>Fire Frequency

Fire Frequency

For certain types of vegetation, the fire return interval has lengthened considerably due to fire suppression. This leads to greater threats of high intensity fires. The naturally-occurring fire regime for different vegetation types includes a range of return intervals, extents and durations. Unnatural conditions include those outside this range.
Natural Disturbance

QT: Indicators

What is it?

This indicator is a comparison of observed fire frequency to expected fire frequency over a series of decades spanning the last 100 years. Fire return interval, a measure of fire frequency, is a measurable property of terrestrial vegetation that tells us about how natural a fire regime is in an area. This measure is used to score the subwatersheds based on expected fire frequency derived from CalVeg vegetation data.

Why is it important?

How many fires occur each year in a particular place is an indicator of the state of the landscape regarding the health of plant communities. Specifically, disease pressure, drought, no-burn management practices and timber harvest can directly impact the health of a natural landscape which can be observed in wildfire activity. Forests in a region damaged by increases in pest activity, dry from drought, laden with excessive fuel can burn more frequently and in greater extent. Other factors are also important such as fire intensity, and must be considered along with this information.

Target or Desired Condition

Fire is a natural part of the Sierra Nevada ecosystem. Historically, fires have naturally ranged from slow-burning under-story fires to raging stand-replacing fires (SNEP, 1996). The target condition for this indicator is for fire patterns and frequencies to oscillate around the central tendency of historical conditions. This is reflected in the vegetation-specific fire return intervals used for this indicator, where return intervals and the corollary frequencies, vary with vegetation type. The undesired condition set for this indicator is both zero fires and greater than twice the natural frequency. Therefore, a score of zero is attained under either of these conditions. A desired trend is for actual fire frequency to return to natural frequencies, depending on the vegetation present and danger to human communities.

What can influence or stress condition?
Fire is affected by climatic variables such as preceeding year’s moisture, El Niño cycles, and the Pacific Decadal Oscillations (Taylor and Beaty, 2005, Norman and Taylor, 2003, Morgan et al., 2008, Skinner et al., 2008). In fact, fire is so strongly determined by these climatic factors that land management seems to play only a minor role in regional fire patterns, except in changing vegetation patterns (through logging), locations of fire suppression (urban development), and soil moisture (thinning and clear-cutting).

Data Sources

Fire data were sourced from the CalFire FRAP as a GIS layer that logged each known fire occurrence since 1900. The location and extent are stored as polygons with attributes such as date, cause and cost of fighting the fire (if available). Fire return intervals were obtained from Nagel et al. (2005) and Stephens et al., (2007) and are presented in Table 3. Vegetation classes were acquired from the USFS CalVeg statewide natural vegetation database.

Data Transformations

The fire return intervals from the literature were converted to expected 10-year and 100-year fire frequency. These values were assigned to the CalVeg vegetation polygons.


Fire boundary data were located for the watershed. The number of fires was calculated per year for each of the planning watersheds (referred to as observed fire frequency). These values were combined into 10-year and 100-year summaries to compare decadal trends in fire activity over the century. Ten-year and 100-year fire frequency was calculated based on CalVeg vegetation classes using an area weighted average of the vegetation polygons for each planning watershed. Actual frequencies were compared to expected frequencies at the planning watershed resolution.

The comparison is standardized by generating a value (score) ranging from zero to 100 depending on how close the observed fire rate was to the expected fire frequency. A linear relationship between observed fire frequency and the score was established from zero to the expected fire rate (Figure 2). This relationship was mirrored for values greater than the expected frequency until the score reached zero (and remained zero for all greater values). The same scoring scenario was used for the 100-year analysis.

The scores were calculated for all planning watersheds, then aggregated to the subwatershed scale using an area weighted average score of the planning watershed scores. This was completed for each decade and for the entire century of records.

This indicator's data resource has not been uploaded yet.

What did we find out/How are we doing?

Fire frequency condition scores for the last decade were low for all subwatersheds, reflecting departure of the current fire patterns from historical conditions (Table 1, Figure 1). Only the Lower Feather subwatershed had anything approaching the historical fire frequencies. In this case, 7 of 19 planning watersheds (smaller creek drainages) had fires in the last decade. Throughout the watershed over the last century, there have been large areas of every subwatershed that has not had a mapped fire.

Trend Analysis

In order to find out if there was any change in fire frequencies over the century of record for the watershed, number of fires per decade were summed per subwatershed and the Mann-Kendall and Regional-Kendall trends analyses conducted (section 4.3). There were no significant trends for any subwatershed in numbers of fires per decade, where significance was measured as p < 0.05. this means that the number of fires per decade is neither increasing or decreasing over the last 100 years in any subwatershed.

Temporal and spatial resolution

Fire data are updated on an annual basis and are available as an online GIS dataset. The fire perimeter resolution is high and fires less than one acre are recorded in the dataset. The CalVeg vegetation data were created in 1979, and were recently updated (2000). The spatial resolution is based on the LandSat Multispectral Scanner (MSS) images with a resolution of 80 meters. The smallest CalVeg areas reported are less than one hectare.

How sure are we about our findings (Things to keep in mind)

The method used here of evaluating the last decade for condition (2000 – 2007) is prone to one systematic error — if fire frequencies are low or high in that decade for climatic reasons, the score will reflect that. The scores were in fact lower than the equivalent scores for the last century for 9 of the 11 subwatersheds (Table 2), suggesting that the fire frequencies have recently been low. Measuring fire frequency at the decade time scale is also challenged by the presence of vegetation types that would naturally not burn at greater than once every 20-30 years (Table 3). Thus decadal scores calculated in the way we did would tend to be variable.