Accepted Manuscript

Modeling the effects of surface storage, macropore flow and water repellencyon infiltration after wildfire

Petter Nyman, Gary J. Sheridan, Hugh G. Smith, Patrick N.J. Lane

PII: S0022-1694(14)00152-8DOI: http://dx.doi.org/10.1016/j.jhydrol.2014.02.044Reference: HYDROL 19436

To appear in: Journal of Hydrology

Received Date: 8 May 2013Revised Date: 15 December 2013Accepted Date: 12 February 2014

Please cite this article as: Nyman, P., Sheridan, G.J., Smith, H.G., Lane, P.N.J., Modeling the effects of surfacestorage, macropore flow and water repellency on infiltration after wildfire, Journal of Hydrology (2014), doi: http://dx.doi.org/10.1016/j.jhydrol.2014.02.044

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customerswe are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting proof before it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Title: 1

Modeling the effects of surface storage, macropore flow and water repellency on 2

infiltration after wildfire 3

Keywords: infiltration; water repellency; macropore flow; soil moisture; wildfire; erosion 4

5

Author names and affiliations: 6

Petter Nymana,b,c (nymanp@unimelb.edu.au) 7

a Melbourne School of Land and Environment, The University of Melbourne, 221 Bouverie St, 8

Parkville, 3010 Victoria, Australia 9

b eWater Cooperative Research Centre, The University of Melbourne, Parkville, Victoria 3010, 10

Australia 11

c Bushfire Cooperative Research Centre, 340 Albert Street, East Melbourne, Victoria 3002, 12

Australia 13

14

Gary J Sheridana,c (Sheridan@unmelb.edu.au), 15

a Melbourne School of Land and Environment, The University of Melbourne, 221 Bouverie St, 16

Parkville, 3010 Victoria, Australia 17

c Bushfire Cooperative Research Centre, 340 Albert Street, East Melbourne, Victoria 3002, 18

Australia 19

20

Hugh G Smithd (Hugh.Smith@liverpool.ac.uk) 21

d School of Environmental Sciences, University of Liverpool, Liverpool, United Kingdom 22

23

24

2

Patrick NJ Lanea,c (patrickl@unimelb.edu.au) 25

a Melbourne School of Land and Environment, The University of Melbourne, 221 Bouverie St, 26

Parkville, 3010 Victoria, Australia (nymanp@unimelb.edu.au) 27

c Bushfire Cooperative Research Centre, 340 Albert Street, East Melbourne, Victoria 3002, 28

Australia 29

30

31

Corresponding author: 32

Petter Nyman (nymanp@unimelb.edu.au) 33

221 Bouverie St 34

Parkville 35

3010 Victoria 36

Australia 37

Ph: +61 408584676 38

39

40

3

Abstract 41

Wildfires can reduce infiltration capacity of hillslopes by causing i) extreme soil drying, ii) 42

increased water repellency and iii) reduced soil structure. High severity wildfire often results in a 43

non-repellent layer of loose ash and burned soil overlying a water repellent soil matrix. In these 44

conditions the hydraulic parameters vary across discrete layers in the soil profile, making the 45

infiltration process difficult to measure and model. The difficulty is often exacerbated by the 46

discrepancy between actual infiltration processes and the assumptions that underlie commonly used 47

infiltration models, most of which stem from controlled laboratory experiments or agricultural 48

environments, where soils are homogeneous and less variable in space and time than forest soils. 49

This study uses a simple two-layered infiltration model consisting of surface storage (H), macropore 50

flow (Kmac) and matrix flow (Kmat) in order to identify and analyze spatial-temporal infiltration 51

patterns in forest soils recovering from the 2009 Black Saturday wildfires in Victoria, southeast 52

Australia. Infiltration experiments on intact soil cores showed that the soil profile contained a 53

region of strong water repellency that was slow to take on water and inactive in the infiltration 54

process, thus restricting flow through the matrix. The flow resistance due to water repellent soil was 55

represented by the minimum critical surface tension (CSTmin) within the top 10 cm of the soil 56

profile. Under field conditions in small headwaters, the CSTmin remained in a water repellent 57

domain throughout a 3-year recovery period, but the strength of water repellency diminished 58

exponentially during wet conditions, resulting in some weather induced temporal variation in 59

steady-state infiltration capacity (Kp). An increasing trend in macropore availability during recovery 60

was the main source of temporal variability in Kp during the study period, indicating (in accordance 61

with previous studies) that macropore flow dominates infiltration processes in these forest soils. 62

Storage in ash and burned surface soil after wildfire was initially high (~ 4 mm), then declined 63

exponentially with time since fire. Overall the study showed that the two layered soil can be 64

4

represented and parameterized by partitioning the infiltration process into surface storage and flow 65

through a partially saturated and restrictive soil layer. Ash, water repellency and macropore flow are 66

key characteristics of burned forest soils in general, and the proposed model may therefore be a 67

useful tool for characterizing fire impact and recovery in other systems. 68

69

Keywords: infiltration; water repellency; macropore flow; soil moisture; wildfire; erosion 70

71

72

5

1 Introduction 73

Fire can increase overland flow by reducing interception, infiltration and surface roughness (Martin 74

and Moody, 2001; Robichaud, 2000; Shakesby and Doerr, 2006; Sheridan et al., 2007). Increased 75

production of overland flow can in turn lead to increased erosion rates (Cerda and Lasanta, 2005; 76

Lane et al., 2006a; Moody and Martin, 2001; Robichaud et al., 2008a; Shin et al., 2013; Smith et al., 77

2011; Wagenbrenner and Robichaud, 2013) and increased frequency of threshold driven responses 78

such as flash floods and debris flows (Cannon, 2001; Cawson et al., 2012; Kean et al., 2013; Nyman 79

et al., 2011). High severity wildfire (i.e. crown fire) removes vegetation, burns the topsoil and 80

deposits ash on hillslopes. Under these conditions the surface roughness and the rates of 81

interception are low and relatively homogenous across hillslopes, irrespective of the catchment 82

conditions prior to burning (Johansen et al., 2001). Infiltration however can be highly variable due 83

to a strong dependency on pre-fire soil properties. Soil properties such as porosity, pore-size 84

distribution, macroporosity and water repellency are therefore important controls on variation in 85

hydrological responses across burned landscapes (Larsen et al., 2009; Nyman et al., 2011; 86

Robichaud et al., 2007; Shakesby and Doerr, 2006). 87

88

Infiltration models use theory of flow in porous media to estimate the rate at which water enters the 89

soil. Essentially the infiltration rate is modeled as a function of i) the pore-size distribution of the 90

soil matrix, ii) the initial soil moisture and iii) the rate at which water is supplied at the surface 91

(Green and Ampt, 1911; Philip, 1957). Hydraulic conductivity (mm h-1), sorptivity (mm h-0.5), or 92

the suction at the wetting front (mm) are infiltration parameters that reflect the combined effects of 93

these properties on flow and retention of water within the soil (Smith et al., 2002). These infiltration 94

parameters can be obtained from laboratory studies, field experiments or pedotransfer functions 95

(Cook, 2007; Moody et al., 2009; Rawls et al., 1983; Risse et al., 1994; Robichaud, 2000). Models 96

6

of infiltration are process-based and represent the physical processes contributing to flow and 97

storage of water in the soil. However, the models are highly idealized and there remain large gaps in 98

their capacity to represent the actual infiltration processes in forest soils, where infiltration is 99

characterized by preferential flow and non-uniform wetting fronts (Beven and Germann, 2013). 100

101

The effect of burning on infiltration rates is well documented in the literature. Fire impacts on 102

infiltration parameters by i) adding surface storage capacity as deposits of fine ash and burned soil 103

(Bodí et al., 2012; Cerdà and Doerr, 2008; Woods and Balfour, 2008; Woods and Balfour, 2010), ii) 104

reducing soil structure and macropore flow (Nyman et al., 2010; Onda et al., 2008) and iii) reducing 105

pore-space availability and wettability due to water repellent soils (Cerdà and Doerr, 2007; Doerr 106

and Moody, 2004; Moody and Ebel, 2012; Nyman et al., 2010). Soil profiles on fire affected 107

hillslopes typically consist of heated and burned soil that is sandwiched between ash at the surface 108

and an underlying soil matrix that is unaffected by the fire. The layered soil profile means that there 109

is often strong variability with depth for soil properties such as porosity, particle size distribution 110

and water repellency (Bodí et al., 2012; Ebel, 2012; Ebel et al., 2012; MacDonald and Huffman, 111

2004; Moody and Ebel, 2012; Moody et al., 2009; Stoof et al., 2010; Woods et al., 2007). 112

Characterizing soil hydrological properties and their effects on the infiltration process in these 113

systems is challenging because it requires simultaneous examination of flow processes in soil layers 114

with different media properties (Ebel and Moody, 2013; Moody et al., 2013). 115

116

The properties that dominate infiltration change depending on the spatial and temporal scales at 117

which processes are measured. Water repellency for instance can be quantified as a spatial 118

distribution of a point-based measurement of water drop penetration times (Doerr et al., 1998; 119

Robichaud et al., 2008b; Woods et al., 2007). The water repellency has strong effect on the 120

7

behavior of water drops, reducing the ability of the soil to absorb water (Doerr et al., 2000). 121

However, the strength or persistence of water repellency at points may not translate to large impacts 122

on infiltration if water bypasses the matrix as preferential flow through wettable patches, cracks, 123

and macropores and along roots and rocks (Burch et al., 1989; Doerr and Moody, 2004; Granged et 124

al., 2011; Imeson et al., 1992; Nyman et al., 2010; Shakesby and Doerr, 2006; Urbanek and 125

Shakesby, 2009). Similarly, the temporal scale of measurement is important. Moisture-induced 126

changes to water repellency for instance is important when infiltration is modeled across different 127

seasons, but it might be negligible within rain storms since the time scale of imbibition in water 128

repellent soil may be in the order of several hours to days (Crockford et al., 1991; Ebel et al., 2012; 129

