Introduction
Alaska’s forests are primarily boreal forests comprised of black spruce. When fire enters such a forest the resulting fires can reach hundreds of thousands of acres in size. Since the Alaska fire service does not actively control these fires unless they pose a threat, Alaska is an excellent place to study the effects of forest fires.
Our project sought to better estimate the amount of carbon consumed in a forest fire. The amount contributed by the above ground biomass is better understood than the contribution of the organic layer. In boreal forests biomass builds up on top of permafrost. Depending on the age of the stand this biomass layer can be quite deep or quite shallow. Depending on burn severity, a fire may burn the entire ground biomass layer or it may only scorch the top and leave the layer intact. We are combining remote sensing techniques with in situ severity measurements in order to classify the severity classes within a fire, which in turn will give us a better estimate of the amount of carbon consumed and released. Also presented here is work completed by MTRI intern Luke Spaete on the use of MODIS hot spot products for improving data on fire regime in Alaska.
Introduction
Burn Severity Mapping
Use of MODIS for Understanding Fire Regime
Burn Severity Mapping
Assessment of the spatial patterns of fire severity in relation to vegetation structure can help quantify fire emissions for carbon cycle studies. In a study funded by NASA, we are investigating the use object-based classification techniques to develop a method of assessing the spatial patterns of fuel consumption. The method we develop will be implemented at fire sites across Northern and Western North America to assess how fuel consumption varies.
We present the preliminary results of an object-based image analysis conducted within one fire complex from the 2004 Alaska fire season. The Boundary Complex covers a 203,703-hectare area located northwest of Fairbanks. The site includes a variety of fuel conditions and burn severity levels.
Use of MODIS for Understanding Fire Regime
Each year the Alaska Interagency Coordination Center (AICC) actively monitors wildland fires throughout Alaska. At the conclusion of each fire year the AICC compiles the yearly data into fire specific information. Historically this information has proven useful to scientists studying the historical effects of fire but has fallen short when it comes to studying seasonal aspects of fire occurrence i.e. start date, end date and duration of fires. Until recently the AICC was the only source where fire occurrence information could be obtained. Moderate Resolution Imaging Spectroradiometer (MODIS) satellites AQUA and Terra pass over the same location several times a day capturing information about the landscape with every pass. This increased temporal resolution can be used to look at seasonal aspects of fire occurrence, information that the AICC data fails to capture.
Data: MODIS Rapid Response Fire Data
MODIS detects fires using the fire detection algorithm based on algorithms developed for the AVHRR and TRMM VIRS (Giglio et al.,2003). It was developed for the need of the fire community for MODIS fire data shortly after acquisition to aid in fire management. It has a resolution of 1 km. The fire detection algorithm is:
Data: AICC Fire Polygons
- Vector based fire boundaries
- Historical coverage: 1950 – present
- Can be linked to Start/End date information for fires starting in 2001
Left: Example of a MODIS fire product with classification
Right: Example of the AICC fire polygons historical coverage. Each year is a different color.
Methods: GIS Workflow
Above: MODIS Rapid Response Fire points selected by AICC fire polygons. Earliest fire pixel (highlighted) becomes start date.
Results:
MODIS & AICC Start Date Comparisons
MODIS & AICC End Date Comparisons
MODIS & AICC Fire Duration by Acres Burned Comparisons
Above left: F=70 a functional relationship between variables
Above right: F=-108 No functional relationship between variables
Results show that MODIS derived fire data and AICC fire data yield similar start dates but fail to yield similar end dates for all years studied. From these results we can conclude that the end date discrepancies are not determined by the intensity of the fire year i.e. monitoring resources spread to thin, but rather by regulatory and bureaucratic issues i.e. funding timelines, that influence AICC fire data end dates.