Available via license: CC BY 4.0
Content may be subject to copyright.
1
Augmentation and Use of WRF-Hydro to Simulate Overland Flow- 1
and Streamflow-Generated Debris Flow Hazards in Burn Scars 2
3
Chuxuan Li1, Alexander L. Handwerger2,3, Jiali Wang4, Wei Yu5,6, Xiang Li7, Noah J. 4
Finnegan8, Yingying Xie9,10, Giuseppe Buscarnera7, and Daniel E. Horton1 5
1 Department of Earth and Planetary Sciences, Northwestern University, Evanston, IL, 60208, USA 6
2 Joint Institute for Regional Earth System Science and Engineering, University of California, Los Angeles, 7
CA, 90095, USA 8
3 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, 91109, USA 9
4 Environmental Science Division, Argonne National Laboratory, Lemont, IL, 60439, USA 10
5 Cooperative Institute for Research in Environmental Sciences, University of Colorado Boulder, CO, 11
80309, USA 12
6 NOAA/Global Systems Laboratory, 325 Broadway Boulder, Denver, CO, 80305-3328, USA 13
7 Department of Civil and Environmental Engineering, Northwestern University, Evanston, IL, 60208, USA 14
8 University of California Santa Cruz, Department of Earth and Planetary Sciences, Santa Cruz, CA, 95064, 15
USA 16
9 Program in Environmental Sciences, Northwestern University, 2145 Sheridan Road, Evanston, IL, 60208, 17
USA
18
10 Department of Biological Sciences, Purdue University, 915 W State St, West Lafayette, IN 47907, USA 19
20
Correspondence to: Chuxuan Li (chuxuanli2020@u.northwestern.edu) 21
22
23
24
25
26
27
28
29
30
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
2
Abstract 31
In steep wildfire-burned terrains, intense rainfall can produce large volumes of runoff that can 32
trigger highly destructive debris flows. The ability to accurately characterize and forecast debris-33
flow hazards in burned terrains, however, remains limited. Here, we augment the Weather 34
Research and Forecasting Hydrological modeling system (WRF-Hydro) to simulate both overland 35
and channelized flows and assess postfire debris-flow hazards over a regional domain. We perform 36
hindcast simulations using high-resolution weather radar-derived precipitation and reanalysis data 37
to drive non-burned baseline and burn scar sensitivity experiments. Our simulations focus on 38
January 2021 when an atmospheric river triggered numerous debris flows within a wildfire burn 39
scar in Big Sur – one of which destroyed California’s famous Highway 1. Compared to the 40
baseline, our burn scar simulation yields dramatic increases in total and peak discharge, and shorter 41
lags between rainfall onset and peak discharge. At Rat Creek, where Highway 1 was destroyed, 42
discharge volume increases eight-fold and peak discharge triples relative to the baseline. For all 43
catchments within the burn scar, we find that the median catchment-area normalized discharge 44
volume increases nine-fold after incorporating burn scar characteristics, while the 95th percentile 45
volume increases 13-fold. Catchments with anomalously high hazard levels correspond well with 46
post-event debris flow observations. Our results demonstrate that WRF-Hydro provides a 47
compelling new physics-based tool to investigate and potentially forecast postfire hydrologic 48
hazards at regional scales. 49
50
Short Summary 51
In January 2021 a storm triggered numerous debris flows in a wildfire burn scar in California. We 52
use a hydrologic model to assess debris flow hazards in pre-fire and postfire scenarios. Compared 53
to pre-fire conditions, the postfire simulation yields dramatic increases in total and peak discharge, 54
substantially increasing debris flow hazards. Our work demonstrates the utility of 3-D hydrologic 55
models for investigating and potentially forecasting postfire debris flow hazards at regional 56
scales. 57
58
1 Introduction 59
Following intense rainfall, areas with wildfire burn scars are more prone to flash flooding (Neary 60
et al., 2003; Bart & Hope 2010; Bart 2016) and runoff-generated debris flow hazards than 61
unburned areas (Moody et al., 2013; Ice et al., 2004; Shakesby & Doerr, 2006). After wildfire, 62
reduced tree canopy interception, decreased soil infiltration due to soil-sealing effects (Larsen et 63
al., 2009), and increased soil water repellency – especially in hyper-arid environments (Dekker 64
and Ritsema, 1994; Doerr and Thomas, 2000; MacDonald and Huffman, 2004) – increases excess 65
