Wildfire impacts on soil-water retention in the Colorado Front Range, United States

Abstract

[1] This work examined the plot-scale differences in soil-water retention caused by wildfire in the area of the 2010 Fourmile Canyon Fire in the Colorado Front Range, United States. We measured soil-water retention curves on intact cores and repacked samples, soil particle-size distributions, and organic matter content. Estimates were also made of plant-available water based on the soil-water retention curves. Parameters for use in soil-hydraulic property models were estimated; these parameters can be used in unsaturated flow modeling for comparing burned and unburned watersheds. The primary driver for measured differences in soil-water retention in burned and unburned soils was organic matter content and not soil-particle size distribution. The tendency for unburned south-facing soils to have greater organic matter content than unburned north-facing soils in this field area may explain why unburned south-facing soils had greater soil-water retention than unburned north-facing soils. Our results suggest that high-severity wildfire can “homogenize” soil-water retention across the landscape by erasing soil-water retention differences resulting from organic matter content, which for this site may be affected by slope aspect. This homogenization could have important implications for ecohydrology and plant succession/recovery in burned areas, which could be a factor in dictating the window of vulnerability of the landscape to flash floods and erosion that are a common consequence of wildfire.

1. Introduction

[2] The water retention of soils is commonly represented using the constitutive relationship between volumetric soil-water content and matric potential (i.e., the soil-water retention curve). This water retention relationship is critical for the dynamics of soil moisture, soil-plant-water relations, net infiltration/groundwater recharge, and water exchange between the soil and atmosphere [e.g.,Nimmo, 2009]. Soil-water retention curves depend on many factors such as soil texture (i.e., the particle-size distribution) [Salter et al., 1966; Gupta and Larson, 1979; Arya et al., 1999; Nimmo et al., 2007], organic matter content [Gupta and Larson, 1979; Vereecken et al., 1989], soil structure [Jamison, 1953; Nimmo, 1997], and ambient soil temperature [Nimmo and Miller, 1986].

[3] Wildfire can profoundly impact soil properties by alteration or consumption of organic matter and reduction of structure and porosity [Certini, 2005; Mataix-Solera et al., 2011], which in turn impacts the soil-water retention curve. The first-order controls on fire effects on soil properties are the fire intensity (i.e., rate of thermal energy production) and duration; these variables are determined by factors such as fuels, atmospheric conditions, and topography [Certini, 2005]. The combination of intensity and duration (i.e., fire severity), along with second-order controls like soil-water content at the time of wildfire, determines the warming of the soil at depth. The heat impulse from fire rarely reaches 20–30 cm below land surface and temperatures greater than 150°C are seldom achieved deeper than 5 cm [DeBano, 2000]. Organic matter consumption, which is known to affect water retention in fire-impacted soils [e.g.,Stoof et al., 2010], commences between 200°C and 250°C and full combustion can occur by 460°C [Giovanni et al., 1988], with intermediate temperatures like 220°C still producing almost 40% loss of organic matter [Fernàndez et al., 1997]. At temperatures between 250°C and 500°C, black carbon is produced [Baldock and Smernik, 2002]. The organic geochemistry of the conversion and destruction of organic matter in the presence of soil heating is highly complex [González-Pérez et al., 2004; Miltner and Zech, 1997] and beyond the scope of this work.

[4] Soil structural stability can decrease because of the destruction of organic cements [Badìa and Martì, 2003], which changes the particle-size distribution [Giovannini and Lucchesi, 1997]. Aggregation of fines (i.e., clay and silt size fractions) from wildfire can lead to a “coarsening” of the particle-size distribution, resulting in a distribution with a higher sand fraction [Ulery and Graham, 1993; Molina and Sanroque, 1996; Giovannini and Lucchesi, 1997]. The study by Stoof et al. [2010]found a fining of the particle-size distribution, with a shift to more clay and silt sizes. Some studies have not observed particle-size distribution shifts after wildfire [Oswald et al., 1999].The prewildfire particle-size distribution may be the primary control on the directionality and magnitude of wildfire effects on particle-size distributions. Pore-size distribution changes can be restricted to the very near surface soils, for example,Mallik et al. [1984]found that particle-size distribution changes were restricted to the upper 2 cm of soil. Shifts in particle-size distribution could be responsible for soil-water retention differences following wildfire.Boix-Fayos [1997]attributed reduced postwildfire soil-water retention to decreased clay contents (i.e., coarsening of the particle-size distribution). Increases in bulk density reported after wildfire [Giovannini et al., 1988; Andreu et al., 2001; Stoof et al., 2010] also imply a loss of soil porosity, which should decrease the volumetric water content at saturation.

[5] The incorporation of ash into soil is an additional factor potentially impacting soil-water retention following wildfire. FollowingScott [2010] and Bodí et al. [2011], ash is defined as solid residual material from biomass combustion consisting of charred organic matter, charcoal, and mineral particles. As noted by Woods and Balfour [2010], the impact of ash incorporation into soil on soil-hydraulic properties (such as soil-water retention) hinges on the soil particle-size distribution relative to the ash particle-size distribution, which determines whether the ash has a “clogging” effect.

[6] Wildfire-affected soil can have enhanced soil-water repellency, which in turn affects water retention. Effects of soil-water repellency on soil-water retention are not binary in that soil-water retention curves affected by soil-water repellency can lie in a “transition zone” between the end members of being completely hydrophilic versus hydrophobic [Dekker et al., 2001]. Phenomena such as hysteresis can also be enhanced by soil-water repellency [Kobayashi et al., 1996; Miyata et al., 2007]. As a further complication, soil-water repellency is soil-water content dependent [Goebel et al., 2004; Regalado and Ritter, 2005; de Jonge et al., 2007], suggesting that a soil-water repellency characteristic curve may be needed to fully characterize the hydraulic properties of water-repellent soils [Bachmann et al., 2007; Karunarathna et al., 2010a, 2010b]. An alternative is to measure soil-water retention using traditional methods and then effects of repellency (if present in a given soil) may be incorporated in an “effective” sense [e.g.,Miyata et al., 2007].

