Fire pits play an important role in erosion associated with the
Cave Creek Complex Fire, Sonoran Desert, Arizona
Case Allen1* and Ronald I. Dorn1
1
School of Geographical Sciences, Arizona State University, Tempe, USA
* Correspondence to:
Case Allen, School of Geographical Sciences, Arizona State University, PO Box 870104, Tempe
AZ 85287-0104, USA, E-mail: case@asu.edu
ABSTRACT
The June 2005 Cave Creek Complex Fire is one of the largest historic wildfires to affect
Arizona’s Sonoran Desert. The wildfire burned sufficiently hot to burn the roots of desert scrub
vegetation, leading to a pitted landscape of rimmed fire pit depressions. Monitoring erosion in a
small representative drainage basin after three summer precipitation seasons reveals that fire pits
have played a major roll in the initiation of rilling and gullying. Rilling initiates when these
depressions overflow, and the steep slope of the pit rims start the rilling process that grows
down-basin. Deeper gullying takes place through headcutting up the initial rills to the fire pits
and then further upslope. Although rilling and gullying also occurred through more classic
overland flow processes, we observed that at least a third of the rills and gullies in the small
basin in the burn resulted from this process.
Keywords: biogeomorphology; gully; fire; rill; Sonoran Desert; wildfire
INTRODUCTION
Wildfires dramatically increase soil erosion rates, increase gully formation and alter the behavior
of streams (Blake et al. 2005, Doerr et al. 2000, Germanoski and Miller 1995, Moody and Kinner
2006, Shakesby and Doerr 2006). This is true even in arid regions (Germanoski and Miller
1995, Shakesby and Doerr 2006). Despite a substantial history of over 200 wildfire fires in
central Arizona over the last half-century (Arrowsmith 2004, Schmid and Rogers 1988), only
been a handful of studies have analyzed the erosional impact of fires in the Sonoran Desert
(Arrowsmith 2004, Desilets et al. 2007, Misner et al. 2002, O'Dea and Guertin 2003, Pearthree
2004). Understanding the erosional impact of fires will become increasingly important as
populations continue to expand in the arid southwestern United States. This paper, thus, presents
initial results of monitoring gullying associated with a major fire in the Sonoran Desert, the 2005
Cave Creek Complex fire.
STUDY AREA BACKGROUND
The Cave Creek Complex fire resulted from the combination of the Bronco and
Humboldt fires. Starting June 21, 2005 from lightning, the wildfire burned 248,310 acres (USFS
2005). The fire was so spectacular that it was seen from space (Fig. 1) (Kekesi 2005), and
thermal satellite imagery of the fire was used by the National Weather Service to illustrate how
fire impacts the surrounding atmosphere (Flagstaff_NWS 2005). The wildfire burned over a
geological mix of intrusive granitic, extrusive basalt, and metamorphic rock types that host
colluvial, alluvial and pediment landforms.
Dr. Cecil Schwalbe of the U.S. Geological Survey explained that vegetation change
played a substantial role in providing fuel including:
"Thirty years ago, if a patch of desert caught fire, areas would burn here and there in a mosaic-type pattern,
said Schwalbe, adding that today the dried grasses [red brome, cheat grass and buffelgrass] continuously
carry the fire, causing it to spread out of control.” (Balazs 2005)
These invasive grasses generated mats of dried remains that turned normally small mosaicproducing fires into much larger burn areas (Knapp 1998, Lambrakis 2006, Schmid and Rogers
1988, Wentz et al. 2001). In addition to burning invading grasses, the fire fed on droughtimpacted desert scrub of creosote bush (Larrea tridentata), jojoba (Simmondsia chinensis),
crucifixion-thorn (Canotia holacantha), bur sage (Ambrosia dumosa), brittlebush (Encelia
farinosa), and ocotillo (Fouquieria sp). Small trees, including species of palo verde
(Cercidium), some catclaw (Acacia), and mesquite (Prosopis), burned in the swales and washes.
Figure 1. The Cave Creek Complex Fire on June 28, 2006 as seen by MODIS, a sensor that picks
up thermal signature. The fire burned the northern edges of metropolitan Phoenix (Kekesi 2005).
Image courtesy of NASA.
Post-fire erosion resulted from a very limited number of precipitation events (Fig. 2).
Between the June 2005 fire and the latest observations reported in this paper from October of
2007, there were three precipitation events greater than an inch per day. Most of the largest rain
events occurred during the summer “Arizona Monsoon” months where convective storms lead to
rain events that could have occurred in short bursts of only tens of minutes. It is important to
note that all of the erosion discussed here took place during drought years where annual rainfall
amounts were far below the long-term historic average.