surface water, and on sloped terrains leads to overland flow (Shakesby & Doerr, 2006; Stoof et al., 66
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
3
2012). As water moves down hillslopes and erosion adds sediment to water-dominated flows, clear 67
water floods can transition to turbulent and potentially destructive debris flows (Meyer & Wells, 68
1997; Cannon et al., 2001, 2003; Santi et al., 2008). In contrast to debris flows initiated by shallow 69
landslides, this rainfall-runoff process has been identified as the major cause for postfire debris 70
flows in the western U.S. (Cannon, 2001; Cannon et al., 2003, 2008; Kean et al., 2011; Nyman et 71
al., 2015; Parise & Cannon, 2012), and in other regions with Mediterranean climates (Mitsopoulos 72
& Mironidis, 2006; Bisson et al., 2005; Parise & Cannon, 2008, 2009; Rosso et al., 2007). In 73
California, because climate change is projected to increase the intensity and frequency of wet-74
season precipitation (Swain et al., 2018; Polade et al., 2017), increase wildfire potential (Swain, 75
2021; Brown et al., 2021), and extend the wildfire season (Goss et al., 2020), occurrence and 76
intensity of postfire debris flows are likely to increase (Cannon et al., 2009; Kean & Staley, 2021; 77
Oakley, 2021). 78
To assess postfire debris flow hazards, statistical approaches including empirical models and 79
machine-learning techniques are commonly used in both research and operational settings 80
(Gardner et al., 2014; Cannon et al., 2010; Staley et al., 2016; Cui et al., 2019; Nikolopoulos et al., 81
2018; Friedel 2011a, 2011b). Statistical approaches are useful for identifying and characterizing 82
relationships amongst contributing environmental factors and are helpful in operational settings 83
due to low computational costs and the potential for rapid assessment. For example, the U.S. 84
Geological Survey (USGS) currently employs a statistical approach in their Emergency 85
Assessment of Postfire Debris-flow Hazards that consists of a logistic regression model to predict 86
the likelihood of post-wildfire debris flows (e.g., Staley et al., 2016; Cannon et al., 2010), and a 87
multiple linear regression model to predict debris flow volumes (Gartner et al., 2014). Machine-88
learning techniques have also been used to predict postfire debris flows in the western U.S. 89
(Nikolopoulos et al., 2018; Friedel 2011a, 2011b). For example, self-organizing maps and genetic 90
programming were used to predict postfire debris flow occurrence (Friedel 2011b) and volumes 91
(Friedel 2011a), respectively. Compared to the current USGS predictive models, genetic 92
programming was posited to be more useful in solving non-linear multivariate problems (Friedel 93
2011a), while a random forest algorithm demonstrated increased performance in predicting 94
postfire debris flow occurrence (Nikolopoulos et al., 2018). Despite the utility and advantages of 95
data-driven hazard prediction approaches, these techniques do not simulate the underlying physics, 96
which limits their utility in developing a better process-based understanding of debris flow 97
mechanics, limits their applicability in climatological and geographic settings different than their 98
training sites, and limits their use in non-stationary conditions (e.g., under changing climatic 99
conditions). 100
In contrast, physics-based models that simulate spatially-explicit hydrologic and mass wastage 101
processes are well-suited for mechanistic sensitivity analyses in diverse settings, but applications 102
of these models have tended to focus on landslide-induced debris flows (e.g., Iverson & George, 103
2014; George & Iverson, 2014), rather than runoff-generated debris flows which are more common 104
in postfire areas (Cannon et al., 2001, 2003; Santi et al., 2008). Studies that have investigated 105
postfire hydrologic responses using process-based models have largely focused on short-term 106
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
4
responses in individual catchments at high spatiotemporal resolutions (McGuire et al., 2016, 2017; 107
Rengers et al., 2016) or long-term runoff responses at coarse temporal resolutions (Rulli & Rosso, 108
2007; McMichael & Hope, 2007). For example, process-based models have employed shallow 109
water equations to understand the triggering and transport mechanisms of postfire debris flows in 110