[7] Previous research on fire impacts on water retention has relied on both laboratory experiments [e.g., Badìa and Martì, 2003; García-Corona et al., 2004; Stoof et al., 2010] and field-sampling studies [e.g.,Boix-Fayos, 1997; Alauzis et al., 2004; Kitzberger et al., 2005; Silva et al., 2006]. Results from some of these selected studies are summarized in Table 1. Comparison of the different studies in Table 1shows that nearly all investigations have found reductions in soil-water retention following wildfire based on soil-water contents (θ) at saturation, field capacity, and permanent wilting point.

Table 1. Summary of the Findings of Selected Studies Examining Wildfire Impacts on Soil-Water Retention^a

Study	Soil-Water Retention Impact		θPermanent wilting point	Full Range of ψ	Number of Samples	Sample Type	Fire Type	Δ Unburned − Burned PAW^b^,^c (%)
	θSaturation	Field Capacity
^a θSaturation, soil-water content at saturation;θPermanent wilting point, soil-water content at permanent wilting point;ψ, matric potential; Exp., experimental/prescribed fire; Wild, wildfire; Lab., laboratory experiment; Dist., disturbed samples (i.e., repacked); Int., intact samples (i.e., cores). a In terms of water retention, G denotes greater after wildfire, = denotes equal after wildfire, and L denotes less after wildfire. b The change (Δ) in unburned − burned plant-available water (PAW) is calculated as (unburned PAW − burned PAW)/unburned PAW × 100. c Plant-available water is given as the difference in soil-water content at −100 cm and at −1.6 × 10^4 cm. d High severity. e Moderate fire severity. f 0–5 cm depth. g 5–10 cm depth. h North-facing slope. i South-facing slope.
Mallik et al. [1984]	G	G	G		6	Int.	Exp.	−42.6
Boix-Fayos [1997]	L	L	L		60	Dist.	Wild
Badìa and Martì [2003]	L	L			8	Dist.	Lab.
Alauzis et al. [2004]		L			6	Dist.	Wild
Kitzenberger et al. [2005]	L	L	L		4	Dist.	Wild	11.1
González-Pelayo et al. [2006]					9	Dist.	Exp.
Under vegetation canopy		L,^d G^e	L^d^,^e					−34.4,^d −118.0^e
Bare soil		L,^d G^e	L,^dG^e					7.4,^d 0^e
Silva et al. [2006]	L	L	L		4	Int.	Exp.	0
Are et al. [2009]	=	=	=		20	Int.	Exp.	8.0,^f 4.8^g
Stoof et al. [2010]				L	25	Dist.	Lab.	−9.3
This study	G,^d^,^h L,^e^,^h L^i	G,^h L^i	G,^h L^i	G,^d^,^h L,^e^,^h L^i	7	Int.	Wild	−15.7,^h −49.4,^h 32.9,^i 28.8^i

[8] The finely controlled conditions of soil heating in laboratory experiments used to simulate wildfire impacts are useful for precise examination of temperature effects on soil-water retention and such controlled experiments are difficult to replicate in field studies [Stoof et al., 2010]. Controlled experiments, however, typically use sieved and repacked samples where soil structure has been substantially altered. Recent work has also suggested that furnace heating may not adequately represent wildfire conditions [Bodí et al., 2011]. Therefore, the importance of field studies to complement laboratory experiments motivated us to collect and analyze field samples of soils both affected and unaffected by wildfire. Because the field area selected in this work has mountainous topography, wildfire may differentially impact soils across the landscape because of differences in soil development, vegetation communities, soil organic matter content, and soil-water retention.

[9] The principal objective of this work was to examine soil-water retention differences between burned and unburned soils along with factors that impact water retention such as particle-size distribution and organic matter content. Here we use “burned” to describe soils that have been fire affected and “unburned” for soils in a prefire state. Water-retention differences are organized with respect to the two disparate aspects that dominate the field site, which are north- (NF) and south-facing (SF). An additional objective of this work was to provide unsaturated hydraulic property model parameters for use in unsaturated flow modeling [e.g.,Šimůnek et al., 2008] and models of coupled surface water/groundwater flow [e.g., VanderKwaak, 1999; Kollet and Maxwell, 2006; Qu and Duffy, 2007; Sudicky et al., 2008; Furman, 2008] for application to wildfire-impacted areas. The paucity of unsaturated hydraulic property model parameters for burned watersheds makes this study unique. This work was conducted using single cores taken from six research plots (Figure 1). On the NF slope, there was one unburned plot (UBNF), two burned plots at different hillslope positions at the ridge (NFR) and midslope (NFM), and an ash sample from the NFM plot. On the SF slope, there was one unburned plot (UBSF) and two burned plots at ridge (SFR) and midslope (SFM) positions.

Figure 1. Map of the 2010 Fourmile Canyon Fire area in Colorado, United States, near the city of Boulder. Locations of north- and south-facing soil plots in burned and unburned soils are shown. Base map by Sheila Murphy, US Geological Survey.

