With the recently observed and projected trends of growing wildland fire occurrence in high northern latitudes, satellite-based burned area mapping in these regions is becoming increasingly important for scientific and fire management communities. Coarse- and moderate-resolution remotely sensed data products are the only viable source of comprehensive and timely estimates of burned area in remote, sparsely populated regions. Several MODIS (Moderate Resolution Imaging Spectroradiometer)-based burned area products for Alaska are currently available. However, our research shows that the existing burned area products underestimate the extent of the effect of fire by 15-70%. Environmental conditions limit the effective observation of land surface in Alaska to the period between May and September. These limitations are particularly noticeable in mapping late-season fires. Here we present an ecosystem-based modification to a previously developed burned area mapping approach designed to enhance the algorithm performance in Alaska. The mapping results show a consistently high performance of the adjusted algorithm in mapping burned areas in Alaska during large (2004 and 2005) and small (2006 and 2007) fire years. The adjusted burned area product maps burned areas identified by the Monitoring Trends in Burn Severity products with the overall accuracy of 90-93% and Kappa of 0.67-0.75%.