Moody and Ebel, 2012). 130

131

Representing the interactions between macropore flow, matrix flow and imbibition is important for 132

understanding and predicting fire-impacts on infiltration processes. Most infiltration models, 133

however, are based on theory and data from systems where the dominant processes and key 134

properties are different from what is typically observed in fire-affected soils, particularly with 135

regards to wetting behavior (imbibition) and macropore flow (Ebel and Moody, 2013; Nyman et al., 136

2010). In this paper we therefore aim to develop a model for hydraulic conductivity which 137

incorporates moisture dependent water repellency dynamics and which accounts for changes in 138

macropore flow during recovery from wildfire. The study combines field campaigns and laboratory 139

measurements to: 140

1. Model the interactions between imbibition and hydraulic conductivity of intact soil cores 141

that were water repellent. 142

2. Quantify the effects of seasonal weather, water repellency, surface storage and macropore 143

flow on storage and steady-state infiltration in catchments recovering from wildfire. 144

8

The study was conducted at sites in Victoria, southeast Australia, burned by the 2009 Black 145

Saturday wildfires and the 2006 Great Divide Wildfires. Previous work from the region shows that 146

macropore flow and water repellency are important controls on infiltration (Burch et al., 1989; 147

Crockford et al., 1991; Lane et al., 2006b; Nyman et al., 2010; Prosser and Williams, 1998; 148

Shakesby et al., 2003; Sheridan et al., 2007; Smith et al., 2011). 149

150

2 Methods 151

2.1 Infiltration model 152

A large proportion of post-fire erosion tends to occur in response to high intensity rainfall events 153

(Ebel et al., 2012; Kean et al., 2011; Nyman et al., 2011; Smith et al., 2011). This study assumes 154

that infiltration during these types of events is determined by storage of water in surface material 155

and the flow of water out of this storage and into the soil matrix. Once storage is depleted, the 156

maximum infiltration capacity (Kp) occurs under ponded conditions and is either controlled by i) 157

supply rate of water (R), ii) the hydraulic conductivity of ash/burned soil mixture (Kash), or iii) the 158

sum of steady-state infiltration through macropores (Kmac) and the matrix (Kmat) (Figure 1). 159

160

Field- and laboratory-based infiltration measurements were used to parameterize and analyze a 161

storage based infiltration model which represents infiltration, f(t), as a two stage process; 162

163

���� = ��� �⁄ + �� � ����� < �������� �� > �� ��

���� = �� � + �� � ����� > �������� �� > �� ��

(1)

164

where t is time, tp is time to ponding, H is the surface storage (mm), θe is the effective saturation, 165

Kmat is the effective hydraulic conductivity of the soil matrix (mm h-1) and Kmac is the macropore 166

9

flow (mm h-1) given unlimited supply (Figure 1a). Effective saturation (θe) is the difference between 167

soil moisture at saturation (or porosity, θs) and the initial soil moisture (θi) relative to θs, θe = (θs - 168

θi)/θs. Macropore flow, Kmac, is the difference in steady-state infiltration between h = -15 mm and h 169

= 5 mm. 170

171

<<<<<<<<<<<<<<<<<<Figure 1 here >>>>>>>>>>>>>>>> 172

173

The surface storage volume, V (mm3), above and within the water repellent layer was estimated 174

from: 175

� =

����

� − �� �

∗ � (2)

176

Where R is the supply rate of water (mm h-1) to the soil surface and A is the area under infiltration. 177

The potential supply from infiltrometers is high relative to Kmat (R >> Kmat), hence V → AHθe 178

(Kirkby, 1975; Scoging, 1979). 179

180

Water repellency can vary with changes in soil moisture. A separate model was used to represent 181

the infiltration into this layer since wetting processes are slow compared to non-repellent soil 182

(Moody and Ebel, 2012). Water repellency was represented as a distribution of critical surface 183

tension, CST (dyn cm-1) at soil depths, ds, between 0 and 10 cm using the function: 184

185

!"���� = 72.7 + &�� ∗ '() (3)

10

The fitted parameters v and u determine the maximum strength and the distribution of water 186

repellency (CST) in the soil profile. Eq. (3) represents the soil surface as wettable (i.e. CST = 72.7 187

dyn cm-1 when ds = 0), an assumption that was supported by data from field measurements on 188

burned soil. Using Eq. (3) to minimize the error in the spatially distributed CST measurements 189

means that both vertical and planar variability were captured in a single function. The maximum 190

strength of water repellency within the soil profile, CSTmin, (the ‘bottleneck’ if ash is not limiting 191

flow) was obtained by setting the first derivative of CST(ds) in Eq. (3) to 0, that way calculating the 192

depth of maximum repellency (dmax): 193

194

�� * = −1/log�&� (4)

The slow wetting process in water repellent soils means that the interaction between soil moisture 195

and CSTmin are mostly dependent on weather conditions spanning over days to weeks. For field 196

conditions, the empirical Keetch-Byram Drought Index (KBDI; Keetch and Byram, 1968) was used 197

as a predictor of moisture status in the catchments and hence temporal dynamics in CSTmin: 198

199

200

201

202

P is the daily precipitation (mm), Tmax is the daily maximum temperature (°C) and Pyrs is the annual 203

precipitation at the site (mm). The temporal variability in Kmat was measured and represented as an 204

empirical function of CSTmin and KBDI. 205

206

In the following section we describe a set of laboratory and field experiments that were designed to; 207

�012� = �012�34 − 106 +

�2000− �012�34��0.9679�:.:;;<=>?@4.AAB� − 0.83

�1 + 108893:.::4EFGH)�1000

(5)

11

1. Evaluate the assumption that hydraulic conductivity of the soil matrix is restricted by water 208

repellent layer (Figure 1) and, 209

2. Quantify the effects of i) regional weather conditions, ii) soil moisture status, iii) water 210

repellency, and iv) macropore availability on infiltration processes in soils recovering from 211

wildfire. 212

2.2 Laboratory and field measurements: overview of methods 213

Laboratory- and field-based experiments were used to measure infiltration at three fire affected 214

hillslopes in Victoria, southeast Australia (Figure 2 and Table 1). The primary study sites were 215

Sunday Creek and Stony Creek, which were burned at high severity during the Black Saturday 216

wildfires in February 2009 (Cruz et al., 2012). A third site, Ella Creek, was burned by the Great 217

Divide wildfires December 2006, and was included to represent similar forest environments in a 218

later stage of recovery. All sites are located in the eastern uplands of Victoria and were burned 219

under wildfire conditions where the forest canopy was completely burned (burn severity class 1; 75-220

100 % Crown consumption) (see definitions in Cruz et al., 2012). The southeast Australian region 221

experiences a Mediterranean climate with hot and dry summer and cool and wet winters. There is 222

large variability in ecosystem properties across the study area, but the three sites used in this study 223

were all characterized by dry eucalyptus forest and were similar in terms of rainfall, aspect and 224

solar exposure. See Nyman et al., (2011) for a general description of the geomorphology, vegetation 225

and fire regimes in the region. 226

227

Intact soil cores were sampled from each site in April 2010 and used for laboratory-based 228

infiltration measurements that quantify the relation between imbibition rate and hydraulic 229

conductivity. Laboratory experiments were best suited for this type of study because soil moisture 230

conditions could be controlled. Infiltration rates were also sampled in the field during campaigns 231

12

that were aimed at identifying key controls on infiltration in soils that were recovering from 232

wildfire. Initial soil moisture and water repellency were measured alongside measurements of 233

surface storage, H (mm), matrix flow, Kmat (mm h-1) and macropore flow Kmac (mm h-1). The field 234

measurements were made in headwater catchments (~ 2 ha) during a 3-year recovery period from 235

wildfire. Daily temperature and rainfall were measured at each site and used to calculate KBDI 236

using Eq. (5). 237

238

<<<<<<<<<<<<<<<<Figure 2 here>>>>>>>>>>>>>>>> 239

240

2.3 Laboratory study: Flow and imbibition in water repellent soil 241

Intact cores were collected from Ella Creek, Sunday Creek and Stony Creek and used to measure 242

hydraulic conductivity and absorption rate at different stages of imbibition on a tension table. Nine 243

cores (~8 cm deep and 5.3 cm in diameter) were sampled from each site at upper, mid and lower 244

hillslope positions in April 2010. The soil cores were left to air dry in the laboratory (4 weeks in the 245

lab at 20 degrees), then weighed before the first set of infiltration measurements. The infiltration 246

rate was measured at 15 mm tension (h = -15 mm) using Mini-disc (MD) infiltrometer (Decagon) 247

(Moody et al., 2009; Robichaud et al., 2008b) and a lab procedure for measuring unsaturated 248

hydraulic conductivity (Cook, 2007). Measuring infiltration under slight tension ensures that values 249

were representative of flow processes in the soil matrix, excluding gravity driven flow in 250

macropores, which can obscure the effect of water repellency on flow within the matrix. 251

252

The cores and the infiltrometer were held in place using clamp and stands during infiltrations. 253

Cheesecloth and a shallow layer of pre-wetted contact sand (300 um < D < 600 um and Ks > 1000 254

mm h-1) was placed on the soil surface to ensure good contact between the disc and the soil. In the 255

13

early stages of infiltration, the rate was usually high, and then declined towards some low steady-256

state, usually within 5 minutes. The water level was read from the reservoir every 30 seconds for 35 257

minutes (min) and only recorded if there was a change from the previous reading. The core was 258

reweighed at 35 min and replaced in the clamp for another 15 min of infiltration measurements if 259

the wetting front had not appeared at the bottom of the core. If the wetting front had reached the 260

bottom of the core at 35min, then the core was placed on a buchner funnel set to a tension of -15 261

mm (same as the mini-disc tension chamber). Having the same pressure at the bottom and top of the 262

core ensures a uniform hydraulic gradient throughout the core and allowed for direct estimation of 263

hydraulic conductivity (Cook, 2007). The last 15 min of infiltration was used as a measure of the 264

effective hydraulic conductivity of the matrix (Kmat) for given initial moisture conditions (θi). 265