2. Field Site

[10] The field site for this effort was the 2010 Fourmile Canyon Fire near Boulder, Colorado, United States [Ebel et al., 2012; Moody and Ebel, 2012]. The fire began on 6 September 2010 and was contained on 13 September 2010 [Fourmile Emergency Stabilization Team (FEST), 2010]. High winds, rugged topography, and spatially variable vegetation contributed to a patchy network of low, moderate, and high soil burn severities (after Keeley [2009]) spread over approximately 2500 ha [FEST, 2010]. Soil burn severity focuses on wildfire-caused changes in physical, hydrologic, and biologic properties and includes, for example, color, structure, and infiltration capacity [Parsons et al., 2010].This investigation of the differences in soil-water retention caused by wildfire and aspect was facilitated by selection of burned and unburned sites with NF and SF aspect slopes at the southern edge of the fire (Figure 1). Despite being located at the edge of the fire perimeter, the burned sites were of relatively high severity [FEST, 2010] owing to a reversal of wind direction coupled with high wind velocities. The soils at the area are part of the Allens Park member of the Fern Cliff-Allens Park-Rock outcrop complex and are nonsaline and nonsodic [Moreland and Moreland, 1975]. The soils are classified as frigid Lamellic and Typic Haplustalfs [United States Department of Agriculture (USDA), 2010]. Aspect affects soil horizon development in the field area because SF aspects tend to be warmer and drier while NF aspects tend to be cooler and wetter [Veblen and Lorenz, 1991], resulting in NF soils having relatively thin O and E horizons underlain by B and C horizons (Cryalfs or Ustalfs) and SF soils having thicker A and E horizons underlain by B and C horizons (Cryolls or Ustolls) [Birkeland et al., 2003]. The parent material for the colluvial/residual soils was the Boulder Creek granodiorite, with very limited exposures of a diabase from a feature known as the Iron Dike [Gable, 1980]. The prewildfire vegetation was typical of Montane ecosystems in the Rocky Mountain Foothills and aspect dependent, with NF slopes consisting of aspen (Populus tremuloides), Rocky Mountain douglas fir (Pseudotsuga menziesii subspecies glauca), and limber pine (Pinus flexilis) and with SF slopes having more sparse forests of ponderosa pine (Pinus pondersosa) and Rocky Mountain juniper (Juniperus scopulorum). The most recent major fire in the Fourmile Canyon area was in 1860 [Graham et al., 2012], which is longer than the mean wildfire return interval in low elevation Northern Colorado forests [Veblen et al., 2000; Sherriff and Veblen, 2007]. The mature stands of vegetation (i.e., trees) and robust understory that developed during this long period between major fires may have contributed to fire severity.

3. Methods

[11] To cover the full range of soil-water contents (θ) from saturation to near-oven dryness, a variety of measurement techniques were used for the burned and unburned soils and the ash. The majority of retention measurements were conducted in the laboratory using intact cores with a combination of a hanging column [Dane and Hopmans, 2002a], pressure plate [Dane and Hopmans, 2002b], dewpoint potentiometer [Gee et al., 1992], and a relative-humidity-controlled chamber [Nimmo and Winfield, 2002]. Portions of each sample were removed from the core for the dewpoint potentiometer and relative-humidity-controlled chamber measurements. The matric potential (ψ) ranges for each laboratory technique and sample equilibration times are given in Table A1. Samples (N= 1 per experimental plot) were collected just below the surface (i.e., between one and two and 7–8 cm depth) by digging a small trench and then driving in the coring device parallel to the ground surface in the upslope direction and then excavating the intact core. The presence of many stones in the subsurface required multiple coring attempts to retrieve an intact sample without abundant rocks. At the UBNF plot, cores were taken below the several-centimeter thick litter duff layer into soil that was primarily in the B horizon. The UBSF plot cores were from the A horizon. The north-facing burned cores from the NFM and NFR plots appeared to be mixed E and B horizons and the south-facing burned cores appeared to be primarily from the A horizon. Cores from the field plots were 6 cm in diameter and 20 cm long. These larger cores were subsampled for the laboratory measurements to average dimensions of 4 cm diameter and 7–10 cm length. The ash-retention curve measurements were conducted on disturbed samples that were repacked to a field-measured bulk density, rather than on intact cores because the ash layer present immediately after the fire when ash samples were collected was only 1.8 cm thick, on average [Ebel et al., 2012]. Core samples were saturated for 24 h under vacuum and then left saturated for an additional 2 days. Miyata et al. [2007] noted that saturation of samples longer than 24 h tended to overcome water repellency.

[12] Additional retention measurements were conducted on disturbed soil samples to better characterize dry-end retention (i.e., soil-water content below 0.05 m³ m⁻³). These additional measurements on disturbed samples, used separately for calibration of Decagon 5TE soil-water content sensors (Decagon Devices), were taken at depths 5–10 cm deeper than the intact cores, which may have slight impacts on physical and hydrologic properties of those samples. These repacked samples were wetted in a controlled fashion, rather than dried like the intact cores, although the repacked samples were taken in the dry end of the soil-water retention curve with the largest matric potential near −1 × 10⁴ cm where hysteresis effects are less pronounced. Retention curve analysis of the intact cores was conducted by D. B. Stephens and Associates in Albuquerque, New Mexico. The dewpoint potentiometer analysis of the repacked samples was conducted at the U.S. Geological Survey Soil Laboratory in Boulder, Colorado. The computer program RETC [van Genuchten et al., 1991] was employed to estimate van Genuchten [1980] parameters. Parameters for the Rossi and Nimmo [1994]junction model approach for representing soil-water retention were fit using nonlinear least-squares optimization in the MATLAB software package. Field capacity remains problematic to define because benchmark pressure approaches do not apply across the full range of soils [Meyer and Gee, 1999], and field capacity can be a dynamic rather than a static hydraulic property [Ahuja et al., 2008]. Field capacity was taken to be the soil-water content at a matric potential of −340 cm, which is 0.33 bars, a commonly used value [Twarakavi et al., 2009] and the permanent wilting point was taken to be the soil-water content at a matric potential of −1.6 × 10⁴ cm [Koorevaar et al., 1983]. While field capacity can differ between soils [e.g., Romano and Santini, 2002] and the permanent wilting point can differ between plant species, using −340 cm and −1.6 × 10⁴ cm, respectively, was consistent with other studies in Table 1, and thus facilitated comparison between our work and results from previous researchers across the scientific community.

