Starbuck Fire

Location of state-owned land within the Starbuck Fire perimeter. No federal land listed.

Land cover within the Starbuck Fire perimeter is mostly rangeland with sizeable areas of agricultural land.

Severity data

Categorical MTBS severity classifications

The categories here are specified by MTBS using cutoff values standard across the US.

Burn severity by MTBS classification.

Take-away: Most of the area within the fire perimeter is classed by MTBS as unburned-low.

Continuous severity metric (\(\Delta NBR\))

The severity data are initially derived from pixel-level differences in remotely-sensed reflectance products (Normalized Burn Ratio, NBR) compared before and after the event. MTBS then assigns these values to categories. Below are the raw \(\Delta NBR\) for the Starbuck Fire; the higher the value, the greater the burn severity. Colors here reflect a scale defined by the Starbuck Fire itself–the lowest observed to the highest observed \(\Delta NBR\).

Actual dNBR values on continuous scale

The categories below are derived from intervals within the above data:

Categorical dNBR scaled to Starbuck Fire specifically.

Take-away: Certainly confirms MTBS conclusion that much of the area was unburned or low, but

• Might discern more unburned areas from low severity that MTBS would lump
• Adds nuance within MTBS low severity categories, revealing more variability within the fire perimeter

Weather

Two sources for historical data:

• GridMET provides daily historical values in 4 km^2 grids interpolated from national-scale weather observation networks. Access via climateR package.
  
• MesoWest One-stop shop for current and hourly historical data from weather stations around the country. Access via mesowest package.

Daily mean data aren’t fine enough to characterize changes in wind direction:

Daily mean wind values from the gridMET dataset, 6-11 March 2017.

Hourly data do capture wind shifts that could be associated with frontal passage and changes in fire spread direction and speed.

It would be interesting to see if either/both of the spike-and-drop and wind direction patterns were consistent across wildfires in the Plains. Angie Reid had a paper on weather and wildfire in Oklahoma but it didn’t get into wind direction or trends within operational periods.

Hourly wind values from weather station in Clark County, KS, 6-11 March 2017. Unfortunately hourly format of X axis data confounds arrow plotting.

Other fire weather variables

These data would be useful to realistically model previous wildfire events and simulate future events.

Daily values of select variables relevant to fire behavior from gridMET.