single catchments (McGuire et al., 2016, 2017) and to investigate the timing of postfire debris 111
flows in three separate catchments (Rengers et al., 2016), the latter of which also assessed the 112
efficacy of a simplified kinematic wave approach. In addition to individual catchment applications, 113
process-based models often adopt simplifications that can limit effective prediction and hypothesis 114
testing to overcome computational limits. For example, the kinematic runoff and erosion model 115
(KINEROS2) simplifies drainage basins into 1-dimensional channels and hillslope patches 116
(Canfield & Goodrich, 2005; Goodrich et al., 2012; Sidman et al., 2015), and the Hydrologic 117
Modeling System (HEC-HMS) uses an empirically-based curve number method to estimate 118
saturation excess water (Cydzik et al., 2009), which cannot resolve infiltration excess overland 119
flow, a critical process in burn scars (Chen et al., 2013). 120
Given the current state of debris flow hazard assessment and prediction in previously burned 121
terrains, in addition to the growing influence of anthropogenic climate change on wildfire and 122
extreme precipitation, development of physics-based hazard assessment tools that can be used in 123
both hindcast investigations and forecasting applications is needed. Furthermore, due to the diverse 124
morphology of precipitation events and their interaction with geographically distributed wildfire 125
burn scars, development of tools that can assess hazards over regional domains, particularly in 126
operational forecasting applications, is critical. Here to advance the field of burn scar debris flow 127
hazard assessment, we explore the use of the physics-based and fully-distributed Weather Research 128
and Forecasting Hydrological modeling system version 5.1.1 (WRF-Hydro). WRF-Hydro is an 129
open-source community model developed by the National Center for Atmospheric Research 130
(NCAR). It is the core of National Oceanic and Atmospheric Administration’s (NOAA) National 131
Water Model forecasting system, and has been used extensively to study channelized flows (e.g., 132
Lahmers et al., 2020; Wang et al., 2019). Here, we modify WRF-Hydro to output high temporal 133
resolution fine-scale (100 m) debris flow-relevant overland flow; a process computed using a fully 134
unsteady, explicit, finite difference diffusive wave formulation. Previous efforts, employing 135
shallow water equations, diffusive, kinematic, and diffusive-kinematic wave models, have 136
demonstrated that water-only models can provide critical insights on runoff-driven debris flow 137
behavior (Arattano & Savage, 1994; McGuire & Youberg, 2020; Arratano & Franzi, 2010; Di 138
Cristo et al., 2021), even in burned watersheds (Rengers et al., 2016). 139
To test and demonstrate the utility of WRF-Hydro in debris flow studies, we investigate the 140
January 2021 debris flow events within the Dolan burn scar on the Big Sur coast of central 141
California (Fig. 1a–b). We first identify multiple debris flow sites using optical and radar remote 142
sensing data and field investigations. We then calibrate WRF-Hydro against ground-based soil 143
moisture and streamflow observations and use it to study the effects of burn scars on debris flow 144
hydrology and changes in hazard potential. The paper is organized as follows. Section 2 describes 145
our debris flow identification approach and historical context. Section 3 presents a description of 146
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
5
WRF-Hydro. Section 4 describes the simulation, calibration, and validation of WRF-Hydro. 147
Section 5 presents the results. Section 6 discusses the results and Sect. 7 provides a conclusion. 148
149
Fig. 1| WRF-Hydro model domain and Dolan burn scar. (a) WRF-Hydro model domain depicting 150
topography, 2020 wildfire season burn scars, and PSL soil moisture and USGS stream gage 151
observing sites. The black rectangle outlines (b) the Dolan burn scar inset, in which debris flow 152
locations and major streams are marked and labeled. 153
154
155
2 Study domain and debris flow identification methodology 156
The Dolan wildfire burned from August 18th till December 31st, 2020. 55% of areas within the fire 157
perimeter were burned at moderate-to-high severity (Burned Area Emergency Response, 2020). 158