[13] Loss on ignition (LOI) measurements to determine organic matter content [Dean, 1974; Heiri et al., 2001] were conducted at 550°C for 2 h based on laboratory experiments, to determine the duration to achieve no appreciable change in mass. The LOI technique, while not the most robust method for analyzing the amount of soil organic matter, was useful here for comparing relative differences between samples. Measurement of the particle-size distribution was conducted using standard dry-sieving techniques for standard class sizes [Guy, 1969] down to 63 μm. Particle size distributions for the fraction less than 63 to 0.04 μm were analyzed using a Coulter LS-230 optical diffraction apparatus [Gee and Or, 2002]. Further information on this particle-size determination technique can be found inWinfield [2003]. Analysis of particle size and LOI was performed on bulk samples taken 0–5 cm below the surface adjacent to core samples to achieve enough sample volume to run replicate samples.

4. Results

4.1. Particle Size and Organic Matter Content

4.1.1. North-Facing Soils

[14] There were not obvious differences between the burned and unburned NF soils based on particle-size data. The particle-size data indicated that the NF soils were primarily sand (92.8%–97.6%), with a few percent silt-size particles (2.2%–6.7%) and almost no clay-sized particles (Table 2). At the NFR site, three out of four samples had >15% gravel giving the soil a USDA classification of gravelly sand. At the NFM site, two out of four samples had >15% gravel giving the soil a USDA classification of gravelly sand. All of the UBNF samples had >15% gravel. The ash had a finer textural class than the NF soil samples, with 85.2% sand and 14% silt with less than 1% clay, giving the ash a loamy-sand USDA classification. Organic matter content, based on LOI, was about a factor of 3–4 larger (by % mass) in unburned soils. Ash had a similar organic matter content amount in burned soils on NF slopes. Dry bulk density,ρ_b, was 6%–7% greater in unburned soils and ash had a much lower ρ_b than soil.

Table 2. Particle Size Distribution and LOI Data

Soil Sample	Condition	Sand^a (%)	Silt^a (%)	Clay^a (%)	LOI^b (%)	N (Number of Samples)	^c (g cm^−3)	N (Number of Samples)
^a Locations are shown in Figure 1. a With gravel fraction taken out, three out of four samples at NFR and SFR had >15% gravel, two out of four samples at NFM and SFM had >15% gravel, four out of four samples at UBNF had >15% gravel. b LOI ± one standard deviation. c is the dry bulk density.
UBNF soil	Unburned	97.6	2.2	0.2	11.2 ± 0.67	4	1.35	1
NFM soil	Burned	94.2	5.4	0.4	3.9 ± 0.33	4	1.26	1
NFR soil	Burned	92.8	6.7	0.5	3.1 ± 0.14	4	1.27	1
Ash at NFM	Burned	85.2	14.0	0.8	3.9 ± 0.34	4	1.07	1
UBSF soil	Unburned	94.1	5.7	0.2	20.0 ± 3.4	4	1.03	1
SFM soil	Burned	97.1	2.7	0.2	3.1 ± 0.35	4	1.37	1
SFR soil	Burned	94.2	5.5	0.3	6.6 ± 0.41	4	1.33	1

4.1.2. South-Facing Soils

[15] Similar to the NF soils, the SF soils did not show major differences between burned and unburned soils in the particle-size data. SF soils were primarily sand (94.1%–97.1%) with a few percent silt-size particles (2.7%–5.7%) and almost no clay-sized particles (Table 2). Similar to the NF soils, the SF soils at the ridge position (i.e., SFR) had three out of four samples with >15% gravel, giving the soil a USDA classification of gravelly sand and at the midslope site (i.e., SFM) site, two out of four samples had >15% gravel giving the soil a USDA classification of gravelly sand. None of the UBSF samples had >15% gravel. Organic matter content amounts were a factor of 3–7 larger (by % mass) in unburned soil. ρ_b was 29%–33% smaller in unburned soils.

4.1.3. Cross Aspect Comparisons

[16] The differences between the soils on NF versus SF aspects were essentially indistinguishable based on the particle-size data (Table 2). The UBNF soil tended to have more sand than the UBSF soil, although only by 3.5%. Gravel content was the largest difference between unburned soils on different aspects, in terms of particle size with four out of four of the UBNF samples having >15% gravel and none of the UBSF samples having >15% gravel. In contrast, the burned soils had similar gravel contents when compared across aspects. The unburned SF soils had about twice as high of LOI compared to unburned NF soils, indicating that SF unburned soils had twice as much organic matter content (Table 2). In contrast, burned soils had similar LOI values regardless of aspect, and the burned-soil LOI values were similar to the ash values (Table 2).