Figure 2. Precipitation (in inches) in the study area, as measured by the “Cave Creek Fire”
automated rain gauge (Maricopa_County_Flood_Control_District 2007). This graph portrays all
days with rain events greater than 0.10 inches of precipitation per day.
The focus of this paper rests in a post-fire landscape appearance that we termed “bombpit
topography,” and we call here the individual depressions “fire pits” (Fig. 3). Instead of
hummocks normally seen underneath desert scrub, the center of a former plant is now a pit. The
only root pieces that we could identify in these fire pits appear to be from former creosote bushes
(Larrea tridentata), although fire pits can certainly be associated with a wide variety of other
burned perennial plants (Letnic et al. 2005). Because we could not identify a single creosote
bush that had resprouted in our study area, we speculate that the fire was hot enough to ignite the
volatile oils (Arteaga et al. 2005) in the creosote roots.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Figure 3. Depression left behind after a creosote bush (Larrea tridentata) burned. The
topographic elements include a distinct depression and an irregular rim of dirt, likely a part of the
former hummock. A crust of ash and silt coated many of the pits and rims. In this paper, we call
this topographic feature a “fire pit”. This photograph was taken in a few weeks after the fire in
July 2005, but prior to the onset of summer precipitation.
METHODS
This study is part of a larger project quantifying erosion and the gullying effects of postfire precipitation events; we are still gathering and analyzing data for this larger analysis. We
report here on findings associated with ground-based surveys of one small basin that provides
insight into ongoing geomorphic processes.
In order to obtain a field-based understanding of post-fire erosion processes, we selected
a small oval-shaped drainage basin of approximately 12,000 m2 for field study (Fig. 4).
Immediately after the fire on July 15, 2005, prior to the onset of the Summer 2005 summer
convective precipitation, we established a variety of survey sites. We measured: pre-existing
gullies (e.g. Fig. 5); cross-sections in pre-existing swales downslope from fire pits (e.g. Fig. 6
and 7); gates to measure rainsplash erosion on slope convexities (e.g. Fig. 8); and we measured
the volume of fire pit depressions (e.g. Fig. 9) that collected sediment.
The pre-rain volume of fire pits were calculated using a pegboard and measurement rod,
and then the changes were remeasured after the Monsoon in October of 2005. Sediment traps in
nearby unburned sites served as “controls” to compare erosion rates for the same time period.
We also surveyed a stream cross-section (see Fig. 4) immediately downstream of the small basin
to understand channel changes in response to sediment transport. Also, prior to the summer
2006 precipitation season, fifty green-painted cobbles were placed in one gully, fifty greenpainted cobbles were placed in another gully, and three radiotransmitters were placed in each of
these gullies in order to determine the distance of bedload particle transport.
Figure 4. A small drainage basin informally termed the “bombpit basin” was the subject of field
monitoring of erosion processes. The basin is centered at approximately N33.8567 W 111.6677.
Annotations indicate locations of locations surveyed three times after the Arizona Monsoons of
2005, 2006, and 2007. Left frames illustrate pre-fire and November 2006 appearances. Width of
upper image is ~760 and lower image is ~220 m. The two-lane Bartlett Lake Road in the upper
frames also provides scale.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Figure 5. Example of a pre-existing gully surveyed July 15, 2005 immediately after the June
2005 and before (Pre) the start of the summer Arizona monsoon and again in October of 2005
after the end of the Arizona Monsoon (Post).
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Figure 6. Example of a swale downslope from a fire pit surveyed and identified with a gated
system to assist in relocation. The left frame was imaged July 2005 prior to the Monsoon, and
the right frame is from October 2005 after the end of the Arizona Monsoon showing the
development of a rill starting from overflow of the fire pit.
Figure 7. Example of a burned paloverde occupying only a swale that would have collected
some subsurface storm flow prior to the burn (July 2005 upper left), but that swale was the site
of rill development immediately downslope from a fire pit during the 2005 summer convective
season (lower left October of 2005). Then, over the next two years the rill developed into gully
with depths reaching 3 m and a length of over 120 m. The morphology and location of the root
collar served as a basis for relocation and measuring rill and gully development. Note extensive
resprouting of the Paloverde.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Figure 8. A two-dimensional understanding of rainsplash transport was measured by placement
of a gate system on randomly selected inter-rill slope convexities. Our assumption was that such
a slope position would be the location of rainsplash detachment (Parsons et al. 1994).