After the fire, USGS Emergency Assessment of Postfire Debris-flow Hazards produced a debris 159
flow hazard assessment using a design storm based statistical model (USGS, 2020). On January 160
27–29, 2021, an atmospheric river (AR) made landfall on the Big Sur coast, bringing more than 161
300 mm of rainfall to California’s Coast Ranges (Fig. 2), with a peak rainfall rate of 24 mm h-1. 162
During the AR event, a section of California State Highway 1 (CA1) at Rat Creek was destroyed 163
by a debris flow. CA1 was subsequently closed for three months and rebuilt at a cost of ~$11.5M 164
(Los Angeles Times, 2021). 165
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
6
166
167
Fig. 2| The topography (shading; m) and MRMS accumulated precipitation (contour lines; mm) 168
during the AR event from January 27th 00:00 to 29th 23:00 in the Dolan burn scar. Contour line 169
interval for accumulated precipitation is 20 mm, and lines of 220, 260, and 300 mm are labeled. 170
The red polygon outlines the perimeter of the Dolan burn scar. 171
172
173
2.1 Debris flow identification from remote sensing and field work 174
In addition to the Rat Creek debris flow, which made national news (Los Angeles Times, 2021), 175
we identified three other debris flows using a combination of field investigation, and open access 176
satellite optical and synthetic aperture radar (SAR) images (Fig. 3 and Fig. B1). We examined 177
relative differences in normalized difference vegetation index (rdNDVI) defined by (Scheip & 178
Wegmann, 2021): 179 𝑟𝑑𝑁𝐷𝑉𝐼=
×100 (1) 180
where 𝑁𝐷𝑉𝐼 and 𝑁𝐷𝑉𝐼 are the pre- and post-event normalized difference vegetation index 181
(NDVI) images computed following: 182 𝑁𝐷𝑉𝐼=
(2) 183
where NIR is the near-infrared response and Red is the visible red response. rdNDVI was calculated 184
from Sentinel-2 satellite data using the HazMapper v1.0 Google Earth Engine application (Scheip 185
& Wegmann, 2021). HazMapper requires selection of an event date, pre-event window (months), 186
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
7
post-event window (months), max cloud cover (%) and slope threshold (°). These input 187
requirements filter the number of images used to calculate the rdNDVI. We set the event date to 188
27 January 2021 and used a 3 month pre- and post-event window with 0% max cloud cover and a 189
0° slope threshold to identify vegetation loss associated with the debris flows. We then created a 190
binary map to highlight debris flows (and other vegetation loss) pixels above a rdNDVI vegetation 191
loss threshold. We removed all pixels with rdNDVI > -10. 192
Lastly, we searched for debris flows (and other ground surface deformation) by examining SAR 193
backscatter change with data acquired by the Copernicus Sentinel-1 (S1) satellites (see full 194
description in Handwerger et al., in review). We measured the change in SAR backscatter by using 195
the log ratio approach, defined as 196 𝐼 =10×𝑙𝑜𝑔(
) (3) 197
where 𝜎
is a pre-event image stack (defined as the temporal median) of SAR backscatter and 198 𝜎
is a post-event image stack. Similar to the HazMapper method, our approach requires 199
selection of an event date, pre-event window (months), post-event window (months) and slope 200
threshold (°). No cloud-cover threshold is needed since SAR penetrates clouds. We used a 3 month 201
pre- and post-event window and 0° slope threshold to identify ground surface changes associated 202
with the debris flows. We then created a binary map to highlight debris flows by removing all 203
pixels with Iratio < 99th percentile value. 204
Identified debris flow source areas and deposition sites were confirmed by field investigation (N.J. 205
Finnegan) and named after the locations where they deposited (i.e., Big Creek, Mill Creek, and 206
Nacimiento). We note that there were likely more debris flows triggered during the AR event. 207
However, given the primary goal of this study – to demonstrate the utility of WRF-Hydro – a 208
comprehensive cataloging of debris flows is beyond this study’s scope. 209
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
8
210
Fig. 3| Identified debris flow sites using rdNDVI vegetation change within the Dolan burn scar. 211
We convert the rdNDVI data into a binary map by setting a threshold value, which yield only the 212
likely debris flow locations. (a)–(d) Sentinel-2 rdNDVI vegetation change at (a) Rat Creek, (b) 213