266

After the first set of infiltration experiments, the cores were immediately placed onto a tension table 267

with the tension set so that the average water pressure (h) for the cores was equal to -15 mm. The 268

tension table was set up with large reservoir of water and constant head burette to maintain constant 269

tension and accommodate any water that was lost due to the water uptake by the cores. Each core 270

was re-weighed after 36 hrs on the tension table and a new set of infiltrations were carried out. The 271

cores were replaced onto the tension table and reweighed at 60, 100, 148, 268 and 316 hrs. The 272

multiple time steps at which the mass was obtained indicated changes in moisture content with time, 273

and helped determine a suitable point at which to run a new set of infiltrations. 274

275

A new set of infiltrations required a new initial soil moisture status of the cores. However, after 268 276

hrs of wetting at h = -15 mm, it was evident that some cores were still dry on the surface. After this 277

time step the water level in the tension table was therefore raised to the midpoint of the core (mean 278

h = 0 mm) in attempts to reach a higher moisture levels and ultimately soil saturation. A third and 279

14

fourth set of infiltration measurements were carried out after 374 hrs and 681 hrs of wetting. After 280

the final infiltration, all cores appeared to be saturated and there was no further increase in mass of 281

cores from this point onwards. The final set of infiltrations was taken to represent the saturated 282

hydraulic conductivity (Ks) of the soil matrix when h = -15mm. The small tension means that 283

measurements exclude the effect of macropores that contribute to flow when h > -15mm. 284

285

The cores were finally replaced onto the tension table and drained at 5 tension intervals (-2.5 < h < -286

50 cm) in order to characterize the size distribution of pores in the meso-macropore domain. The 287

porosity of the soil at each site was estimated from the saturated volumetric water content θs (cm3 288

cm-3) of the 9 cores that were used in the laboratory infiltration experiment. The pore-sixe 289

distribution was estimated from the change in water content at 5 increments of h between 0 and -50 290

cm, and using the capillary equation to relate h to pore radius (r, mm): 291

292

� =

2IJℎ

cos�N� (6)

Where ρ is the density of water, g is gravity and α is the contact angle. A cumulative distribution 293

function was then used to quantify the proportion of pore-space occupied by macropores (r > 0.5 294

mm) (Nyman et al., 2010). 295

296

2.4 Infiltration properties of headwater catchments during recovery from wildfire 297

Infiltration was measured in the field under both positive (h = 5 mm) and negative (h = -15mm) 298

pressure, using the min-disc tension infiltrometer (Decagon) and a custom-designed ponded 299

infiltrometer (Perroux and White, 1988) with same disc diameter as the tension infiltrometer. The 300

measurements were made in three headwater catchments during 4 sampling campaigns, each taken 301

to represent burned systems in different stages of recovery and in different seasons (Figure 3). 302

15

Infiltration in each headwater catchment was measured at 4 points in 3 quadrats (1 m × 1 m) along 3 303

transects (80 m long) running perpendicular to the contour (n = 36). Quadrats were positioned at 304

upper- (0 m), mid- (40 m) and lower (80 m) sections of the transects. Each campaign was conducted 305

on allocated strips, 2 m to the side of the previous campaign to avoid disturbance from previous 306

campaigns. 307

308

A retention ring (5.3 cm in diameter) was inserted 2-3 cm into the soil in order to prevent lateral 309

flow through non-repellent surface material which often covered the hillslope. The soil surface was 310

relatively smooth and the infiltrometer disc was small and no contact material was required to 311

achieve contact between the soil and the disc of the tension infiltrometer. The depth of infiltrated 312

water was measured with the tension infiltrometer at 15 s intervals in the first minute, then at 30 s 313

intervals for the remaining 10-15 min of infiltration, after which the infiltrometer was removed and 314

replaced with the ponded infiltrometer. The soil within the retaining cylinder was then flooded to 315

achieve 5 mm of ponding before measuring the ponded infiltration rate at 15 s intervals for 5-10 316

min. 317

318

<<<<<<<<<<<<<<<<<<Figure 3>>>>>>>>>>>>>>>>> 319

320

Soil was collected at 5 depth intervals (0-1, 1-3, 3-5, 5-7.5, 7.5-10 cm) at 10 m intervals along each 321

of the three transects where infiltration was measured. The samples were a composite of 2 cores (10 322

cm x 5.5 cm) collected in the 1 m quadrat where infiltration was measured. The samples were 323

placed in sealed bags and transported back to the laboratory and weighed. Water repellency was 324

measured on samples in the laboratory using ethanol solutions of different concentrations (0 - 6 M; 325

0.4 M intervals), and calculating the critical surface tension (CST) once three drops of solution 326

16

penetrated the soil in 3 s (CST test) (King, 1981; MacDonald and Huffman, 2004). The soil was 327

mixed inside the sealed sampling bag before a subsample was placed on a petri dish. Gravel and 328

large organics were removed manually from the petri dish before placing the drops of ethanol 329

solution on the soil surface. 330

331

The sub-sampled soil was returned to the original sample, which was then oven dried at 105 degrees 332

for 48 hrs in order to calculate the volumetric water content. For samples collected in March 2010, 333

the soil was left to air-dry first so that the air-dry CST of each site could be obtained, using the same 334

methods described above. The water repellency was often high in the 0-1 cm depth interval. 335

However, the undisturbed soil surface always seemed to be wettable and absorbed some water 336

during infiltration measurements in the field. Measurements of CST were therefore carried out at the 337

soil surface during each field campaign in order to determine CST at ds = 0. Litter was removed 338

while ensuring that the soil was not disturbed prior to the test. The surface material was usually a 339

mix of mineral soil, ash and organic material and was non-repellent in 95-100 % of sampled 340

locations at each site. 341

342

2.5 Linking point and plot scale infiltration processes 343

Infiltration was measured in March 2010 at plots (2 m x 1.5 m) during 3 replicate rainfall 344

simulations at Sunday Creek (Figure 2). The plots were located on steep hillslopes (28°) in dry 345

eucalyptus forest with clay loam soil that was burned by wildfire in February 2009. The rainfall 346

simulation procedure has been described in Sheridan et al., (2007). The rainfall rate (target 100 mm 347

h-1) was first calibrated with steady-state runoff from a plastic sheet covering the plot. The sheet 348

was removed and discharge was measured in 0.5 l containers at regular intervals during a 30-minute 349

17

runoff period. An adjacent 1m x 1m plot was used as a test area for measuring infiltration at 4 350

points per rainfall simulation plot. 351

352

3 Results 353

3.1 Laboratory study: Flow and imbibition in water repellent soil 354

The soils were water repellent at all sites under air dry conditions (Figure 4). The critical surface 355

tension (CST) was highly variable at each sampling depth apart from the lowest depth interval (7.5-356

10) were the CST was usually close to 72 (i.e. non-repellent). The spatial variability in CST within 357

the most hydrophobic region of the soil was more strongly skewed at Ella Creek (Figure 4c) than at 358

the other two sites (Figure 4a and b). The different patterns of variability essentially show that water 359

repellency was stronger and more homogenous in the recently burned sites. The CST(ds) function, 360

Eq. (3), explained 31, 17 and 24 % of variability in water repellency of air-dry soils at Sunday 361

Creek, Stony Creek and Ella Creek, respectively, when fitted across all sampling points (n = 36). 362

The parameter optimization was highly significant (p << 0.01) for all sites. 363

364

When fitted to the mean CST, the function, Eq. (3), consistently explained more than 80 % of the 365

variability in CST with depth, indicating that spatial variability across the hillslope, and not 366

variation with depth, was the main source of residual error when the function was fitted across all 367

sampling points. The depth of minimum CST (i.e. maximum water repellency) in air dry soils was 368

estimated from Eq. (4) to be 1.2, 1.8 and 1.3 cm for Sunday Creek, Stony Creek and Ella Creek 369

respectively. The corresponding minimum value for CST, the CSTmin, was 40.1, 46.5 and 52.4 dyn 370

cm-1. 371

372

<<<<<<<<<<<<<<<<Figure 4>>>>>>>>>>>>>>> 373

18

374

The porosity and pore-size distribution values were obtained from the intact cores once the 375

infiltration experiments were completed and the cores were completely saturated (h = 0 mm). The 376

porosity at Sunday Creek and Stony Creek was 0.34 and 0.41, respectively (Table 2). A higher 377

proportion of macropores (r > 0.5 mm = 4.7 % of θs) at Sunday Creek resulted in higher matrix flow 378

(Kmat at saturation) at Sunday Creek (66 mm h-1) than at Stony Creek (52 mm h-1) despite Stony 379

Creek having higher overall porosity (Table 2). The porosity at Ella was higher (0.55) and the 380

matrix flow at saturation was ~4 times higher than the two other sites (Table 2). Porosity was 381

inversely related to gravel content. The hydraulic conductivity of repacked cores with wettable 382

surface material (ash, gravel and soil) was 41 mm h-1 and 21 mm h-1 for Sunday Creek and Stony 383

Creek respectively (Table 2). At saturation the mixture of ash, gravel and burned soil had a water 384

holding capacity of 0.40 cm3cm-3. 385

386

The volumetric soil moisture content, θi (cm3cm-3), of air dry cores from Stony Creek, Sunday 387

Creek and Ella Creek were 0.014, 0.022 and 0.029, respectively. In subsequent measurements the 388

soil moisture was consistently higher at Stony Creek than Sunday Creek but highest at Ella Creek 389

(Figure 5a), reflecting an increasing trend in porosity. The effective hydraulic conductivity of the 390

matrix Kmat was low relative to the saturated conductivity of the matrix (Ks) for all sites in the three 391

first three sets of infiltration measurements (Figure 5b). The largest change in Kmat occurred 392

between the third and the fourth set of infiltration experiments with cores from Ella Creek 393

displaying a larger increase than the two other sites (Figure 5b). 394

395

The volumetric water uptake (∆θ) is the change between the initial and final water content (θi and θf 396