4.2. Soil-Water Retention

4.2.1. North-Facing Soils

[17] The soil-water retention data andvan Genuchten [1980] fits (Table 3) for the NF burned and unburned soils showed conflicting results in terms of wildfire impacts. At matric potentials wetter than −10 cm, the burned NFR site had greater soil-water retention than the unburned UBNF site while the NFM site had lower soil-water retention (Figures 2a and 2b). Shifting toward drier matric potentials, between −10 and −10³cm, both of the burned NF soils had lower soil-water retention than the unburned UBNF soil (Figures 2a and 2b). At matric potentials drier than −10³cm, differences in soil-water retention between burned and unburned soils declined (Figures 2a and 2b). The van Genuchten [1980]fits became less accurate at the dry end because of the residual soil-water content parameter, which does not allow the curve to reach zero, especially drier than −10⁵cm. The ash soil-water retention was greater than the unburned soil-water retention from matric potentials from −0.1 to −10⁴ cm (Figures 2a and 2b).

Figure 2. Soil-water retention curves for north-facing soils. FCap. denotes field capacity (i.e., soil-water content at −340 cm). PWP denotes the permanent wilting point (i.e., −1.6 × 10⁴ cm). (a) Data and model fit using the van Genuchten (VG) [1980]approach. (b) The difference in soil-water content at a given matric potential (Δθ) between unburned and burned soil samples (c) Data and model fit using the Rossi and Nimmo (RN) [1994] approach. (d) Close up of data and model fit using the Rossi and Nimmo [1994] approach for dry conditions (i.e., matric potential < −1.0 × 10⁴ cm).

Table 3. Unsaturated Model Parameters^a from Intact Soil Cores, Disturbed Samples, and Repacked Ash Sample^b

Soil Sample	Condition	θS^a^,^c (cm^3 cm^−3)	θr^a^,^c (cm^3 cm^−3)	α^a^,^c (cm^−1)	β (N)^a^,^c	RMSE^d (%)
a Parameters for the van Genuchten [1980]model of soil-water retention.θSis saturated soil-water content andθris residual soil-water content. b Locations are shown in Figure 1. c Estimated using the RETC [van Genuchten et al., 1991] program to fit to the measured soil-water retention data. d RMSE is calculated by comparing the van Genuchten [1980]model fits to the measured soil-water retention data from D.B. Stephens, not to the disturbed samples from the Decagon 5TE sensor calibration.
UBNF soil	Unburned	0.46	0.010	0.0705	1.446	0.011
NFM soil	Burned	0.42	0.019	0.0370	1.457	0.010
NFR soil	Burned	0.49	0.008	0.0397	1.401	0.004
Ash at NFM	Burned	0.58	0.027	0.0068	2.028	0.009
UBSF soil	Unburned	0.58	0.026	0.0330	1.392	0.015
SFM soil	Burned	0.50	0.021	0.0417	1.483	0.010
SFR soil	Burned	0.52	0.020	0.0582	1.352	0.007

[18] The Rossi and Nimmo [1994] fits (Table 4) were nearly identical to the van Genuchten [1980]fits from saturation to the permanent wilting point, but differed from the permanent wilting point to oven-dry, where the Rossi-Nimmo model better represents soil-water retention from adsorption processes [Rossi and Nimmo, 1994]. The Rossi and Nimmo [1994]model slightly misfit the soil-water retention data in NF soils, for example, theRossi and Nimmo [1994]fit for the ash had the lowest soil-water retention (Figure 2d). From −10⁴ to −10⁷cm, the unburned NF soil had the highest soil-water retention using theRossi and Nimmo [1994]model fit. The laboratory data based on the repacked samples, however, showed nearly indistinguishable differences in soil-water retention between the different samples when soil was dryer than field capacity. Field capacity and soil-water content at the permanent wilting point were smaller in unburned soils relative to burned soils on NF slopes, which resulted in smaller plant-available water amounts in unburned soils (Table 5). Soil-water content at saturation,θ_S, showed conflicting trends, with 9% greater θ_S in unburned soils (UBNF versus NFM) and 7% less θ_Sin unburned soils (UBNF versus NFR). The ash soil-water retention data showed that ash is capable of storing large amounts of water from saturation to a matric potential of −1.0 × 10³ cm. Ash had the highest θ_S of the soil samples (Table 3). In terms of plant-available water, ash had the highest field capacity and the second lowest soil-water content at the permanent wilting point, thus the largest plant-available water.

Table 4. Unsaturated Model Parameters for the Rossi and Nimmo [1994]Soil-Water Retention Approach^a from Intact Soil Cores, a Repacked Ash Sample, and Disturbed Samples^b

Soil Sample	Condition	θs^a (cm^3 cm^−3)	ψd^a (cm)	ψ0^a (cm)	λ^a	RMSE^c (%)
a Parameters for the Rossi and Nimmo [1994]junction model of soil-water retention.θSis saturated soil-water content. b Locations are shown in Figure 1. c RMSE is calculated by comparing the van Genuchten [1980]model fits to the measured soil-water retention data from D.B. Stephens, not to the disturbed samples from the Decagon 5TE sensor calibration.
UBNF soil	Unburned	0.446	−1 × 10^7	−6.663	0.280	0.017
NFM soil	Burned	0.419	−1 × 10^7	−13.84	0.313	0.009
NFR soil	Burned	0.493	−1 × 10^7	−17.07	0.346	0.013
Ash at NFM	Burned	0.583	−1 × 10^7	−67.04	0.488	0.013
UBSF soil	Unburned	0.584	−1 × 10^7	−19.06	0.328	0.013
SFM soil	Burned	0.501	−1 × 10^7	−12.28	0.325	0.008
SFR soil	Burned	0.521	−1 × 10^7	−12.10	0.314	0.015

Table 5. Soil-Water Content at Field Capacity, Soil-Water Content at the Permanent Wilting Point, and Plant-Available Water (PAW) for the Burned and Unburned Soil Samples^a

