Demystifying surface fire behavior fuel models

[Megan Sebasky]

Background: I have received a lot of feedback on LANDFIRE-assigned fire behavior fuel models (mainly FBFM40) being incorrect in certain areas, especially in terms of unnatural fuels such as agriculture (see below). What I have learned is when investigating a discrepancy in fuel model, there are three major lines of investigation:

1.       Input data: Are other LANDFIRE datasets that contribute to fuel model assignment accurately represented? This mainly includes existing vegetation type, cover, height, and disturbance.

2.       Rule sets: Are the rule sets based on combinations of EVT, EVH, EVC, and disturbance accurate? You can view the rule sets for assigning fuel models based on these factors in a database or through working with the LANDFIRE Total Fuel Change Tool.

3.       Assumptions and conditions: If other supporting datasets accurately represent the landscape, review the ‘assumed’ environmental conditions used for FBFM mapping and average fuel properties driving the models, as explained in Reeves et al., 2009, quoted below:

o   “The ‘assumed’ environmental conditions used for the FBFM mapping process are those that typify the fire weather normally encountered during the peak of the burning season in the geographic region being evaluated.”

o   “Both surface fire behavior fuel model (FBFM) classifications mapped by the LANDFIRE Project represent average fuel properties needed to drive the surface fire spread model created by Rothermel (1972, 1983). These properties include fuel load by category (live and dead) and size class (0 to 0.64 cm (0 to 0.25 in) 0.64 to 2.54 cm (0.25 to 1.0 in) and 2.54 to 7.62 cm (1.0 to 3.0 in) diameter), surface-area-to-volume ratio of each size class, heat content by category, fuel bed depth, and moisture of extinction (i.e. the moisture limit beyond which fire cannot spread) (Scott and Burgan 2005). These surface FBFMs enable estimates of expected fire behavior under specific moisture conditions (Burgan and Rothermel 1984).”

What we can do: In order for LANDFIRE to make changes to any of their datasets, models, or rulesets, a general trend or widespread issue needs to be identified and described. This is because LANDFIRE is mapping many complex relationships throughout the entire United States, and cannot incorporate all local-level phenomena. The fire behavior fuel model (FBFM 13 and 40) rulesets are unique for each map zone, and therefore changes need to be relevant at that level of geography. I have proposed collecting detailed information on prescribed and wild fires to address the three lines of investigation above and document this properly for myself, all of you, and the LANDFIRE mappers to start to understand trends on a broader scale (see working draft below). In addition, there is the potential to incorporate a version of Provisional Modeling Dynamic Fuels with an Index System (MoD-FIS) in the Northeast if we can identify the main factor(s) driving changes in fuels. In the Southwest it was level of herbaceous cover, so MoD-FIS in that region incorporates seasonal variability in herbaceous cover. In the Southeast, drought was determined to be the main factor driving changes in fuels. Therefore, LANDFIRE created drought-based fuel dynamic fuel model datasets for that region. If we can demonstrate the need for a similar product, we may just get it! I am working with the LANDFIRE mappers to develop a table to capture the information they would need to make a determination like this. You will hear from me soon on how to provide information for that table.

Draft table fields:

[Megan Sebasky]

Hi everyone,

I have continued to work on this spreadsheet with input from several folks (thank you!). See attached. I believe it is at a point that we can start testing it out. I will be contacting several people individually who I have chatted with this about already, but please feel free to send around and ask me if you'd like to work through it together.

I'm also attaching a modified fuels comprehensive plan from LANDFIRE that I found online, which gives really great descriptions of what was done in the Southwest and Southeast, and even talks about how KBDI is being tested in the Northeast. I am trying to find out who is doing this work and if we can collaborate, but I still think it will be helpful to start this investigation separately. 

[Megan Sebasky]

Another note - to investigate EVT and fuel models if you do not have ArcGIS, you can use the LANDFIRE Data Product Review website, here: https://landfire.nkn.uidaho.edu/fbfm40?map=conus

If you click the (i) information button to the right of the map, and click 'Layers,' when you click on a pixel you will get a pop-up with the EVT or FBFM in that pixel. In the popup box, you can use the arrows to toggle between FBFM and EVT attributes (see image below). You can also use the handy search tool to search for an area of interest.

(and if while you're there you could create a login and fill out the form to send info to LANDFIRE that would be awesome! though filling out the spreadsheet will get to them eventually, too :) )

  

[Megan Sebasky]

Hi all,

There have been some more modifications to the spreadsheet for it to best fit our needs. See attached.

To reiterate the goal and provide some new information:

LANDFIRE is very supportive of creating dynamic fuels datasets through a Mod-FIS system for the Northeast, as well as creating the Canadian fuels dataset for the Great Lakes States (as they have done for Alaska). These products would be produced after all of the other scheduled Remap products, so we shouldn't expect them for a few years. The creation of these datasets will take a lot of work on OUR part to determine what the index should be for a MoD-FIS system in our region, and to create the crosswalks from vegetation types to Canadian fuel models. The attached spreadsheet helps us collect data for both of those goals. Most of you will not have information to fill out all fields, but anything is helpful to get us started and we can work with it. Even if it is a dump from a database of locations of fires and a few attributes, potentially including an observed fuel model (in any system - we can crosswalk). In a perfect world, the fires would have associated polygons so we could query the LANDFIRE pixels within the polygon, and we would have someone who was on the ground at the fire look over each one and provide comments, but there are still things we can do without all of that information. If you have any questions, please let me know!