, respectively) during the first 35 minutes of infiltration, excluding the storage component, V, on the 397

19

soil surface, ∆θ = θi - θf - V. Surface storage volume during early (non-steady) stage of infiltration 398

was obtained from H, a fitted infiltration parameter in Eq. (1). The change in water content of cores 399

before and after 35 min infiltrations was relatively large (0.08 < ∆θ < 0.15) for first set of 400

infiltration experiments (Figure 5c).The water uptake ∆θ during infiltrations was similar for the 401

second and third set of measurements (0.04 < ∆θ < 0.07) (Figure 5c). In the fourth and last set of 402

infiltrations the matrix was completely saturated, and ∆θ approached zero (Figure 5c). At the last 403

stage of wetting the Kmat was considered to represent Ks of the soil matrix when h = -15mm (Table 404

2). 405

<<<<<<<<<<<<<<<Figure 5 here>>>>>>>>>>>>>>>> 406

407

The relationship between initial soil moisture (θi), effective hydraulic conductivity (Kmat) and rate of 408

water uptake (∆θ/thrs) during different stages of imbibitions were combined across the three study 409

sites by normalizing Kmat by Ks, and ∆θ and θi by total porosity, θs (Figure 6a). The rate of water 410

uptake, (∆θ/θs)/thrs, declined linearly with increasing initial soil moisture θi/θs (Figure 6b) This 411

decline represents the effects of i) decreasing pore-space availability due to water repellent soil and 412

ii) an overall decline in available pore space . The linear relation means that the water uptake is 413

proportional to initial soil moisture (∆θ/∆t = - λθi), thus giving rise to an exponential relation 414

between soil moisture and the duration of infiltration. However, the hydraulic conductivity 415

remained relatively low (Kmat/Ks < 0 .4) and constant while θi/θs < 0.6, indicating that some sections 416

of the soil remained dry and hence inactive in the overall flow process. The hydraulic conductivity 417

Kmat increased exponentially with increasing soil moisture when θi/θs > 0.6. The change in hydraulic 418

conductivity with θi/θs could be represented by an exponential function which asymptoted at some 419

minimum flow in dry and repellent soils. In summary the patterns of flow and absorption shows that 420

the hydraulic conductivity was dependent on the rate at which pore space was activated and 421

20

introduced to the infiltration process. The rate of at which the soil matrix was activated was strongly 422

dependent on initial soil moisture conditions. The data support the assumptions underlying the 423

conceptual model of water repellency acting as a restriction on flow through the soil matrix (Figure 424

1). 425

426

<<<<<<<<<<<<<<<<<<Figure 6 here>>>>>>>>>>>>>>>> 427

428

3.2 Infiltration in burned headwater catchments 429

Storage, H, was obtained for headwaters at each measurement campaign by fitting Eq. (1) to the 430

mini-disc infiltration measurements (mm h-1) (n= 36) (Table 3). The matrix flow, Kmat in Eq. (1) was 431

obtained from steady-state infiltration (t > 6 min) (Table 3) and treated as a fixed parameter in the 432

fitting procedure. Soil moisture was highest for all sites in September 2010 after a relatively wet 433

spring period (Table 3). The normalized soil moisture decreased linearly (R2 = 0.61) with increasing 434

KBDI (Figure 7a). Overall, the moisture content remained low relative to θs despite KBDI 435

approaching 0, and the range of soil moisture contents (0.10 < θi/θs < 0.43) (Figure 7a) was low 436

compared to the range achieved under laboratory conditions (Figure 6a). There was no relationship 437

between initial soil moisture and the matrix flow Kmat (Figure 7b), a result which may have been 438

expected given that laboratory measurements (Figure 6) showed that Kmat is invariant with θi when 439

θi/ θs < 0.6. 440

441

The CSTmin (dyn cm-1) ranged from 24.8 - 43.1, 42.3 - 63.8 and 29.2 - 58.2 at Sunday Creek, Stony 442

Creek and Ella Creek respectively (Table 3). The relation between CSTmin with KBDI was scattered 443

when sites were combined without adjusting for the different background repellency. The aim 444

however was to generalize the relation between KBDI, CSTmin and hydraulic conductivity across dry 445

21

eucalyptus forest recovery from wildfire. The CSTmin was therefore normalized by the lowest CSTmin 446

value (i.e. strongest water repellency) measured under field conditions, that way producing a 447

relative measure (CST+

min) which could be represented as a function of KBDI (Figure 7c). Water 448

repellency increased with increasing KBDI although the soil remained repellent even when KBDI 449

was close to 0, indicating the water repellency is likely to be present during most weather 450

conditions for these forest systems. In terms of infiltration, the persistence of water repellency 451

meant that flow was always restricted by water repellent soil, and that the true saturated hydraulic 452

conductivity (Ks) could not be measured under field conditions. 453

454

The strong relation between Kmat and CSTmin indicate that water repellency was an important source 455

of variability when soil moisture levels were lower than water repellency thresholds (Figure 7d). 456

The matrix flow Kmat was most sensitive to changes in water repellency status for CSTmin > 40. The 457

flow under non-repellent conditions (Ks) was estimated to be ~50 mm h-1 by extrapolating a value 458

for Kmat at CSTmin = 72.7 dyn cm-1 (Figure 7d). The extrapolated value corresponds well with 459

laboratory measurements of Ks for Stony Creek and Sunday Creek (Table 2) but is much smaller 460

than the Ks measured on soil cores from Ella Creek. 461

462

<<<<<<<<<<<<<<<Figure 7 here>>>>>>>>>>>>>>>> 463

464

Macropores can contribute to a high proportion of flow in forest soil when given abundant supply of 465

water. Macropore flow was quantified by subtracting matrix flow (Kmat) from ponded flow (Kp; h = 466

5 mm) (Table 2). The difference in flow was then expressed relative to Kmat (∆K/Kmat), and scaled 467

by macroporosity ( r > 0.5 mm) in order to provide a normalized measure of macropore flow 468

(K+mac) which was independent of site attributes and water repellency status. The normalized 469

22

macropore flow was also calculated for a south-facing headwater at Sunday Creek (n = 24) and a 470

wet eucalyptus forest using infiltration data from Nyman et al. (2010). 471

472

The data from all the sites was plotted as a function of time since fire, tsf (Figure 8a). In general 473

there was a linear increase in the macropore flow during recovery of burned catchments, indicating 474

that burning can have a large impact on macropore availability. The variable cluster of points at 2-3 475

yrs after fire is from Ella Creek, which was the site that had recovered the most. The macropore 476

flow at Stony Creek was lower than expected at two time-steps. This site had the lowest overall 477

macroporosity (Table 2) and was the only site located on a metamorphic geology. 478

479

The storage (H) was normalized by θs of the surface material (depth 0-1 cm) to produce a 480

normalized storage variable, H+, shown as a function of time since fire in Figure 8b. This is the 481

storage capacity of the soil before the flux approaches a steady-state. Storage declined exponentially 482

with time since fire. The initial storage after the burn was not measured and some material had been 483

lost through wind and water erosion prior to the first field campaign. An extrapolated value at t = 0 484

would yield an H+ of 11.1 which is equal to 4.4 mm of actual storage (H) given an average 485

volumetric storage in surface material of 0.40 cm3 cm-3. At Ella Creek (> 30 months after fire) the 486

storage, H, was on average ~1 mm. 487

488

<<<<<<<<<<Figure 8 here>>>>>>>>>> 489

490

3.3 Point- to plot-scale infiltration 491

The infiltration parameters derived from the mini-disc infiltration measurements at Sunday Creek 492

were evaluated against runoff data from rainfall simulation plots. The aim was to determine how 493

23

point scale infiltration measurements compare with infiltration rates measured in large plots (3 m2) 494

under simulated rainfall. The average steady-state infiltration was higher and more variable for the 495

mini-disc infiltrometer (Kmat = 10.1 mm h-1 ±10.1) than under the rainfall simulation plot (steady-496

state infiltration = 7.9 mm h-1 ± 3.65 SD) (Figure 9a). The time to ponding tp for a steady-state 497

rainfall input R was estimated from tension infiltrometer data as; 498

499

�� = �/�� − �� �� 7

The time to ponding was then combined with steady-state infiltration model and kinematic wave 500

equations in order to model the hydrograph from steady-state rainfall (Figure 9b) (Brutsaert, 2005; 501

p. 201). The infiltration excess was routed assuming a relationship between water depth (hw, mm), 502

friction slope (Sf) and average velocity (u, mm h-1); 503

504

' = OℎP!Q

4 8

where Cr is a resistance factor (Brutsaert, 2005; p. 172). The time to equilibrium, the build-up, and 505

the decay-phase were modeled through approximate solutions to the kinematic wave equation under 506

steady lateral rainfall input (R). The hydrograph modeled using mini-disc parameters (H and Kmat) 507

corresponded well with the runoff measurements from plot-scale rainfall simulations (Figure 9b). 508

The ponded infiltration, Kp, (25 mm h-1) reflects the maximum steady-state infiltration capacity of 509

the soil, and was ~2.5 time higher than the steady-state infiltration rate under simulated rainfall. 510

511

<<<<<<<<<<<<<Figure 9 here>>>>>>>>>>>> 512

513

4 Discussion 514

24

4.1 Contribution of the soil matrix to infiltration in burned soils 515

The effective hydraulic conductivity of the soil matrix, Kmat, was strongly dependent on the 516

availability of actively conducting pore-space. Laboratory experiments on intact cores showed that 517

Kmat remained at < 40 % of Ks while the relative soil moisture (θi/θs) was < 0.8. The magnitude of 518

the water repellency effect on hydraulic conductivity (Kmat/ Ks) is similar to the 60-70 % reduction 519

measured in wet eucalyptus forest (Nyman et al., 2010; Sheridan et al., 2007) and the range of 520

impacts reported for forest soils in other places (Imeson et al., 1992; Leighton-Boyce et al., 2007). 521

