Tracing how Landsat data is used to inform burnt acreage in an emergency stabilization Burned Area Emergency Response BAER for the 2012 High Park Fire in Colorado

From Earth Science Information Partners (ESIP)

Context

Wildfires can cause complex problems, from severe loss of vegetation and soil erosion, to a decrease in water quality and possible flash flooding (www.nifc.gov). The US Department of Interior defines "emergency stabilization" as “Planned actions to stabilize and prevent unacceptable degradation to natural and cultural resources, to minimize threats to life and property resulting from the effects of a fire, or to repair/replace/construct physical improvements necessary to prevent degradation of land or resources" (Bureau of Land Management Handbook H-1742-1, Rel. 1-1702). In this use-case, we examine how an emergency stabilization Burned Area Emergency Response (BAER) report for the 2012 High Park Fire (east of Fort Collins, CO) uses Landsat data to inform an estimate of burn severity quoted in the report. This BAER report documents the initial damage assessment, and proposes remediative actions to mitigate identified threats to critical values like roads, water diversion infrastructure, water quality, flooding, debris flow, invasive species, and cultural resources.

Vision

Using a combination of algorithmic techniques that may involve ingesting concept maps, we envision providing BAER teams with context-aware discovery tools that help them find and re-purpose existing BAER reports. These BAER teams often operate under time constraints to assess the damage to the land after a wildfire is contained, thereafter which an emergency stabilization BAER report is required to obtain approval for mitigative risk-reduction actions. This is one of many possible scenarios of how context-aware discovery may help BAER teams assemble knowledge relevant to a BAER report:

After total containment of fire, an interagency Burned Area Emergency Response (BAER) team commences documentation of the wildland fire in a BAER report. The report starts with a description of the fire history, observed fire behavior, values at risk, physical characteristics of the burned landscape, and other typical elements that describe the incident. Once this is completed, the BAER team submits the incomplete BAER report to a web tool called “SideBURN” (SIDEkick for burn response). SideBURN uses the draft report to compute an “incident fingerprint” of the current incident.

SideBURN regularly ingests BAER reports to harvest background information about any given fire incident to compute corresponding incident fingerprints. While ingesting incident background information in old BAER reports, SideBURN also ingests information like:

  1. treatment methods (e.g., land treatments, channel treatments, road and trail treatments),
  2. modeling tools (e.g., ERMiT, Peak Flow Calculator, Values-At-Risk Calculation Tool, etc) used to inform the choice of treatment,
  3. data (BAER Imagery Support website, USGS EROS data center, NOAA weather forecasts).

SideBURN has therefore captured previous response strategies proposed for use in a given incident. Using the current incident fingerprint, SideBURN next selects old BAER reports that roughly resemble the current situation on the ground. Previous incidents may have occurred in a different US state, but share some combination of weather, topographic, fuel characteristics, edaphic characteristics, etc as the current incident. Using information from these down-selected old BAER reports, SideBURN presents the BAER team with a visualization that allows the humans to perform an assessment of how previously used treatments, tools, and data can be re-used / re-purposed for the current incident.



Data to decisions narrative

The narrative below, expressed in two equivalent forms (as text and as a concept map), are assembled from concept maps that are constructed to be more amenable for machine processing. The narrative below is an example of informal story-telling of how Landsat data can be used to inform the "Burn Severity" section of a BAER report. One can therefore imagine telling these data-to-decisions stories in email updates, for example, either as hyperlinked text, or as a diagram. In both cases, the narrative is supported by knowledge captured in concept maps that can be retrieved through the hyperlinks.

Option 1: Narrative in the form of text

This wildfire emergency response narrative, used to demonstrate how concepts can be chained together into a data-to-decisions narrative, starts with the 2010 wildfire season in Colorado which ravaged major parts of the State. Those fires include the High Park fire that broke out in 2012 just east of Fort Collins, CO. The 2012 fires ultimately contributed to major flooding in Larimer County in September 2013, despite mitigation actions by the local community. The flood impacts included the evacuation of 18,000 people, damage to 16,000 structures, destruction of 1882 structures, and 10 human fatalities.


The containment of the High Park Fire was followed almost immediately by the issuance of an emergency stabilization BAER report that included a science-informed, data-driven initial assessment of the fire impacts. BAER reports are likely to include a section to estimate burn severity that estimates the area of land unburned, lightly burned, moderately burned, and intensely burned. These burn severity estimates are generated with the help of the Burned Area Reflectance Classification (BARC) program. The BARC uses data derived from Landsat reflectance data products, including the Landsat Normalized Burn Ratio and the difference Normalized Burn Ratio. Burn severity estimates are also used in the estimation of the likelihood of post-fire debris flow.



Option 2: Narrative in the form of a (non machine-readable) concept map

The first three "steps" in the narrative is similar to the first three steps in the narrative "Tracing how Landsat data helps foster fire-adapted communities in Colorado through data-driven, science-informed planning." Clickable node in the concept map point to detailed concept maps. These underlying concept maps will not tell the whole story: references in each concept map point to documents and websites for future information. You may also notice that the underlying concept maps use slightly formalized expressions compared to the narrative below. This is because those underlying concept maps are drafted with the intent of possible re-use by machine algorithms in the future.

The nodes below are clickable, but you would be left scrolling around madly within the restricted size window. Click here instead to render the full-scale concept map below in your browser window.