|
Mapping fractional coverage of major fuel types for wildland fire research in Alaskan tussock tundra
Jiaying
He, University of Maryland, hejiaying0608@gmail.com
(Presenter)
Wildland fires can cause dramatic changes in vegetation, soil and water properties of ecosystems in a short period of time, leading to strong impacts on ecosystem functions and services. Though comparatively less frequent, wildfires in Alaskan tundra are a major disturbance agent with significant impacts on climate through carbon cycle and surface albedo. Tundra fires could not only release vast underground carbon from permafrost thawing, but also introduce shrub expansion, which affects the regional surface energy balance and carbon storage capability. Additionally, recent studies indicate that climate warming has the potential to increase future wildfire frequency and burned area in the tundra. Hence global warming could further enhance tundra fires' impacts on the local ecosystems and climate change. However, better understanding of factors and mechanisms governing wildland fires in the Alaskan tundra is needed to improve our predictive capabilities. As a major components of the fire environmental triangle, fuel conditions have been commonly considered as significant factor affecting fire ignitions and spread. Paleorecords suggest that change of vegetation type could play an important role in tundra fires. To enhance the understanding of tundra fire activities, this study aims to explore the spatial distribution of major fuel types in Alaskan tundra. We develop continuous fractional coverage maps of shrub, sedge and grass in multiple years with Level-2 Landsat 8 Operational Land Imager (OLI) multispectral surface reflectance data, field observations collected in the Noatak River Basin and the Seward Peninsula in our ABoVE project and other remote sensing products. Using spectral bands and metrics calculated with Landsat 8 data, we apply Random Forest algorithm to develop models for the fractional coverage of major fuel types with 70~80% accuracies. We further map the spatial distribution of these fuel types in Alaskan tussock tundra with Landsat 8 data. This result will function as important input data for future research of wildfire behaviors in the Alaskan tundra. Associated Project(s): Poster Location ID: 113 Session Assigned: Fire Disturbance
|