Hydraulic conductivity (Kmat) increased exponentially with θi when θi/θs > 0.8, which means that the 522

last 20 % of pore-space within the core was contributing disproportionately to flow through the soil 523

matrix. The extremely slow wetting rate for θi/θs > 0.8 and the exponential dependency between 524

Kmat and θi for θi/θs > 0.8 suggests that flow was restricted by a water repellent layer that acted as a 525

‘throttle’. 526

527

Water repellency at the study sites in southeast Australia displayed consistent trends with soil depth. 528

Water repellency peaked at depths of 1 - 3 cm, and remained largely non-repellent on the surface 529

and at depths > 8 cm. Similar patterns of variability with depth have been observed in water 530

repellent soils elsewhere (Benavides-Solorio and MacDonald, 2001; DeBano, 2000; Doerr et al., 531

1996; Doerr et al., 2000; MacDonald and Huffman, 2004; Miyata et al., 2007; Nyman et al., 2011), 532

indicating that the water repellency function, CST(ds) in Eq. (3), may be useful more generally for 533

characterizing water repellency in burned soils. Overall the wildfire-affected sites in southeast 534

Australia displayed moderate to extreme repellency, with air-dry soils on the most recently burned 535

sites at Stony Creek and Sunday Creek (0.5 years after fire) being more repellent than air-dry soils 536

from Ella Creek (2-3 years after fire). Under field conditions, the variability in water repellency 537

(CSTmin) was driven by meteorological conditions (or seasonality) which is consistent with previous 538

25

studies of temporal variations in water repellency (Burch et al., 1989; Crockford et al., 1991; 539

Leighton-Boyce et al., 2005; Sheridan et al., 2007). Field measurements showed that the maximum 540

strength water repellency (CSTmin) in soil profile was driving weather dependent changes in 541

hydraulic conductivity (Kmat) of the soil matrix. 542

543

Moisture induced changes to CSTmin took place over long timescales which meant that the soil 544

required long exposure to wet conditions before approaching a true Ks. A dry and water repellent 545

soil is therefore unlikely to reach saturation at the timescale of convective storm events that result in 546

major runoff events after wildfire. A parameter such as Kmat that represents infiltration in the active 547

region of the soil is more effective than Ks at describing the steady-state infiltration in these systems 548

(Hardie et al, 2011; Liu et al, 2005; Moody and Ebel, 2013). Water uptake in dry and water 549

repellent soil was driven by longer term meteorological conditions, and was therefore represented 550

separately to the infiltration processes taking place at the time-scale of rain storms. Separating 551

between the two processes is consistent with the concept presented in Moody and Ebel (2012), 552

where infiltration into water repellent and dry soils is controlled by diffusion-adsorption processes 553

(timescale: days to weeks) rather than capillary and gravity driven processes (timescale: minutes to 554

hours). Distinguishing between these two infiltration processes based on their respective timescales 555

is an important conceptual development, which contrasts with common infiltration models such as 556

Green-Ampt that assume a uniform wetting front, where infiltration rate equals the saturated 557

hydraulic conductivity once the capillary potential approaches zero. 558

559

The slow wetting rate of the water repellent layer meant that the matrix never reached a saturated 560

hydraulic conductivity (Ks) under field conditions, and that water repellency was restricting flow 561

even under relatively wet conditions. Temporal variability in Kmat due to meteorological related 562

26

water repellency dynamics was modeled by linking CSTmin with the Keech-Byram Drought Index 563

(KBDI), which varies over time as function of daily rainfall and temperature. Both soil moisture and 564

CSTmin declined with increasing KBDI. Water repellency, CSTmin, displayed consistent trends with 565

KBDI and was causing temporal variability in Kmat with values ranging from 16 - 35 mm h-1. There 566

was considerable scatter caused by different background conditions at the three study sites so the 567

model of KBDI-dependent changes in water repellency was improved by normalizing CSTmin values 568

by CSTmin for the most severe water repellency conditions at the site (e.g. Karunarathna et al., 569

2010). Soil moisture (θi/θs) declined linearly with increasing KBDI, but the range of soil moisture 570

conditions was too low to drive temporal variability in Kmat. 571

572

4.2 Surface storage during recovery from wildfire 573

Ash and burned soil on the surface of recently burned hillslopes was generally non-repellent and 574

was capable of buffering subsurface soils from participating in the infiltration process in a similar 575

way to what has been described for burned systems elsewhere (e.g. Ebel et al., 2012; Kinner and 576

Moody, 2010; Woods and Balfour, 2008). The surface material contained ash, but was actually a 577

mixture of ash, gravel, burned soil and charred organics that eroded or was incorporated back into 578

the soil profile during recovery. Initially after fire, the storage, H, was ~4 mm which is in the lower 579

range of values for recently burned hillslopes in Colorado (3.6 - 5.5 mm) (Ebel et al., 2012). The 580

wettable surface material was on average 1 cm deep and had a water holding capacity of 0.40 581

cm3cm-3 which was low compared to the range of values (0.58 – 0.95) reported for ash in the 582

literature (Cerdà and Doerr, 2008; Ebel et al., 2012; Leighton-Boyce et al., 2007; Woods and 583

Balfour, 2008). The lower storage capacity may be explained partly by the high gravel content of 36 584

and 62 % (by mass) at Stony and Sunday Creek, respectively. In the field, the surface storage, H, 585

declined exponentially with time since fire and asymptoted towards background H of ~1.2 mm. 586

27

This decline during recovery was primarily caused by erosion of the wettable (and largely non 587

cohesive) surface layer. The erosion process occurring on surface material at Sunday Creek and 588

Stony Creek is described in Nyman et al., (2013). 589

590

Hydraulic conductivity of surface material, Kash, at Stony Creek and Sunday Creek was high (21 - 591

41 mm h-1) compared to 8.6 mm h-1 in Colorado (Ebel et al., 2012); lower than 51 mm h-1 in 592

Montana (Woods and Balfour, 2008); similar to 36 mm h-1 in Wyoming (Balfour and Woods, 593

2013); lower than 150 mm h-1 in British Columbia (Balfour and Woods, 2013); and much lower 594

than values for other laboratory measurements (138 - 600 mm h-1) (Bodí et al., 2012; Woods and 595

Balfour, 2010). In experiments on ash from Coniferous fuels that were burned in a muffle furnace, 596

Balfour and Woods (2013) found that Kash (12 - 24 mm h-1) produced from moderate burn 597

temperatures (500 - 700 °C) was much lower than the Kash (180 - 720 mm h-1) of ash from low 598

(300°C) and high (900 °C) burn temperatures. The large range of Kash values across the various 599

studies may therefore partly be attributed to the effects of burn temperature on ash composition. 600

With gravel (D < 2 mm) removed, the Kash at Stony Creek became much lower (6.8 mm h-1) while it 601

remained high at Sunday Creek (36.4 mm h-1). 602

603

The large variability in hydraulic properties of ash combined with variable subsurface soil 604

properties makes it difficult to generalize the role of ash (Kash) in controlling steady-state 605

infiltration. At Stony Creek the Kmat (16.0 – 21.5 mm h-1) was equal to the hydraulic conductivity of 606

surface material (Kash = 21 mm h-1) on two occasion during the sampling campaign within the first 607

year of the fire (Table 2), suggesting the Kash may have been restricting steady-state infiltration. In 608

dry conditions however, the effect of water repellency at Stony Creek was stronger than ash effects 609

(Kmat < Kash), thus resulting in subsurface controls on steady-state infiltration, which similar to the 610

28

results in Kinner and Moody (2010). At Sunday Creek, the matrix flow (Kmat: 16.7 - 19.6 mm h-1) 611

was always less than the saturated hydraulic conductivity of ash/surface material (Ks = 41 mm h-1), 612

suggesting the surface layer of ash was not limiting the steady-state infiltration in the soil matrix. 613

614

4.3 Macropore availability during recovery from wildfire 615

Macropores were largely inactive during the plot-scale rainfall experiments and Kmat was therefore a 616

better predictor of steady-state infiltration than the infiltration capacity (Kp), which is the sum of 617

matrix and macropore flow (Kmat + Kmac). However, the activation of macropore flow depends on 618

supply of water at the soil surface (Leonard et al., 2004; Nyman et al., 2010) and therefore depends 619

on spatial scale and rainfall intensity. The contribution of macropore flow, Kmac, to infiltration is 620

likely to increase with ponding, which increases with rainfall intensity and flow accumulation 621

(Karssenberg, 2006; Langhans et al., 2011, 2013). Thus, the effective infiltration rates on hillslopes, 622

when the supply of water at the surface is greater than the infiltration capacity of the matrix (R > 623

Kmat), is likely to be somewhere between Kmat and Kp (Smith and Goodrich, 2000). 624

625

The linear increase in macropore availability during recovery shows that wildfire cause an initial 626

reduction in infiltration capacity of macropores. The exact mechanism underlying this effect was 627

not established but previous studies have shown that ash can limit flow by acting as a restricting 628

layer and/or by clogging of pores within the soil profile (Ebel et al., 2012; Mallik et al., 1984; 629

Nyman et al., 2010; Onda et al., 2008; Woods and Balfour, 2010). This effect may not be that 630

pronounced for flow in the soil matrix where capillary processes dominate. However, an ash layer 631

may be much more important in reducing infiltration capacity (Kp) when the surface is inundated, 632

because a layer of unstructured material (i.e. ash) on the soil surface would limit the rate at which 633

ponding water can be delivered to macropores in the underlying soil. Hence the structural 634