Mill Creek, (c) Big Creek, and (d) the Nacimiento River. 214
215
3 WRF-Hydro 216
3.1 Model description 217
WRF-Hydro is an open-source physics-based community model that simulates land surface 218
hydrologic processes. It includes the Noah-Multiparameterization (Noah-MP) land surface model 219
(LSM; Niu et al., 2011), terrain routing module, channel routing module, and a conceptual 220
baseflow bucket model. The Noah-MP LSM is a 1-dimensional column model that calculates 221
vertical energy fluxes (i.e., sensible and latent heat, net radiation), moisture (i.e., canopy 222
interception, infiltration, infiltration excess, deep percolation), and soil thermal and moisture states 223
on the LSM grid (1 km in our application). The infiltration excess, ponded water depth, and soil 224
moisture are then disaggregated using a time-step weighted method (Gochis & Chen, 2003) and 225
sent to the terrain routing module which simulates subsurface and overland flows on a finer terrain 226
routing grid (100 m in our application). According to the mass balance, local infiltration excess, 227
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
9
overland flow, and exfiltration from baseflow contribute to the surface head which flows into river 228
channels if defined retention depth is exceeded. The channel routing module then calculates 229
channelized flows assuming a trapezoidal channel shape (Fig. B2). Parameters related to the 230
trapezoidal channel, such as channel bottom width (Bw), Manning’s roughness coefficient (n), and 231
channel side slope (z) are functions of channel stream order (Fig. B3 and Table B1). Computed 232
streamflow is then output on the 100 m grid. Equations used to compute infiltration excess, 233
overland flow, and channelized flow are provided in Sect. 3.3 and 3.4. 234
By default, WRF-Hydro uses Moderate Resolution Imaging Spectroradiometer (MODIS) 235
Modified International Geosphere-Biosphere Program (IGBP) 20-category land cover product as 236
land cover (Fig. B4) and 1-km Natural Resources Conservation Service State Soil Geographic 237
(STATSGO) database for soil type classification (Fig. B5; Miller & White, 1998). Land surface 238
properties including canopy height (HVT), maximum carboxylation rate (VCMX25), and overland 239
flow roughness (OV_ROUGH2D) are functions of land cover type (Table B2 & Fig. B4). Default 240
soil hydraulic parameters in WRF-Hydro (i.e., soil porosity, grain size distribution index, and 241
saturated hydraulic conductivity) are based on Cosby et al.’s (1984) soil analysis (Table B3) and 242
are used to map onto the STATSGO 16 soil texture types (Fig. B5). 243
244
3.2 Meteorological forcing files 245
WRF-Hydro is used in standalone mode (i.e., it is not interactively coupled with the atmospheric 246
component of WRF), but rather is forced with a combination of Phase 2 North American Land 247
Data Assimilation System (NLDAS-2) meteorological data and Multi-Radar/Multi-Sensor System 248
(MRMS) radar-only quantitative precipitation (Zhang et al., 2011, 2014, 2016). A description of 249
the MRMS dataset and uncertainties therein can be found in Appendix A. NLDAS-2 provides 250
hourly forcing data including incoming shortwave and longwave radiation, 2-m specific humidity 251
and air temperature, surface pressure, and 10-m wind speed at 1/8-degree spatial resolution. 252
MRMS provides hourly precipitation rate at 1-km resolution. 253
254
3.3 Overland flow routing and output 255
The Noah-MP LSM calculates rate of infiltration excess following Chen & Dudhia (2001): 256
257
=
1− ∑∆()
– (
)
∑∆()
– (
) (4) 258
259
where h (m) is the surface water depth and t is the time. 𝑃 (m) is the precipitation not intercepted 260
by the canopy; ∆𝐷 (m) is the depth of soil layer i; 𝜃 is the soil moisture in soil layer i; 𝜃 is the 261
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
10
soil porosity; 𝐾 (m s-1) is the saturated hydraulic conductivity; 𝐾 is 2 × 10−6 m s-1 which 262
represents the saturated hydraulic conductivity of the silty–clay–loam soil texture chosen as a 263
reference; 𝛿 (s) is the model time step; and k which is equal to 3.0 is the runoff–infiltration 264
partitioning parameter [the same as 𝑘𝑑𝑡 in Chen & Dudhia (2001)]. 265
266
Noah-MP passes excess water to the terrain routing module, which simulates overland flow using 267