Soil Sample	Condition	θField capacity^b (cm^3 cm^−3)	θPermanent wilting point^c (cm^3 cm^−3)	PAW^d (cm)	Δ Unburned-Burned PAW^e (%)
a Locations are shown in Figure 1. b θField capacityis the soil-water content at −0.33 bars, which corresponds to a matric potential of approximately −340 cm [Twarakavi et al., 2009]. c θPermanent wilting pointis the soil-water content at −15 bars, which corresponds to a matric potential of approximately −1.6 × 10^4 cm [Koorevaar et al., 1983]. d Plant-available water is defined as the difference in soil-water content from field capacity (i.e., −340 cm) to the permanent-wilting point (−1.6 × 10^4 cm). e The change (Δ) in unburned − burned plant-available water (PAW) is calculated as (unburned soil-water content − burned soil-water content)/unburned soil-water content × 100.
UBNF soil	Unburned	0.118	0.029	0.089
NFM soil	Burned	0.143	0.040	0.103	−15.7
NFR soil	Burned	0.177	0.044	0.133	−49.4
Ash at NFM	Burned	0.243	0.031	0.212
UBSF soil	Unburned	0.241	0.074	0.167
SFM soil	Burned	0.153	0.041	0.112	32.9
SFR soil	Burned	0.194	0.065	0.129	22.8

4.2.2. South-Facing Soils

[19] The soil-water retention results in the SF soils showed large decreases at all matric potentials in burned, relative to unburned soils (Figure 3a). Differences between burned and unburned soil-water retention were greatest at matric potentials of −30 (SFR) to −90 cm (SFM) and were much less at matric potentials drier (i.e., less) than the permanent wilting point (Figure 3b). The Rossi and Nimmo [1994] fits (Table 4) were essentially the same as the van Genuchten [1980] fits until matric potentials were less than the permanent wilting point (Figure 3c). At matric potentials below the permanent wilting point, SF unburned soils had greater soil-water retention, based on both the laboratory data and theRossi and Nimmo [1994] fits (Figure 3d). Field capacities were 20%–30% greater in unburned soils and at the permanent wilting point, soil-water contents were 12%–41% greater in unburned soils. Plant-available water was greater in unburned soils, from 22% to 33%, in unburned relative to burned SF soils (Table 5). South-facing soils also showed differences hydraulic properties that impacted soil-water retention.θ_S was between 10% and 14% greater in unburned soils.

Figure 3. Soil-water retention curves for south-facing soils. FCap. denotes field capacity (i.e., soil-water content at −340 cm). PWP denotes the permanent wilting point (i.e., −1.6 × 10⁴ cm). (a) Data and model fit using the van Genuchten (VG) [1980]approach. (b) The difference in soil-water content at a given matric potential (Δθ) between unburned and burned soil samples (c) Data and model fit using the Rossi and Nimmo (RN) [1994] approach. (d) Close up of data and model fit using the Rossi and Nimmo [1994] approach for dry conditions (i.e., matric potential <−1.0 × 10⁴ cm).

4.2.3. Cross Aspect Comparisons

[20] This section compares the burned soils and unburned soils separately to examine differences in soil-water retention before and after wildfire between the experimental plots. Soil-water retention after wildfire had relatively small differences between experimental plots (Figure 4). The NF burned plot NFM was the exception, which had lower soil-water retention than the other burned plots. The NFM soil had very similar soil-water retention to the unburned plot, UBNF, which may be the result of the relatively low burn severity at the NFM plot, which was discerned based on partially, rather than fully, combusted litter and duff just below the soil surface. The other burned plots (i.e., NFR, SFR, and SFM) had relatively high-to-moderate burn severity at ground level [seeParsons et al., 2010], which may explain why soil-water retention was similar at these plots. Field capacity was similar between the burned plots, regardless of aspect (Table 5). This similarity between burned plots held true for the soil-water content at the permanent wilting point and plant-available water (Table 5). Soil-water retention was quite different between the unburned plots (Figure 4). The NF soil had lower soil-water retention than the SF soil, at matric potentials wetter than the permanent wilting point. Field capacity and soil-water content at the permanent wilting point were more than a factor of 2 larger for unburned SF soil compared to unburned NF soil (Table 5). Plant-available water was almost twice as large for unburned SF relative to unburned NF soil.

Figure 4. Cross-aspect comparison of soil-water retention curve data and model fit using theRossi and Nimmo (RN) [1994]approach. (a) Burned soils on north- and south-facing aspects. (b) Unburned soils on north- and south-facing aspects. Note these are the same data asFigures 2 and 3, just shown to more easily compare aspect controls before and after wildfire.

5. Discussion

[21] The following sections interpret the major versus minor factors responsible for soil-hydraulic and bulk property differences before and after wildfire.

5.1. Major Wildfire Impacts on Soil-Water Retention

[22] Wildfire impacts on soil organic matter content were likely the largest driver of changes in soil-water retention at the experimental plots. This is consistent with previous workbyBoix-Fayos [1997]. The conflicting results from different previous research on wildfire impacts on soil-water retention, showing both increases and decreases (Table 1), may potentially be explained by wildfire impacts on soil organic matter content. For example, Silva et al. [2006]saw minor changes in soil-water retention and no change in plant-available water, which coincides with their observation of little change in organic matter content resulting from wildfire.Alauzis et al. [2004]measured major declines in organic matter content and soil-water retention resulting from wildfire. Ash addition has been shown to increase organic matter content [Badìa and Martì, 2003], which may explain why the experimental plot NFM saw increases in soil organic matter and coinciding increases in soil-water retention.