29

components of the soil are unlikely to infiltrate at full capacity. At Ella Creek, 2-3 years post-635

wildfire, macropore flow made up 88 - 93 % of the steady-state infiltration capacity, Kp. The 636

increase in macropore availability with time since fire may be a key feature resulting in increased 637

infiltration at during recovery. This effect however may go undetected in small runoff plots or 638

conventional rainfall simulation experiments because of the strong sensitivity of Kp to the supply of 639

water at the soil surface. 640

641

5 Conclusion 642

The overall objective of the study was to identify key properties contributing to variability in 643

infiltration during recovery from wildfire and to quantify their effect. A storage-based infiltration 644

model was used as framework for analyzing infiltration data. The model fitted the data well and 645

could be used effectively to partition the infiltration process into its key components of storage, 646

matrix flow and macropore flow. Infiltration on intact cores showed that a water repellent layer 647

acted as a ‘throttle’ on flow and that the soil moisture status of the soil therefore could have large 648

effects on flow through the soil matrix. The slow wetting process in the water repellent layer meant 649

that only after prolonged exposure to saturated conditions did the soil become fully saturated and 650

able to conduct water at its full potential. In headwater catchments that were recovering from 651

wildfire (tsf < 3 years) the soils always displayed some degree of water repellency resulting in 652

incomplete saturation, which meant that the saturated hydraulic conductivity of the soil (Ks) could 653

not be measured in the field. The variability in the infiltration capacity (Kp) was driven primarily by 654

changes in macropore availability but also by temporal variability in water repellency due to 655

meteorological conditions 656

657

30

The results from infiltration measurements in headwater catchments were used to model the spatial 658

and temporal variability in infiltration processes as a function of time since fire (tsf) and catchment 659

dryness (KBDI), which was calculated from monthly weather patterns. The relations underlying the 660

model can be summarized as follows. Water repellency was represented by the minimum critical 661

surface tension (CSTmin) in the soil profile. The CSTmin increased exponentially with decreasing 662

KBDI (i.e. the strength of water repellency decreased exponentially when the monthly weather 663

resulted in wet conditions). The matrix flow (Kmat) in turn, increased exponentially with increasing 664

CSTmin. The storage (H) declined exponentially with tsf, due to the loss of a discrete layer of 665

wettable surface soil. At the same time, the macropore flow increased linearly with tsf due to 666

increased macropore availability during recovery from the fire impact. The relations are given as 667

empirical equations in Figures 7 and 8. Future work will aim to model infiltration during actual 668

storm events and test these relations against observed runoff responses from burned hillslopes 669

670

Overall the results of the study highlight that models of infiltration into burned soils should take 671

into account i) the surface storage in ash and burnt soil, ii) the effect of ponding on macropore flow, 672

and iii) non-uniform wetting in a soil layer that displays varying levels of water repellency 673

depending on meteorological conditions. 674

675

676

31

6 Acknowledgments 677

Field work was carried out with assistance from Philip Noske and Chris Sherwin at the University 678

of Melbourne. Research funding was provided by Melbourne Water and the eWater Cooperative 679

Research Centre. The authors are grateful for comments and suggestions from two anonymous 680

reviewers. 681

682

32

7 References 683

Balfour, V.N., Woods, S.W., 2013. The hydrological properties and the effects of hydration on 684

vegetative ash from the Northern Rockies, USA. CATENA, 111(0): 9-24. 685

686

Benavides-Solorio, J., MacDonald, L.H., 2001. Post-fire runoff and erosion from simulated rainfall 687

on small plots, Colorado Front Range. Hydrological Processes, 15(15): 2931-2952. 688

689

Beven, K., Germann, P., 2013. Macropores and water flow in soils revisited. Water Resources 690

Research, 49(6): 3071-3092. 691

692

Bodí, M.B., Doerr, S.H., Cerdà, A., Mataix-Solera, J., 2012. Hydrological effects of a layer of 693

vegetation ash on underlying wettable and water repellent soil. Geoderma, 191: 14-23. 694

695

Brutsaert, W., 2005. Hydrology: an introduction. Cambridge University Press, Cambridge 696

697

Burch, G.J., Moore, I.D., Burns, J., 1989. Soil hydrophobic effects on infiltration and catchment 698

runoff. Hydrological Processes, 3(3): 211-222. 699

700

Cannon, S.H., 2001. Debris-flow generation from recently burned watersheds. Environmental & 701

Engineering Geoscience, 7(4): 321-341. 702

703

Cawson, J.G., Sheridan, G.J., Smith, H.G., Lane, P.N.J., 2012. Surface runoff and erosion after 704

prescribed burning and the effect of different fire regimes in forests and shrublands: a review. 705

International Journal of Wildland Fire, 21(7): 857-872. 706

33

707

Cerdà, A., Doerr, S.H., 2007. Soil wettability, runoff and erodibility of major dry-Mediterranean 708

land use types on calcareous soils. Hydrological Processes, 21(17): 2325-2336. 709

710

Cerdà, A., Doerr, S.H., 2008. The effect of ash and needle cover on surface runoff and erosion in 711

the immediate post-fire period. CATENA, 74(3): 256-263. 712

713

Cerda, A., Lasanta, T., 2005. Long-term erosional responses after fire in the Central Spanish 714

Pyrenees - 1. Water and sediment yield. CATENA, 60(1): 59-80. 715

716

Cook, F., 2007. Unsaturated hydraulic conductivity: Laboratory tension infiltrometer. In: Carter, 717

M., Gregorich, E. (Eds.), Soil Sampling and Methods of Analysis. CRC. Retrieved July 27, 2012, 718

from Ebook Library. 719

720

Crockford, H., Topalidis, S., Richardson, D.P., 1991. Water repellency in a dry sclerophyll eucalypt 721

forest - measurements and processes. Hydrological Processes, 5(4): 405-420. 722

723

Cruz, M.G. et al., 2012. Anatomy of a catastrophic wildfire: The Black Saturday Kilmore East fire 724

in Victoria, Australia. Forest Ecology and Management, 284(0): 269-285. 725

726

DeBano, L.F., 2000. The role of fire and soil heating on water repellency in wildland environments: 727

a review. Journal of Hydrology, 231-232: 195-206. 728

729

34

Doerr, S.H., Moody, J.A., 2004. Hydrological effects of soil water repellency: on spatial and 730

temporal uncertainties. Hydrological Processes, 18(4): 829-832. 731

732

Doerr, S.H., Shakesby, R.A., Walsh, R.P.D., 1996. Soil hydrophobicity variations with depth and 733

particle size fraction in burned and unburned Eucalyptus globulus and Pinus pinaster forest terrain 734

in the Águeda Basin, Portugal. CATENA, 27(1): 25-47. 735

736

Doerr, S.H., Shakesby, R.A., Walsh, R.P.D., 1998. Spatial variability of soil hydrophobicity in fire-737

prone eucalyptus and pine forests, Portugal. Soil Sci., 163(4): 313-324. 738

739

Doerr, S.H., Shakesby, R.A., Walsh, R.P.D., 2000. Soil water repellency: its causes, characteristics 740

and hydro-geomorphological significance. Earth-Science Reviews, 51(1–4): 33-65. 741

742

Ebel, B.A., 2012. Wildfire impacts on soil-water retention in the Colorado Front Range, United 743

States. Water Resources Research, 48(12): W12515. 744

745

Ebel, B.A., Moody, J.A., 2013. Rethinking infiltration in wildfire-affected soils. Hydrological 746

Processes, 27(10): 1510-1514. 747

748

Ebel, B.A., Moody, J.A., Martin, D.A., 2012. Hydrologic conditions controlling runoff generation 749

immediately after wildfire. Water Resourses Research., 48(3): W03529. 750

751

35

Granged, A.J.P., Jordán, A., Zavala, L.M., Bárcenas, G., 2011. Fire-induced changes in soil water 752

repellency increased fingered flow and runoff rates following the 2004 Huelva wildfire. 753

Hydrological Processes, 25(10): 1614-1629. 754

755

Green, W.H., Ampt, G.A., 1911. Studies on soil physics, 1. The flow of air and water through soils. 756

Journal of Agricultural Science, 4(1): 1. 757

758

Hardie, M. A., Cotching, W. E., Doyle, R. B., Holz, G., Lisson, S., Mattern, K., 2011. Effect of 759

antecedent soil moisture on preferential flow in a texture-contrast soil. Journal of Hydrology, 398(3-760

4), 191-201. 761

762

Imeson, A.C., Verstraten, J.M., Vanmulligen, E.J., Sevink, J., 1992. The effects of fires and water 763

repellency on infiltration and runoff under Mediterranean type forest. CATENA, 19(3-4): 345-361. 764

765

Johansen, M.P., Hakonson, T.E., Breshears, D.D., 2001. Post-fire runoff and erosion from rainfall 766

simulation: contrasting forests with shrublands and grasslands. Hydrological Processes, 15(15): 767

2953-2965. 768

769

Karssenberg, D. , 2006. Upscaling of saturated conductivity for Hortonian runoff modelling, 770

Advances in Water Resources. 29(5), 735-759. 771

772

Karunarathna, A.K., Kawamoto, K., Moldrup, P., de Jonge, L.W., Komatsu, T., 2010. A Simple 773

Beta-Function Model for Soil-Water Repellency as a Function of Water and Organic Carbon 774

Contents. Soil Science, 175(10): 461-468. 775

36

776

Kean, J.W., Staley, D.M., Cannon, S.H., 2011. In situ measurements of post-fire debris flows in 777

southern California: Comparisons of the timing and magnitude of 24 debris-flow events with 778

rainfall and soil moisture conditions. Journal of Geophysical Research-Earth Surface, 116(F04019). 779

780

Keetch, J.J., Byram, G.M., 1968. A drought index for forest fire control. US Department of 781

Agriculture, Forest Service, Southeastern Forest Experiment Station. 782

783

King, P.M., 1981. Comparison of methods for measuring severity of water repellence of sandy soils 784

and assessment of some factors that affect its measurement. Australian Journal of Soil Research, 785

19(4): 275-285. 786

787

Kinner, D.A., Moody, J.A., 2010. Spatial variability of steady-state infiltration into a two-layer soil 788

system on burned hillslopes. Journal of Hydrology, 381(3-4): 322-332. 789

790

Kirkby, M.J., 1975. Hydrograph modelling strategies, process in physical and human geography B2 791