a 2-dimensional fully-unsteady, explicit, finite-difference diffusive wave equation adapted from 268
Julien et al. (1995) and Ogden (1997). It is considered superior to the traditionally used kinematic 269
wave formulation in that it accounts for backwater effects and flow over adverse slopes. The 270
diffusive wave formulation is the simplified form of the Saint Venant equations, i.e., continuity 271
and momentum equations for a shallow water wave. The 2-dimensional continuity equation for a 272
flood wave is: 273
+
+
= 𝑖 (5) 274
where h is the surface flow depth, 𝑞 and 𝑞 are the unit discharges in the x- and y-directions, 275
respectively, and 𝑖 is the infiltration excess. Manning’s equation which considers momentum loss 276
is used to calculate 𝑞. In the x-direction: 277 𝑞=𝛼ℎ (6) 278
Where 𝛽 is a unit dependent coefficient equal to
, and 279
𝛼=
/
(7) 280
where 𝑛 is the tunable overland flow roughness coefficient. The momentum equation in the x-281
direction is given by: 282 𝑆 = 𝑆−
(8) 283
where 𝑆 is the friction slope, 𝑆is the terrain slope, and
is the change in surface flow depth 284
in the x-direction. 285
Off-the-shelf, WRF-Hydro does not output overland flow at terrain routing grids (100 m), however 286
it is computed in the background to determine channelized streamflow. One key advance made in 287
this work is that we modified WRF-Hydro source code to output overland flow. Overland flow 288
depth (m) was converted to overland discharge (m3 s-1) by multiplying flow depth by grid cell area 289
(10,000 m2) and dividing by the LSM time step (1 h). 290
291
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
11
3.4 Channel routing 292
If overland flow intersects grid cells identified as channel grids [2nd Strahler stream order and 293
above; pre-defined by the hydrologically conditioned USGS National Elevation Dataset 30-m 294
digital elevation model (DEM)], the channel routing module routes the water as channelized 295
streamflow using a 1-dimensional, explicit, variable time-stepping diffusive wave formulation. 296
Similarly, the continuity equation for channel routing is given as: 297
+
= 𝑞 (9) 298
and the momentum equation is given as: 299
+(
)
+ 𝑔𝐴
= −𝑔𝐴𝑆 (10) 300
where s is the streamwise coordinate, H is water surface elevation, A is the flow cross-sectional 301
area calculated as (𝐵+ 𝐻 𝑧)𝐻 (Fig. B2), 𝑞 is the lateral inflow rate into the channel grid, Q is 302
the flow rate, 𝛾 is a momentum correction factor, 𝑔 is acceleration due to gravity, and 𝑆 is the 303
friction slope computed as: 304 𝑆= (
) (11) 305
where K is the conveyance computed from the Manning’s equation: 306 𝐾=
𝐴𝑅/ (12) 307
where n is the Manning’s roughness coefficient, A is the channel cross-sectional area, R is the 308
hydraulic radium (A/P), P is the wetted perimeter, and 𝐶is a dimensional constant (1.486 for 309
English units or 1.0 for SI units). 310
311
4 Model simulation, calibration, and validation 312
4.1 Model domain 313
The WRF-Hydro model domain spans regions in California including the Coast Ranges, Monterey 314
Bay, and the Central Valley, and covers several burn scars from the 2020 wildfire season (Fig. 1a). 315
Here we focus our analysis on the Dolan burn scar where the hazardous debris flows occurred (Fig. 316
1b). According to the USGS 30-m DEM, the Rat Creek debris flow site sits at the base of a 1st 317
order catchment with a drainage area of 2.23 km2. Mill Creek, Big Creek, and Nacimiento debris 318
flows were initiated within extremely steep, intensely burned, 1st order catchments, but were 319
deposited in 2nd, 3rd, and 3rd Strahler stream order channels, respectively. 320
To calibrate and validate WRF-Hydro output, we use soil moisture observations from two Physical 321
Sciences Laboratory (PSL) monitoring stations [i.e., Lockwood (lwd) and Gilroy (gry)] (Fig. 1a). 322
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
12
Due to the Mediterranean climate of California, many USGS stream gages experience low or no 323
flow during the dry season. In addition, many gages are under manual regulation to mitigate wet-324
season flood risks and better distribute water resources. As such, it can be challenging to obtain 325
natural streamflow observations for model calibration. Here, three USGS stream gages [i.e., 326
Arroyo Seco NR Greenfield, CA (ID 11151870), Arroyo Seco NR Soledad, CA (ID 11152000), 327