[23] The organic matter content values we measured are consistent with other sites of similar elevation in the Rocky Mountains, United States. For example, unburned soil LOI values of 7.3% (2000 Hi Meadow Fire area), 6.0% (2000 Cerro Grande Fire area), 6.0% (near Golden, Colorado), and 7.0% (near Allenspark, Colorado) were reported by Moody et al. [2005]. Burned soil LOI values were 5.8% (2002 Hayman Fire) and 7.4% (2002 Missionary Ridge Fire) [Moody et al., 2005]. The reductions in organic matter content we observed are similar to slightly less than those measured by Biswas et al. [2008] at a site in Washington, United States, although the % by mass of organic matter content in unburned and burned soils at our field site was considerably lower. Organic matter consumption as a result of wildfire depends on soil temperatures reached during the fire, which depends on heat transfer [Raison et al., 1984]. Heat transfer depends on soil thermal properties, wildfire intensity, and wildfire duration [DeBano et al., 1998].The substantial reductions in soil organic matter content we observed may result from the high fire severity at some of the burned sites [see Hatten and Zabowski, 2009].

5.2. Lack of Particle Size Differences Between Plots

[24] We did not observe differences in particle-size distributions driven by wildfire at the Fourmile Canyon site (Table 2). Particle size is often cited as a major control on soil-water retention and associated soil-hydraulic properties (e.g., saturated and residual soil-water contents) [e.g.,Gupta and Larson, 1979]. Based on the data presented here, the minimal wildfire impacts on particle size cannot explain the postwildfire differences in soil-water retention. Differences in prewildfire particle-size distributions likely affect postwildfire particle-size distribution shifts. It is possible that the predominantly sand-sized particles that make up the soil at this site (Table 2) were minimally affected by wildfire impacts, compared to smaller silt or clay-sized particles.

5.3. Wildfire-Driven Homogenization of Soil-Water Retention

[25] Our results suggest that soil-hydraulic and bulk property differences between experimental plots in unburned soil, which may potentially result from aspect controls, were reduced after wildfire. This suggests that wildfire may induce a landscape-scale “homogenization” of some soil physical and hydraulic properties, which removes landscape-scale differences in soil-water retention. Our work is not the first to notice this homogenizing of soil properties following wildfire, for example,Boix-Fayos [1997] and González-Pelayo et al. [2006]found similar results for soil-water retention.Cerdà et al. [1995] noted that prefire organic matter contents were higher on NF slopes, relative to SF slopes, but that these differences were greatly reduced after wildfire. Soil microtopography differences have been shown to be homogenized following grassland fires [White, 2011]. It is already known that wildfire removes many of the vegetation community differences (e.g., interception, transpiration, and energy balances related to canopy structure) that can lead to hydrologic differences resulting from slope aspect. The high burn severity at the experimental plots in this work likely impacts the level of homogenization across aspect of soil-hydraulic properties and lower burn severity areas may still retain substantial aspect controls on soil-hydraulic properties. Further work across gradients of burn severity and aspect are needed to elucidate thresholds or continuum relationships for the level of homogenization.

5.4. Observed Trends in Soil Organic Matter Content

[26] The principal difference between the unburned soils at the Fourmile Canyon site was soil organic matter content (Table 2), where the SF soils had nearly twice as much organic matter content as the NF soils. After the wildfire, the NF and SF soils have essentially the same amount of organic matter content (Table 2), thus the reduction in soil organic matter content was much greater in the SF soils. The finding that SF unburned sites had greater organic matter content than unburned NF sites agree with other data within the Fourmile Canyon Fire study area by Moody and Nyman [2012], who measured LOI on NF-unburned soils of 3.9%, NF-burned soils of 3.4%–8.4%, SF-unburned soils of 28.1%, and SF-burned soils of 3.8%–5.8%. The work by L. Rigg (Study of soil properties along a hillslope in Parson's Parcel Boulder, CO, unpublished report, 34 pp., University of Colorado, Boulder, Colorado, 1993) in unburned soils near the Fourmile Canyon Fire study area found that NF organic content matter was 5.9% while SF organic matter content was 7.2%.

[27] The reason why the SF unburned soils have greater organic matter fractions than the NF unburned soils may reflect disparities in soil formation that are aspect driven, noted for this area by Birkeland et al. [2003], that result in a thicker, more organic-rich A horizon on SF slopes.Birkeland et al. [2003]also suggested that erosion after disturbance events in this area may result in less-developed soils in summit catena positions with different organic matter prevalence and wildfire impacts depending on slope position. We speculate that the nearly closed canopy forest with minimal understory vegetation on the NF slopes, in contrast to the sparse open trees and grassy understory on the SF slopes at the Fourmile Canyon site may result in soil development on the SF slopes that retains more organic matter, thus contributing to greater soil-water retention on the unburned SF slopes, although this speculation needs further substantiation beyond the data presented here.

5.5. Limiting Factors of This Investigation

[28] Although this work has shown that organic content can be a major factor in soil-water retention changes following wildfire, one confounding factor for our conclusions is the small number of samples analyzed. It is well known that spatial variability at the plot scale can exist for soil-water retention [Russo and Bouton, 1992], soil organic carbon [Robertson et al., 1997], and soil-particle-size distribution [Famiglietti et al., 1998]. Many investigations of soil-water retention in burned areas have small sample sizes, especially for intact cores for actual wildfire sites (seeTable 1) and eventually the collection of these studies and further work may facilitate elucidating meta-analysis [e.g.,Sankey et al., 2012]. Areas affected by wildfire can also have spatial variability in burn severity [Lewis et al., 2006] that can impact soil-water retention [Stoof et al., 2010]. Because our plot sites were selected for high burn severity based on visual observations using the criteria from Parsons et al. [2010], we believe that our results are reasonably robust with respect to the effects of high severity wildfire, but would benefit from further confirmation at other field sites outside of the Colorado Front Range.