- Process in Physical and Human Geography, pp. 69-90. 792

793

Lane, P.N.J., Sheridan, G.J., Noske, P.J., 2006a. Changes in sediment loads and discharge from 794

small mountain catchments following wildfire in south Eastern Australia. Journal of 795

Hydrology(331): 495-510. 796

797

37

Lane, P.N.J., Hairsine, P.B., Croke, J.C., Takken, I., 2006b. Quantifying diffuse pathways for 798

overland flow between the roads and streams of the Mountain Ash forests of central Victoria 799

Australia. Hydrological Processes, 20(9): 1875-1884. 800

801

Langhans, C., Govers, G., Diels J,. 2013. Development and parameterization of an infiltration 802

model accounting for water depth and rainfall intensity. Hydrological Processes, 27(25), 3777-803

3790. 804

805

Langhans, C., Govers, G., Diels, J., Leys, A., Clymans, W., Putte, A. V. d., Valckx, J., 2011. 806

Experimental rainfall-runoff data: Reconsidering the concept of infiltration capacity. Journal of 807

Hydrology, 399(3-4), 255-262. 808

809

Larsen, I.J. et al., 2009. Causes of post-fire runoff and erosion: Water repellency, cover, or soil 810

sealing? Soil Science Society of America Journal, 73(4): 1393-1407. 811

812

Leighton-Boyce, G., Doerr, S.H., Shakesby, R.A., Walsh, R.P.D., 2007. Quantifying the impact of 813

soil water repellency on overland flow generation and erosion: a new approach using rainfall 814

simulation and wetting agent on in situ soil. Hydrological Processes, 21(17): 2337-2345. 815

816

Leighton-Boyce, G. et al., 2005. Temporal dynamics of water repellency and soil moisture in 817

eucalypt plantations, Portugal. Australian Journal of Soil Research, 43(3): 269-280. 818

819

38

Leonard, J., Perrier, E., Rajot, J.L., 2004. Biological macropores effect on runoff and infiltration: a 820

combined experimental and modelling approach. Agriculture Ecosystems & Environment, 104(2): 821

277-285. 822

823

Liu, H.H., Zhang, R., Bodvarsson, G.S., 2005. An active region model for capturing fractal flow 824

patterns in unsaturated soils: Model development. Journal of Contaminant Hydrology. 80(1-2): 18-825

30. 826

827

MacDonald, L.H., Huffman, E.L., 2004. Post-fire soil water repellency: Persistence and soil 828

moisture thresholds. Soil Science Society of America Journal, 68(5): 1729-1734. 829

830

Mallik, A.U., Gimingham, C.H., Rahman, A.A., 1984. Ecological effects of heather burning 1. 831

Water infiltration, moisture retention and porosity of surface soil Journal of Ecology, 72(3): 767-832

776. 833

834

Martin, D.A., Moody, J.A., 2001. Comparison of soil infiltration rates in burned and unburned 835

mountainous watersheds. Hydrological Processes, 15(15): 2893-2903. 836

837

Miyata, S., Kosugi, K.i., Gomi, T., Onda, Y., Takahisa, M., 2007. Surface runoff as affected by soil 838

water repellency in a Japanese cypress forest. Hydrological Processes, 21(17): 2365-2376. 839

840

Moody, J.A., Ebel, B.A., 2012. Hyper-dry conditions provide new insights into the cause of 841

extreme floods after wildfire. CATENA, 93(0): 58-63. 842

843

39

Moody, J.A., Kinner, D.A., Úbeda, X., 2009. Linking hydraulic properties of fire-affected soils to 844

infiltration and water repellency. Journal of Hydrology, 379(3-4): 291-303. 845

846

Moody, J.A., Martin, D.A., 2001. Post-fire, rainfall intensity–peak discharge relations for three 847

mountainous watersheds in the western USA. Hydrological Processes, 15(15): 2981-2993. 848

849

Moody, J.A., Shakesby, R.A., Robichaud, P.R., Cannon, S.H., Martin, D.A., 2013. Current research 850

issues related to post-wildfire runoff and erosion processes. Earth-Science Reviews, 122(0): 10-37. 851

852

Nyman, P., Sheridan, G., Lane, P.N.J., 2010. Synergistic effects of water repellency and macropore 853

flow on the hydraulic conductivity of a burned forest soil, south-east Australia. Hydrological 854

Processes, 24(20): 2871-2887. 855

856

Nyman, P., Sheridan, G.J., Smith, H.G., Lane, P.N.J., 2011. Evidence of debris flow occurrence 857

after wildfire in upland catchments of south-east Australia. Geomorphology, 125(3): 383-401. 858

859

Nyman, P., Sheridan, G. J., Moody, J. A., Smith, H. G., Noske, P. J., Lane, P. N. J. , 2013. 860

Sediment availability on burned hillslopes. Journal of Geophysical Research: Earth Surface, 861

2012JF002664. 862

863

Onda, Y., Dietrich, W.E., Booker, F., 2008. Evolution of overland flow after a severe forest fire, 864

Point Reyes, California. CATENA, 72(1): 13-20. 865

866

40

Perroux, K.M., White, I., 1988. Design for disk permeameters. Soil Science Society of America 867

Journal, 52(5): 1205-1215. 868

869

Philip, J.R., 1957. The theory of infiltration: 1. The infiltration equation and its solution. Soil 870

Science, 83(5): 345. 871

872

Prosser, I.P., Williams, L., 1998. The effect of wildfire on runoff and erosion in native Eucalyptus 873

forest. Hydrological Processes, 12(2): 251-265. 874

875

Rawls, W., Brakensiek, D., Miller, N., 1983. Green‐ampt Infiltration Parameters from Soils Data. 876

Journal of Hydraulic Engineering, 109(1): 62-70. 877

878

Risse, L.M., Nearing, M.A., Savabi, M.R., 1994. Determining the Green-Ampt effective hydraulic 879

conductivity from rainfall-runoff data for the WEPP model. Anglais, 37(2). 880

881

Robichaud, P., Wagenbrenner, J., Brown, R., Wohlgemuth, P., Beyers, J., 2008a. Evaluating the 882

effectiveness of contour-felled log erosion barriers as a post-fire runoff and erosion mitigation 883

treatment in the western United States. International Journal of Wildland Fire, 17(2): 255-273. 884

885

Robichaud, P., Lewis, S., Ashmun, L., 2008b. New procedure for sampling infiltration to assess 886

post-fire soil water repellency, United States Department of Agriculture Forest Service, Fort 887

Collins. 888

889

41

Robichaud, P.R., 2000. Fire effects on infiltration rates after prescribed fire in Northern Rocky 890

Mountain forests, USA. Journal of Hydrology, 231-232: 220-229. 891

892

Robichaud, P.R., Elliot, W.J., Pierson, F.B., Hall, D.E., Moffet, C.A., 2007. Predicting postfire 893

erosion and mitigation effectiveness with a web-based probabilistic erosion model. CATENA, 894

71(2): 229-241. 895

896

Scoging, H.M., 1979. Infiltration characteristics in a semiarid environment. The hydrology of areas 897

of low precipitation. Proc. Canberra Symp., Canberra, Australia. December 1979: 159. 898

899

Shakesby, R.A. et al., 2003. Fire severity, water repellency characteristics and 900

hydrogeomorphological changes following the Christmas 2001 Sydney forest fires. Australian 901

Geographer, 34(2): 147-175. 902

903

Shakesby, R.A., Doerr, S.H., 2006. Wildfire as a hydrological and geomorphological agent. Earth-904

Science Reviews, 74(3-4): 269-307. 905

906

Sheridan, G.J., Lane, P.N.J., Noske, P.J., 2007. Quantification of hillslope runoff and erosion 907

processes before and after wildfire in a wet Eucalyptus forest. Journal of Hydrology, 343(1-2): 12-908

28. 909

910

Shin, S.S., Park, S.D., Lee, K.S., 2013. Sediment and hydrological response to vegetation recovery 911

following wildfire on hillslopes and the hollow of a small watershed. Journal of Hydrology, 499(0): 912

154-166. 913

42

914

Smith, H.G., Sheridan, G.J., Lane, P.N.J., Bren, L.J., 2011. Wildfire and salvage harvesting effects 915

on runoff generation and sediment exports from radiata pine and eucalypt forest catchments, south-916

eastern Australia. Forest Ecology and Management, 261(3): 570-581. 917

918

Smith, R., Smetten, R.J., Broadbridge, P., Woolhiser, D.A., 2002. Infiltration theory for hydrologic 919

applications. Vol. 15. American Geophysical Union 920

921

Smith, R.E., Goodrich, D.C., 2000. Model for rainfall excess patterns on randomly heterogeneous 922

areas. Journal of Hydrologic Engineering, 5(4): 355-362. 923

924

Stoof, C.R., Wesseling, J.G., Ritsema, C.J., 2010. Effects of fire and ash on soil water retention. 925

Geoderma,159(3): 276-285. 926

927

Urbanek, E., Shakesby, R.A., 2009. Impact of stone content on water movement in water-repellent 928

sand. European Journal of Soil Science, 60(3): 412-419. 929

930

Wagenbrenner, J.W., Robichaud, P.R., 2013. Post-fire bedload sediment delivery across spatial 931

scales in the interior western United States. Earth Surface Processes and Landforms. DOI: 932

10.1002/esp.3488 933

934

Woods, S.W., Balfour, V.N., 2008. The effect of ash on runoff and erosion after a severe forest 935

wildfire, Montana, USA. International Journal of Wildland Fire, 17(5): 535-548. 936

937

43

Woods, S.W., Balfour, V.N., 2010. The effects of soil texture and ash thickness on the post-fire 938

hydrological response from ash-covered soils. Journal of Hydrology, 393(3-4): 274-286. 939

940

Woods, S.W., Birkas, A., Ahl, R., 2007. Spatial variability of soil hydrophobicity after wildfires in 941