and Arroyo Seco BL Reliz C NR Soledad, CA (ID 11152050)] (Fig. 1a) on streams that have 328
measurable flows during our study period and are free of human regulation are used. These gages 329
are located downstream of the Dolan burn scar and hence are useful in calibrating the parameters 330
associated with burn scar effects. The PSL soil moisture observations were recorded at 2-minute 331
intervals and USGS streamflow gage data were recorded at 15-minute intervals, but we perform 332
all observation-model comparisons at hourly-mean resolution. 333
334
335
4.2 Baseline simulation and soil moisture calibration 336
WRF-Hydro was run from January 1–31 of 2021. We performed the baseline simulation by 337
modifying WRF-Hydro default parameters (Table B3) based on a calibration using soil moisture 338
observations from stations lwd and gry. Neither PSL station is located in a burn scar. Since the 339
baseline simulation includes no postfire characteristics, it can also be regarded as the “pre-fire” 340
scenario. Soil moisture at 10 cm below ground in the baseline simulation was calibrated by 341
performing a domain-wide adjustment of soil porosity and grain size distribution index at the 342
simulation start (Table B3). We then allowed the model to spin up from January 1–10 before using 343
January 11–31 for validation. Using a relatively short spin-up period is justified because prior to 344
the AR event, little rain fell on the Dolan burn scar (i.e., ~400 mm of rainfall fell from June to 345
December 2020). As such, in the months preceding the debris flow events, soil moisture 346
observations indicate already dry condition prior to our 10 day spin up. 347
348
After calibration, the simulated soil moisture closely mimics ground-based PSL observations (Fig. 349
4). Both the observed magnitude and variability are well captured, with the simulated ±1 standard 350
deviation envelope largely encompassing PSL observations during the AR. Model performance 351
was evaluated using four quantitative metrics, i.e., correlation coefficient, root mean square error, 352
mean bias, and Kling-Gupta efficiency (KGE; Gupta et al., 2009; Kling et al., 2012). KGE has 353
previously been used in soil moisture calibration applications (e.g., Lahmers et al., 2019; 354
Vergopolan et al., 2020) and is computed as follows: 355
356 𝐾𝐺𝐸= 1 − (𝑟−1)+(𝛼−1)+(𝛽−1) (13) 357
358
where r is the correlation coefficient between the observation and simulation, 𝛼 is the ratio of the 359
standard deviation of simulation to the standard deviation of observation, and 𝛽 is the ratio of the 360
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
13
mean of simulation to the mean of observation. KGEs close to 1 indicate a high-level consistency 361
between the simulation and observation, while negative KGEs indicate poor model performance 362
(Schönfelder et al., 2017; Andersson et al., 2017). 363
364
The model’s ability to simulate soil moisture substantially improves after calibration (Fig. 4; Table 365
1). KGE values approach 1 (0.72 at lwd and 0.88 at gry), indicating that WRF-Hydro adequately 366
simulates the hydrologic environment and its response to meteorological change. 367
368
369
370
371
Fig. 4| Precipitation, observed and simulated soil moisture at two PSL soil moisture stations. 372
January 11–31, 2021 MRMS precipitation (green bars) and observed (black line) and simulated 373
volumetric soil moisture 10 cm below ground in the baseline simulation (purple dashed line) at 374
PSL sites (a) Lockwood (lwd) and (b) Gilroy (gry). Envelope of purple shading depicts ±1 standard 375
deviation of model simulated soil moisture. KGE scores are provided at top left for each station. 376
377
378
379
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
14
Table 1 380
Evaluation metrics of simulated soil moisture and streamflow 381
382
Soil moisture (Default / Baseline)
Station r RMSE Bias KGE
lwd 0.97 / 0.98 7.06 / 4.32 5.21 / 4.16 0.10 / 0.72
gry 0.94 / 0.94 5.19 / 2.53 -4.79 / -1.66 0.80 / 0.88
Streamflow (Baseline / Burn scar)
Station r RMSE Bias NSE
1870 0.28 / 0.93 39.29 / 14.69 1.65 / 3.36 -0.17 / 0.84
2000 0.26 / 0.86 51.22 / 24.92 2.47 / 4.81 -0.15 / 0.73
2050 0.25 / 0.81 49.96 / 27.43 5.70 / 8.24 -0.38 / 0.53
383
Table 1| Quantitative evaluation metrics for the simulated soil moisture and streamflow when 384
compared against observations. The metrics include the Pearson correlation coefficient (r), root 385