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Figure 9. Fire pits created by burning of desert scrub roots provided opportunities to measure
sediment transport through surveying the dimensions of the depressions. The arrow indicates
true north, and the meter-fence provides scale. The upper frame from July 15 pre-dates the
summer Monsoon and the lower frame from October of 2005 after the end of the Arizona
Monsoon. Note rilling that developed downhill from the fire pit.
RESULTS AND DISCUSSION
The first finding of our small basin study is that fire pits are important agents of rill
formation. At least one-third of the rills in the small basin could be directly traced to fire pits
(e.g. Fig. 10). More rills might have developed in association with fire pits, but we could not be
certain of the association in our survey after the 2007 summer monsoon season. The process
appears to be relatively straight forward. Rills develop when the depression spills over the rim of
the fire pit. We speculate that the steeper slope of the rim provided enough shear stress to
initiate rill incision that then propagated downslope. We observed this overspill process
generating rills in the field during a natural precipitation event and artificially by pouring enough
water into a pit to create a rill-forming overflow event.
Figure 10. Examples of rills that can be traced directly to fire pits.
The second finding of our small basin study is the extreme spatial irregularity in sediment
erosion rates. On the low end of erosion, we monitored erosion associated with miniature fences
installed to understand the scale of rainsplash erosion (Fig. 8). The monitoring did not yield any
dramatic results. Sand-sized sediment did erode from placement on microtopographic
convexities where slope angles were greater than about 12˚. However, the maximum amount of
slope erosion was limited to depths of less than a centimeter after two summer convective
precipitation seasons.
A very different picture emerged from analyses of fire pits. For thirteen fire pits that had
tiny upslope drainages of only a few dozen square meters, the erosional story reveals tremendous
place-to-place variability. Each of the thirteen depressions completely filled up and overflowed
after just the 2005 summer convective precipitation season. Thus, the pre-monsoon volume of
each depression could only provide a minimum estimate of sediment eroded from the small
drainages upslope, as well as sediment eroded from the rims of the bomb pits. After subtraction
of the rim erosion (e.g. Fig. 11A), the sediment that filled in each depression was then
mathematically “redistributed” over the upslope area to calculate a minimum rate of erosion.
Whereas the unburned control sediment collection sites and three of the burned sites were within
a factor of two of the long term regional average historic average for similar topography in
central Arizona of about 0.13 acre-feet per m2 per year (Henze 2000), six burned areas
experienced rates of erosion over ten times higher. Four slope areas had erosion rates more than
thirty times the long-term historic average (Fig. 11B).
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
A
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
B.
11A. The sediment that filled the bombpits
was a combination of upslope rainsplashdetached sediment and sediment from pit rims.
Caption 11B. The control sediment traps placed
on unburned slopes showed little site-to-site
change, but the burned sites reveal tremendous
site-to-site variability in minimum erosion rates.
Figure 11. Results of analysis of traps located close to ridge crests collecting from very small
drainages of a few tens of square meters.
The few gullies that existed in the small drainage prior to the fire (e.g. Fig. 5) all grew in
depth (e.g. Fig 12). These pre-existing gullies, however, did not change dramatically in their
dimensions. Rather, they appeared to serve as important conduits for sediment transfer. For
example, after the second summer precipitation season in October 2006, many of these initial
gullies developed an extensive updrainage network of rills (e.g. Figure 13). The initial gully
segment then simply routed the eroded sediment.
Width (cm)
0
20
40
60
80
100
120
0
-5
-10
Depth (cm)
-15
2005 Pre-Monsoon
2005 Post-Monsoon
2006 Post-Monsoon
2007 Post-Monsoon
-20
-25
-30
-35
-40
-45
Fig. 12. Example of the slight deepening of pre-existing gullies.
Figure 13. Development of a rill network upstream of a pre-fire gully.
Rill Length (meters)
0
2
4
6
8
10
12
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
A.
Top end of rill system formed between
end of first 2005 Monsoon and second
2006 Monsoon. Photo from October
2006.
Rill Depth (centimeters)
0
-20
-40
-60
-80
-100
-120
B.
The rill system was almost 14 meters long above the
enlarged gully, with typical depths of under 40 cm and
occasional deep knickpoints.
In other cases, a new gully that developed during the 2005 precipitation season then
developed an extensive gully system upstream the following year (Fig. 14A). Intrabasin
alluvial fans of a few tens of square feet in area (Fig. 14B) deposited downstream of the gully
segments that formed in the summer of 2005. Again, the location of gullying the previous year
ended up being a conduit for sediment excavated through headward extension.