[29] The observation reported here of greater organic matter fractions in SF-unburned soils, relative to NF-unburned soils, conflicts with other findings in the western United States. For example,Franzmeier et al. [1969] and Geroy et al. [2011] showed that unburned NF soils had greater organic matter content than unburned SF soils. Larger amounts of soil organic matter content in unburned NF soils relative to SF soils may explain the findings by Pierson et al. [2002], who reported greater reductions in organic matter content on NF slopes relative to SF slopes after a high-severity wildfire in the sagebrush foothills of the Rocky Mountains, United States. The reversed trend of aspect dependence of soil organic matter content observed in unburned soils in the area of the 2010 Fourmile Canyon Fire likely means that the speculative conclusions regarding trends of how aspect controls postwildfire organic matter contents and thus soil-water retention may be confined to this area of the Colorado Front Range. However, the finding of “homogenization” of soil organic matter contents and soil-water retention after wildfire may be more robust.

6. Conclusions

[30] We examined plot-scale differences in soil-water retention caused by wildfire effects in the area of the 2010 Fourmile Canyon Fire in the Colorado Front Range, United States. Measurements of soil-water retention, soil particle-size distributions, and organic matter contents were combined with estimates of plant-available water. Unsaturated soil hydraulic property model parameters for thevan Genuchten [1980] and Rossi and Nimmo [1994]models were estimated for future use in unsaturated flow modeling because there is a paucity of these parameters available for burned areas. Our data and analysis showed that before the wildfire, unburned soils have substantial differences in soil organic content and soil-water retention, but almost no differences in soil-particle size distribution. Wildfire tended to greatly reduce differences in soil organic matter content, and thus reduce differences in soil-water retention. This research suggests that wildfire may “homogenize” soil-water retention across the landscape for high to moderate burn severities. We speculate that this homogenization was primarily the result of combustion/degradation of organic matter and incorporation of ash into soil. This work also suggests that thevan Genuchten [1980]parameterization can represent soil-water retention in burned soils well until matric potential is <−10⁵cm or soil-water content is <0.02 cm³ cm⁻³, at which point soil-water retention models such asRossi and Nimmo [1994]that better represent dry-end soil-water retention should be employed if increased accuracy is desired.

Appendix A:  

[31] Table A1provides the matric potential ranges for each type of soil--water retention measurement. The mean equilibration times (usually exceeding 1 week equilibration per measurement point) for the hanging column and pressure plate methods are also given inTable A1.

Table A1. Methods Used to Measure Soil-Water Retention Curves^a

Soil Samples	HC	PP	DPP	RH
a HC is the hanging column method; PP is the pressure plate method; DPP is the dewpoint potentiometer method; RH is a controlled relative humidity chamber; UBNF mean equilibration time for the HC method was 7.5 days; NFM mean equilibration time for the HC and PP methods was 6.4 days; NFR mean equilibration time for the HC method was 7.5 days; ash at NFM mean equilibration time for the HC method was 6.4 days; UBSF mean equilibration time for the HC and PP method was 8.2 days; SFM mean equilibration time for the HC and PP methods was 6.7 days; SFR mean equilibration time for the HC method was 7.5 days. ψ is the matric potential.
UBNF soil
ψ (cm)	0, 5, 10, 52, 194		3.16 × 10^3, 2.07 × 10^4, 9.45 × 10^4	8.57 × 10^5
NFM soil
ψ (cm)	0, 7, 20, 76	337	1.37 × 10^4, 5.67 × 10^4, 2.44 × 10^5, 3.54 × 10^5, 4.24 × 10^5	8.57 × 10^5
NFR soil
ψ (cm)	0, 4, 11, 50, 195		2.45 × 10^3, 3.68 × 10^4, 1.53 × 10^5	8.57 × 10^5
Ash at NFM
ψ (cm)	0,12,30,99	337	1.72 × 10^4, 5.56 × 10^4, 1.68 × 10^5, 2.61 × 10^5, 3.81 × 10^5	8.57 × 10^5
UBSF soil
ψ (cm)	0, 6, 19, 73	337	6.02 × 10^3, 1.35 × 10^4, 4.48 × 10^4	8.57 × 10^5
SFR soil
ψ (cm)	0, 4, 10, 49, 197		6.43 × 10^3, 4.63 × 10^4, 9.21 × 10^4	8.57 × 10^5
SFM soil
ψ (cm)	0, 7, 20, 73	337	1.41 × 10^4, 1.99 × 10^4, 5.20 × 10^4, 1.50 × 10^5, 2.63 × 10^5, 3.82 × 10^5, 4.38 × 10^5	8.51 × 10^5

Acknowledgments

[32] The presentation of this work benefitted from discussions with Brian Andraski, Rick Healy, Eve-Lyn Hinckley, Deborah Martin, Ben Mirus, John Moody, John Nimmo, and David Stonestrom. Kim Perkins is gratefully acknowledged for completing the laboratory laser diffraction analysis of the particle-size data and for providing thoughtful comments on the manuscript. Two anonymous reviewers made substantial contributions to increasing manuscript clarity. Brian Ebel received support from the Mendenhall Postdoctoral Fellowship Program in the National Research Program and the Climate and Land Use Change Program of the U.S. Geological Survey. Field assistance from John Moody and Petter Nyman was very helpful. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.