Tuesday, August 3, 2010 - 8:00 AM

COS 18-1: Data-model comparison reveals unprecedented recent burning of Alaskan boreal forests since CE 1860

Ryan Kelly1, Mark Olson2, Philip E. Higuera3, T. Scott Rupp2, and Feng Sheng Hu4. (1) University of Illinois, (2) University of Alaska, Fairbanks, (3) University of Idaho, (4) University of Illinois at Urbana-Champaign

Background/Question/Methods

   Fire is the primary disturbance in the boreal-forest biome, and variations in fire regime have important ecological, biogeochemical, and socioeconomic implications. In Alaska, several of the most active fire seasons on record (i.e. since ca. CE 1950) have occurred in the last decade, possibly as a result of anthropogenic warming. Given that climate change is expected to accelerate at high latitudes, it is important to understand climate-fire linkages in order to predict the trajectory of fire-regime shifts over coming decades.

   Paleo-fire data offer a valuable extension to the short observational record of boreal-forest fires. They are also essential for evaluating models used to predict future changes. We created a composite of eleven high-resolution charcoal records from lakes in the Yukon Flats ecoregion, an area with the highest historic fire frequency in Alaska. The record was used to infer sub-decadal variations in annual area burned (AAB) over the last 150 yrs. We then compared these results to AAB simulated for the same region by ALFRESCO, a stochastic model of boreal-forest fire dynamics. This data-model comparison allows us to test ALFRESCO’s ability to simulate beyond the model-calibration period, and to evaluate recent increased burning in a longer context than the observational record.  
Results/Conclusions

   Results show good agreement between simulated and proxy-based estimates of area burned over the last 150 years. When transformed to a common scale and resampled annually, the two records correlate with r>0.7 (p<0.0001). In both datasets, the greatest AAB occurs during the last 10-20 yr, coincident with the sustained highest temperatures in the climate series input to ALFRESCO. Within 50-yr moving windows, the correlation is strongest from ca. CE 1950-present (0.6<r<0.8, p<0.0001), and weaker but still significant for much of the remainder of the record (e.g. 0.3<r<0.5, p<0.01). Additionally, bootstrapped confidence intervals on the empirical estimate overlap substantially with the range of repeated simulations throughout the record. Taken together, our results suggest that summer temperature alone can account for much of the annual to decadal variability in AAB in interior Alaska over the last 150 yrs, and that the magnitude of burning over the last 10-20 yrs is unprecedented during this period. The overall consistency of ALFRESCO simulations with charcoal-based inferences of burning confirms the utility of the model for elucidating fire-regime controls over a range of time scales, and for projecting patterns and consequences (e.g., carbon and energy fluxes) of future fire-regime shifts in boreal ecosystems.