14
Figure 14. Up-gully and down-gully changes. A. The gully that started in a swale the summer
Monsoon after the 2005 fire (Fig. 5) grew uphill a hundred meters between October 2005 and
October 2006. 15B. Some of the excavated sediment then developed an alluvial fan
downstream from the point of initial gullying.
Gully Reach in Meters
Depth in Centimeters
0-5
A.
15
15-30
30-40
40-45
45-55
55-80
80-100
0
-20
-40
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
-60
-80
-100
-120
B.
Anecdotal observations of the large wash fed by this and other small basins suggests that
waves of sediment appear to be moving through larger washes. We recorded some of this
aggradation in a stream cross-section (Fig. 15). The cross-section located in Fig. 4 revealed up
to a meter of aggradation after the summer 2005 Monsoon. However, just a few tens of meters
downstream of this cross-section much of wash had degraded a half-meter from the pre-fire soil
surface, and a few tens of meters upstream of this cross section displayed up to two meters of
aggradation on top of a pre-fire soil surface. Between 2005 and 2007, further aggradation took
place. Thus, the washes fed by small basins appear to be extremely active corridors for transport
of waves of sediment. We were fortunate enough to observe one small flash flooding event in
the area that displayed areas of both degradation upstream of the stopping point and aggradation
where flow came to a stop. These anecdotal observations of the dynamic nature of sediment
transport in washes were supported by the complete loss (inability to relocate) of painted cobbles
and cobbles hosting radiotransmitters, despite walking the entire length of the wash down to
Bartlett Lake.
Figure 15. Approximate change in the elevation of a wash cross-section located in Fig. 4. This
wash is fed by several small basins, including the one studied here. The first set of
measurements estimate changes that occurred during the 2005 summer Monsoon season,
between a July 15, 2005 and October 2005. Between October of 2005 and 2007, a few meters of
aggradation took place. The image above the cross-section illustrates the aggrading cross-section
October of 2007.
CONCLUSION
This study presents the initial results of an ongoing monitoring of geomorphic changes
associated with the Cave Creek Complex Fire, central Arizona. Our field study of one basin
suggests several mechanisms that led to the initial phase of gullying and its subsequent
expansion. Some of the initial gullying developed from the overflow of depressions made when
the roots of desert scrub plants burned. Other gullies developed after the first Arizona Monsoon
started in swales. Still others started as rills coming off of ridgecrests from the concentration of
overland flow. Then, these initial gullies extended upslope — sometimes as a dendritic complex
of centimeter-scale rills and other times as a linear gully system with depths on the scale of a
meter. Anecdotal observations of the wash downstream from this small basin suggests that the
sediment has been delivered in pulses, where a wave of a few meters of aggradation might
temporarily stop and bury one part of the wash but leave an adjacent wash reach degrading.
Much research on the processes of fire-associated gullying (Moody and Kinner 2006,
Pearthree 2004) tends to be focused on regions wetter than the Sonoran Desert. The general lack
of studies of wildfire-related gullying in true desert settings elevates the importance of the Cave
Creek area. The combination of warmer temperatures (Weiss and Overpeck 2005) and nonnative grasses that fueled this fire (Balazs 2005, Brookes and Pyke 2001, Geiger et al. 2003,
Lambrakis 2006) will undoubtedly lead to large future wildfires outside of metro Phoenix, metro
Tucson, metro Las Vegas and other expanding desert cities. Hence, our continued studies on this
Cave Creek area have potential to be useful to understand the impact of future fires in the desert
southwestern USA.
ACKNOWLEDGMENTS
We thank Todd Willard and Christene Frisbee of the Tonto National Forest for permission
to undertake this study and Douglas Frink for field assistance.
REFERENCES
Arrowsmith, R. 2004. Multi-Spectral Remote Sensing of Brush Fire Scars in Arid Urban
Regions: Analysis of Future Fire and Flooding Hazards.
http://activetectonics.asu.edu/Fires_and_Floods/Fires_and_floods_final_report.pdf.
Arteaga, S., A. Andrade-Cetto, and R. Cardenas. 2005. Larrea tridentata (Creosote bush), an
abundant plant of Mexican and US-American deserts and its metabolite
nordihydroguaiarectic acid. Journal of Enthnopharmacology 98: 231-239.
Balazs, D. 2005. Invasive grass called threat to desert's plans, animals. Arizona Republic,
Phoenix.
Blake, W. H., I. G. Droppo, S. H. Wallbrink, S. H. Doerr, R. A. Shakesby, and G. S. Humphreys.
