productivity, changes in permafrost distribution and soil C decomposition, methane emissions and carbon fluxes in lake systems and wetlands–etc.!  A team lead by Dr. Dave McGuire at UAF has taken on this modeling challenge by applying their Integrated Ecosystem Model (IEM) which includes modules for fire, permafrost, and carbon cycling. Dave recently presented an overview of their findings at an IARPC-WCT/AFSC joint webinar (available HERE).  In a nutshell, they found: 1) tundra holds 2x the carbon that boreal forest does in the same area 2) there has been a net C loss from boreal land area of about 8 Tg/yr over the last 60 years, primarily driven by large fires during the 2000’s 3) arctic tundra and SE Alaska still act as C sinks, compensating for these losses so that overall, Alaska sequesters about 3.7 Tg/yr,  4) increases in fire extent predicted with with warming climate will release even more C, but longer growing seasons and increased plant growth (as much as 8-19% increased productivity throughout the remainder of this century) with warmer climate and higher CO2 concentration in the atmosphere are estimated to offset these losses under most of the climate projection scenarios. Since this nutshell summary glosses over a lot, you should take a look at the presentation and the SNAP projects page with information on scenarios and the individual models used.