Montana and Colorado. Geomorphology, 86(3-4): 465-479. 942

943

44

Tables 944

Table 1 Attributes of the three study sites. 945

Site

Aspect &

Elevation

Burn impact

Annual

Rainfall

Forest Type

and dominant vegetation

Geology Soil texture

Ella

Creek,

North

Moderate to

high severity

Dec-2006

1200-1400 Dry eucalyptus,

Broad-leaved peppermint

(E. Radiata)

Narrow-leaved peppermint

(E. Dives)

Shale,

Marine

Sedimentary

Stony and

gravelly clay

loam

Sunday

Creek

North

High to very

high severity

Feb-2009

1000-1200 Dry eucalyptus,

Broad-leaved peppermint

(E. Radiata)

Siltstone,

Marine

Sedimentary

Stony and

gravelly clay

loam

Stony

Creek

Northwest

High to very

high severity

Feb-2009

1000-1200 Dry eucalyptus,

Broad-leaved peppermint

(E. Radiata)

Phyllite & Gneiss

Metamorphic

Gravelly clay

loam

946

947

45

Table 2 Soil hydrological properties measured on intact cores (~8 cm deep and 5.3 cm in diameter) 948

from Sunday Creek, Stony Creek and Ella Creek in Victoria, southeast Australia 949

Site Bulk

Density

Gravel

D > 2 mm

Porosity

θs

Macroporosity

(r > 0.5 mm)

Air-dry

CSTmin

Ks

a

(h = -15mm)

Kash

b

(h = -15mm)

n = 9 n = 9 n = 9 n = 9 n = 24 n = 9 n = 5

g cm-3 % mass cm3 cm-3 % of θs dyn cm-1 mm h-1 mm h-1

Sunday

Creek 1.29 62 0.34 4.7 40.1 66 (18) 41 (4.0)

Stony

Creek 1.31 36 0.41 1.7 46.5 52 (41) 21 (4.9)

Ella

Creek 0.97 27 0.55 2.6 52.4 240 (81) n/a

a This is the hydraulic conductivity of the cores when completely wetted at h = -15 mm. 950

b Repacked cores contained a mixture of gravel, burned soil and ash. 951

46

Table 3 Summary of soil hydrological properties and initial conditions in four measurement campaigns to headwater catchments in

southeast Australia. Sunday Creek and Stony Creek were burned by wildfire in February 2009 and Ella Creek was burned in December

2006. The standard deviation of the effective hydraulic conductivity of the soil matrix (Kmat) and steady-state infiltration capacity (Kp) are

given in parentheses.

Sunday Creek Stony Creek Ella Creek

Date tsf

KBDI θi CSTmin H Kmat Kp tsf KBDI θi CSTmin H Kmat Kp tsf KBDI θi CSTmin H Kmat Kp

mo - cm3 cm-3 dyn cm-1 mm mm h-1 mm h-1 mo - cm3 cm-3 dyn cm-1 mm mm h-1 mm h-1 mo - cm3 cm-3 dyn cm-1 mm mm h-1 mm h-1

- - n=27 n=27 n=36 n=36 n=36 - - n=27 n=27 n=36 n=36 n=36 - - n=27 n=27 n=36 n=36 n=36

Jul-09 6 89 0.09 27.6 1.9 16.7 (22.1)

56 (93)

6 26 0.14 57.7 3.6 21.7 (32.4)

32 (61)

30 1 0.21 58.2 1.2 34.9 (30.6)

371 (419)

Dec-09 10 66 0.06 24.8 1.6 16.9 (24.6)

63 (134) 10 62 0.12 42.3 2.9

16.0 (17.4)

40 (96) 34 54 0.19 29.2 1.2

19.3 (19.4)

162 (174)

Mar-10 13 32 0.09 26.3 1.8 18.3 (27.8)

83 (188)

13 151 0.04 46.5 3.7 21.5 (26.9)

27 (39)

37 86 0.14 46.6 1.3 20.7 (26.4)

306 (444)

Sept-10 19 0 0.15 43.1 1.2 19.6

(16.3) 142

(163) 19 16 0.12 63.8 1.0 31.9 (27.7)

70 (54)

47

Figure captions

Figure 1. a) Storage (H), matrix flow (Kmat), and macropore flow (Kmac) as three parameters in

the infiltration process. b) A schematic representation of infiltration parameters in a water

repellent soil.

Figure 2. The infiltration measurements were made on soils from three sites in the eastern

uplands of Victoria, southeast Australia.

Figure 3. The date, time since fire and the season when soil hydraulic properties were

sampled in small headwater catchments at Stony Creek, Sunday Creek and Ella. The solid

black horizontal lines show the timing of sampling campaigns at each site.

Figure 4. The strength of water repellency (critical surface tension) at different depths at a)

Sunday Creek, b) Stony Creek and c) Ella Creek in Victoria southeast Australia. Ella Creek

was burned by wildfire in December 2006 and Stony Creek and Sunday Creek were burned

in February 2009. Measurements were obtained on air-dry soil samples (n = 36) at multiple

depths (0-1, 1-3, 3-5, 5-7.5 and 7.5-10 cm) which were collected in March 2010 (Figure 3).

Figure 5. a) Initial soil moisture (θi), b) matrix flow (Kmat) and c) change in soil moisture (∆θ)

during infiltration into intact soil cores from Sunday Creek, Stony Creek and Ella Creek (n =

9) at different stages of wetting on a tension table. All values are for h = -15 mm. The water

uptake in c) was obtained from the mass of soil cores measured before and after each 35

minute infiltration.

48

Figure 6. a) The normalized hydraulic conductivity (Kmat/Ks) as a function of initial soil

moisture (θi/ θs). The normalized hydraulic conductivity was fitted with an exponential

function. Once this function equals 1, the matrix flow Kmat is equal to the saturated

conductivity of the soil, Ks, and there can be no further increase in K. b) Normalized

absorption rate (∆θ/θs)/thrs as a function of initial soil moisture (θi/ θs). A decrease in

absorption with increasing initial soil moisture was represented using a linear relation. The

data in a) and b) are from infiltration measurements on intact cores from Ella Creek, Stony

Creek and Sunday Creek (n = 9) at 4 different stages of wetting (Figure 5). Error bars show

the standard error (SE) of the mean.

Figure 7. Soil hydrological properties of headwater catchments in dry eucalyptus forests

during recovery from wildfire. a) Initial soil moisture, θi, versus KBDI. b) Matrix flow, Kmat,

versus soil moisture c) Normalized water repellency, CST+

min, versus KBDI. d) The change in

Kmat with changing water repellency, CSTmin. Each soil moisture and water repellency (CST)

data point represents the average of 27 measurements made on composite samples collected

from two points (ds = 0 - 10 cm) at 9 locations along 3 transects (80 m) at Sunday Creek,

Stony Creek and Ella Creek (n = 27). Each matrix flow Kmat data point is the average of 4

mini-disc measurement point in quadrates (1 x 1 m) at 3 locations along 3 sampling transects

(n = 36). Error bars show the standard error (SE) of the mean.

Figure 8. a) Macropore flow and b) storage as a function of time since fire, tsf. The changes in

macropore flow during recovery was analyzed using infiltration data from Sunday Creek

South (n = 24) and East Kiewa (Nyman et al., 2010) in addition to data from the three main

49

study site at Sunday Creek, Stony Creek and Ella Creek. Error bars show the standard error

(SE) of the mean.

Figure 9. a) The infiltration rate for tension infiltrometer measurements (n = 12) and rainfall

simulation plots (2 x 1.5 m; n = 3) at Sunday Creek in March 2010. A storage-based

infiltration function, Eq. (1), was fitted to all 12 replicate tension infiltrometer measurements

and shown as a single function representing the average infiltration for the three plots. b) The

runoff from rainfall simulation plots (data points) and the modeled hydrograph using the

parameters from the fitted function in a).

Time (hr)

Cu

mu

lati

ve

infi

ltra

tio

n (

mm

)

f(t) = H/t

f(t) = Kmat + Kmac

Tension

infiltrometer

Ponded

infiltrometer

f(t) = Kmat

Matrix flow Kmat

Water repellent

Surface storage

Subsurface

drainage

Macropore flowKmac

Water supplya) b)

Jul-09

Dec-09

Mar-10

Sept-10

6 12 18 24 36 42

Winter

Summer

Winter

Sunday & stony Ck

Burnt in Feb-09

EllaCk

Burnt in Dec-06

SeasonMonth

Time since fire (months)

Site

c)b)a)

CSTmin

CSTmin CSTmin

a)

Sunday Creek

Stony Creek

Ella Creek

*

b) c)

s

i

hrss t θ

θ

θ

θ337.0356.0 −=

∆

∆

s

i

eK

K

s

mat θ

θ87.5

003.0202.0 +=

Hyd

rau

lic c

on

du

ctiv

ity, K

mat/K

s

Normalized soil moisture, θi/ θs

Normalized soil moisture, θi/ θs

Wa

ter

up

take

,( ∆θ/θ

s)/

t hrs

b)

a)

a) b)

c)

73.0

083.07.162

*083.0 min

=

+=

R

eKCST

mat

d)

61.0

0017.038.0/2

=

−=

R

KBDIsi θθ

62.0

80.012

*026.0min

=

+=−+

R

eCSTKBDI

Stony Ck

Ella Ck

Sunday Ck, south aspect

East Kiewa, Nyman (2010)

Sunday Ck, Stony Ck and Ella Ck

X

60.0

70.223.12

)082.0(

=

+=−+

R

eH sft

80.0

196.02

=

=+

R

tK sfmac

a)

b)

a)

Mini-disc (n=12)

Rainfall simulation

Mini-disc fitted line

Modeled

RFS Runoff

b)

50

Highlights

• Infiltration after wildfire is initially controlled by storage in wettable surface material

• Steady state infiltration is controlled by the hydraulic conductivity of ash or the soil • Soil hydraulic conductivity is restricted by repellency and macropore availability

• The maximum strength of repellency in the soil is a function of monthly weather • Macropore flow is initially low after wildfire but increases during recovery