2005. Impacts of wildfire on effective sediment particle size: implications for post-fire
sediment budgets. Pages 143-150 in D. E. Wallling and A. J. Horowitz, eds. Sediment
Budgets 1. International Association of Hydrological Sciences Publication 291. IAHS
Press, Wallingford U.K.
Brookes, M. L., and D. A. Pyke. 2001. Invasive plants and fire in the deserts of North America.
Pages 1-14 in K. E. M. Galley and T. P. WIlson, eds. Proceedings of the Invasive Species
Workshop: the Role of Fire in the Control and Spread of Invasive Species. Fire
Conference 2000: the First National Congress on Fire Ecology, Prevention, and
Management. Tall Timbers Research Station, Tallahassee.
Desilets, S. L. E., B. Nijssen, B. Ekwurzel, and T. P. A. Ferre. 2007. Post-wildfire changes in
suspended sediment rating curves: Sabino Canyon, Arizona. Hydrological Processes 21:
1413-1423.
Doerr, S. H., R. A. Shakesby, and R. P. D. Walsh. 2000. Soil water repellency: its causes,
characteristics and hydro-geomorphological significance. Earth-Science Reviews 51: 3365.
Flagstaff_NWS. 2005. Satellite Imagery of the Cave Creek Complex Fire Event. NOAA.
Geiger, E., T. Mau-Crimmins, and H. Schussman. 2003. Spread of a nonnative graass across
southern Arizona: Multiple data sources to monitor Change. USDA Forest Service
Proceedings RMRS P-30: 116-120.
Germanoski, D., and J. R. Miller. 1995. Geomorphic response to wildfire in an arid watershed,
Crow-Canyon, Nevada. Physical Geography 16: 243-256.
Henze, M. W. 2000. Sediment yield on Spook Hill pediment, Arizona. Pages 224. Geography.
Arizona State University, Tempe.
Kekesi, A. 2005. The Cave Creek Complex Fire near Phoenix, Arizona. NASA.
Knapp, P. A. 1998. Spatio-temporal patterns of large grassland fires in the Intermountain West,
U.S.A. Global Ecology and Biogeography Letters 7: 259-272.
Lambrakis, C. 2006. Red brome grass fuels wildfires in Arizona. Arizona State University.
Letnic, M., B. Tamayo, and C. R. Dickman. 2005. The responses of mammals to La Nina (El
Nino Southern Oscillation) — associated rainfall, predation, and wildfire in central
Australia. Journal of Mammalogy 86: 689-703.
Maricopa_County_Flood_Control_District. 2007. Single-Sensor Data Report Generator.
Maricopa County.
Misner, T., M. Ramsey, and R. Arrowsmith. 2002. Analysis of brush fire scars in semi-arid
urban environments: Implications for future fire and flood hazard using field and satellite
data. Eos Transactions American Geophysical Union 83 (47): Abstract B61C-0740.
Moody, J. A., and D. A. Kinner. 2006. Spatial structures of stream and hillslope drainage
networks following gully erosion after wildfire. Earth Surface Processes and Landforms
31: 319-337.
O'Dea, M. E., and D. P. Guertin. 2003. Prescribed fire effects on erosion parameters in a
perennial grassland. Journal of Range Management 56: 27-32.
Parsons, A. J., A. D. Abrahams, and J. Wainwright. 1994. Rainsplash and erosion rates in an
interrill area on semi-arid grassland, Southern Arizona. Catena 22: 215-226.
Pearthree, P. A. 2004. Fire and sediment deposition. Arizona Geology 34 (3): 1-2.
Schmid, M. K., and G. F. Rogers. 1988. Trends in fire occurrence in the Arizona upland
subdivision of the Sonoran Desert, 1955 to 1983. The Southwestern Naturalist 33: 437444.
Shakesby, R. A., and S. H. Doerr. 2006. Wildfire as a hydrologic and geomorphological agent.
Earth-Science Reviews 74: 269-307.
USFS. 2005. Cave Creek Complex Fire. Tonto National Forest.
Weiss, J. L., and J. T. Overpeck. 2005. Is the Sonoran Desert losing its cool? Global Change
Biology 11: 2065-2077.
Wentz, E., S. Anderson, W. Stefanov, and J. Briggs. 2001. Desert fire history and effects on the
Phoenix, Arizona, metropolitan area. Proceedings of the IEEE/ISPRS Joint Workshop on
Remote Sensing and Data Fusion Over Urban Areas: 154-158.