Fire in Lowland Puget Sound.

As I write this, approximately 375 square miles of land are burning in Eastern Washington and many Washingtonians are still reeling from the damage of last year’s Carlton Complex wildfire, the largest wildfire in the states recorded history.  This year the Paradise wildfire has burned more than 2500 acres of the temperate rainforests in Olympic National Park and is the largest wildfire in the parks history. In a recent visit to Pack Forest, I observed smoke plumes from the Alder lake wildfire (shown below). Given this very active fire season here in the Pacific Northwest, one may be curious as to what wildland fire looks like in lowland Puget sound.

Alder Lake wildfire

Alder Lake wildfire burns near Elbe, WA.

Rather than large multi-day rip-roaring wildfires, fires in lowland puget sound are usually very small and quickly extinguished. From 1992-2001, the South Puget sound has averaged 182 fires per year burning an average of 81 acres of state-protected land.  Still, large wildland fires are always a possibility where sufficient fuel is available (source).

This year, West Pierce County fire officials are reporting a 700 percent increase in brush fire incidents they respond to.  In Auburn, a 20-acre brush fire shut down SR-18 for about two hours and in little more than a week another brush fire occurred near an apartment complex.  In Seattle, I-5 has been blocked numerous times this summer due to brush fires and a brush fire threatened homes in a West Seattle neighborhood.

In this post we will attempt to compare this years active “fire season” in Puget sound within a historical context. We will do this by looking at firefighter dispatch data from two different sources:

  1. Seattle Fire Department
  2. Tacoma Fire Department

Seattle

Incident report data from the Seattle fire department was aggregated as a single csv file from here. This data represents responses from the Seattle fire department and associated with each incident is:

  1. Incident ID
  2. Date
  3. Incident Type

We are interested in the prevalence of vegetation fires so we will only retain incidents that are of the type (fire response code)

  1. Brush fire (BRSH)
  2. Brush fire with Exposure (BRSHX)

After removing the other incidents, our dataset contained 2806 brush fire observations that have occurred between January 1st 2004 and August 23rd, 2015.

One of the first things we can do with this dataset is get a sense as to when brushfires are likely to occur. Below is a histogram of brush fire ignition dates for the years 2004-2014. We immediately may notice two things. One that there is a seasonal trend in ignition frequency. That is, there are many ignitions in the summer time and few in the winter months. Additionally, we can see that many of the ignitions occurred near the fourth of July.

Frequency of Seattle brush fires

In addition to these seasonal dynamics, we may also want to know something about the annual number of incidents. In the next graph, we have a time series of annual brush fire incidents that were responded to by the Seattle fire department.

Annual number of brush fires that were responded to by the Seattle fire department. The dashed line represents the mean annual number of brush fires.

Annual number of brush fires that were responded to by the Seattle fire department. The dashed line represents the mean annual number of brush fires.

The annual number of incidents range from 104 to 349 brush fires between 2004-2014, with a mean of 225 incidents. Moreover, there is not any clear trend that we can see in the time series. With these facts in mind we would now like to know how this fire season has compared to others.

To answer this question, we will see how many brushfires have occurred by this time of year in other years. The next graph shows the cumulative number of brushfire incidents as each year between 2004-2015 progresses through time.

The cumulative number of incidents through each year. The dashed red line is the year 2015 up until August 23, 2015. The remaining lines represent all other seasons from 2004-2014. The vertical dashed line represents the 4th of July.

The cumulative number of incidents through each year. The dashed red line is the year 2015 up until August 23, 2015. The remaining lines represent all other seasons from 2004-2014. The vertical dashed line represents the 4th of July.

We can see from the above graph that the number of brushfire incidents this year is substantial relative to other years. The largest number of annual brushfire incidents responded to by Seattle fire was in 2006 with a total of 349 events. With a total of 326 brushfire incidents, the 2015 “fire season” is on track to exceeding this record.

Tacoma

Like the Seattle Fire Department the Tacoma Fire Department also keeps records of brush fire occurrence within the City of Tacoma website . Much to my delight this dataset keeps records of approximate burn area and coordinates. On the other hand the Tacoma dataset only goes from January 19th 2011 to present day which is a shorter period of time than the Seattle dataset. When we keep only the “Grass/Brush/Trees fires” we are left with 1446 incidents.

First, lets compare the Seattle and Tacoma distribution of incidents throughout the calendar year.

Back to back histograms of the brushfire frequency for Seattle and Tacoma. Note the uptick in wildfire starts near the fourth of July is seen in both datasets.

Back to back histograms of the brushfire frequency for Seattle and Tacoma. Note the uptick in wildfire starts near the fourth of July is seen in both datasets.

We see from the above figure that we are basically seeing the same brush fire dynamics in Tacoma that we are seeing in Seattle. That is, fire counts have a seasonal effect and the Fourth of July is a particularly bad time for ignitions. Remember we only have four years of data so much of what was done with the Seattle dataset cannot be be done (done well at least) with this limited time frame. However, we can at least explore what sizes of brushfires were observed.  Below is a cumulative distribution function for burn area size for brush fires, as reported by the Tacoma fire department. For any point on the curve, the y-axis value tells us how many of the incidents were smaller than the corresponding x-value. Also, incidents from the current year are colored red.

The empirical cumulative distribution function of brush fire burn areas as reported by the Tacoma fire department.

The empirical cumulative distribution function of brush fire burn areas as reported by the Tacoma fire department.

From this graph, we can see that we have very many small fires (quickly approaches 1 as the burn area increases) and very few large brush fires (long slowly increasing “tail”). Since the distribution is highly skewed we will simply use quantiles to describe the data since the arithmetic mean can be misleading here. The smallest record is 0 sq. ft. the first quartile is 3 sq. ft, the median size was 10 sq. ft. the 3 quartile was 100 sq. ft. and the maximum observation covered approximately 1 square mile. We can also see from this graph that the brush fires of 2015 are fairly large relative to our fairly short reference period, but are not record setting.

This distinction belongs to what we will call the “Swan Creek brushfire”, a large brushfire on Tacoma’s Eastside that was “extinguished” one day prior to rekindling. I suspect that the 1 square mile figure likely refers to the “affected area” rather than the “area burned” since there are reports of much smaller burn areas as of Thursday (one-day after ignition) and the blaze was contained on Friday morning. Nevertheless, even if the blaze was only 5 acres it would still be the largest brushfire incident in the Tacoma dataset. As a brief aside, I have often frequented this park in search of the elusive Forest Scurf Pea (Rupertia physodes) but I digress.

In this post we have looked at some Fire Department records to get some glimpse as to what wild land fire looks like in lowland puget sound. Note the other records may exist that may shed better light on the topics we have investigated in this post and I have side stepped the measurement error issues uncovered with the “Swan Creek brushfire”. Still, we have learned a thing or two about brushfires in the Puget sound lowlands and how this year’s “fire season” compares to other years. Stay tuned, there will be more to come.

About hpodschwit

I am a third year graduate student at the University of Washington's Quantitative Ecology and Resource Management program. My research aims to better describe the cross-sectional wildfire growth data using statistical methods. These techniques can be used to more thoroughly describe wildfire regimes.
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s