mean square error (RMSE), and mean bias (Bias). In addition, the comprehensive metrics Kling-386
Gupta efficiency (KGE) and Nash-Sutcliffe efficiency (NSE) are used to evaluate model-simulated 387
soil moisture and streamflow, respectively. For soil moisture, the numbers in front of “/” are 388
calculated between the default run (i.e., uncalibrated run) and the observations, whereas the 389
numbers following “/” are the corresponding values in the baseline simulation (the purple dashed 390
line in Fig. 4). For streamflow, the numbers in front of “/” are computed between the baseline run 391
(purple dashed line in Fig. 6) and the observations, while the numbers behind “/” are for burn scar 392
simulation (red line in Fig. 6). If the model performance regarding a certain metric is enhanced in 393
the burn scar simulation, the number after “/” is underlined. 394
395
4.3 Burn scar simulation and streamflow calibration 396
To simulate effects of wildfire burn scars on hydrologic processes and debris flow hazards, we 397
made two modifications to the baseline simulation soil moisture calibrated model configuration. 398
First, we changed the land cover type within the burn scar perimeter to its nearest LSM analogue, 399
i.e., “barren and sparsely vegetated”. The switch to barren land causes: (1) height of the canopy 400
(HVT) to decrease to 0.5 m; (2) maximum rate of carboxylation at 25°C (VCMX25) to decrease 401
to 0 𝜇𝑚𝑜𝑙 𝐶𝑂/(𝑚∙𝑠); and (3) overland flow roughness coefficient (OV_ROUGH2D) to decrease 402
to 0.035 (Fig. 5a–c) from default values (Fig. B4 and Table B2). 403
404
The second adjustment was to decrease soil infiltration rates within the burn scar perimeter, 405
achieved by reducing soil saturated hydraulic conductivity (DKSAT; Fig. 5d; Scott & van Wyk, 406
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
15
1990; Cerdà, 1998; Robichaud, 2000; Martin & Moody, 2001) from default values (Table B3). 407
Consistent with the hydrophobicity of burned soils, we calibrate the burn scar simulation by 408
systematically exploring a range of burn scar area saturated hydraulic conductivities (0 to 3×10 -7 409
m s-1 with a 5×10-8 m s-1 increment), with the goal of reproducing streamflow behavior similar to 410
USGS gage observations. We found that a value of 1.5×10-7 m s-1 gives the highest Nash-Sutcliffe 411
efficiency (NSE; Nash & Sutcliffe, 1970) across all three USGS stream gages (Table 1). NSE and 412
KGE are the two most widely used metrics for calibration and evaluation of hydrologic models. 413
The NSE has previously been used in streamflow calibration applications (e.g., Xia et al., 2012; 414
Bitew & Gebremichael, 2011), and it is calculated as follows: 415
416 𝑁𝑆𝐸= 1 − ∑(()())
∑(() )
(14) 417
418
where 𝑇 is the length of the time series, 𝑄(𝑡) and 𝑄(𝑡) are the simulated and observed 419
discharge at time 𝑡, respectively, and 𝑄 is the mean observed discharge. By definition, NSEs of 420
1 indicate perfect correspondence between the simulated and observed streamflow. Positive NSEs 421
mean that the model streamflow has a greater explanatory power than the mean of the observations, 422
whereas negative NSEs represent poor model performance (e.g., Moriasi et al., 2007; Schaefli & 423
Gupta, 2007). When burn scar characteristics are included, NSEs increase from negative values in 424
the baseline to greater than 0.5, and the NSEs at gages 1870 and 2000 reach 0.84 and 0.73, 425
respectively. Higher NSE scores indicate the abovementioned burn scar parameter changes 426
improve the model’s ability to simulate streamflow observations downstream of the burn scar 427
(Table 1). 428
429
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
16
430
Fig. 5| Parameter setting in the WRF-Hydro burn scar simulation. (a) The height of the canopy 431
(HVT; m; shading), (b) maximum rate of carboxylation at 25°C (VCMX25; 𝜇𝑚𝑜𝑙 𝐶𝑂/(𝑚∙𝑠); 432
shading), (c) overland flow roughness coefficient (OV_ROUGH2D; shading), and (d) saturated 433
hydraulic conductivity (DKSAT; m s-1; shading) in the burn scar simulation. 434
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
17
435
436
Fig. 6| Precipitation, observed and simulated streamflow at three USGS stream gages. January 26–437
31, 2021 MRMS precipitation (green bars), observed (black dash dotted line) and simulated 438
streamflow in baseline simulation (purple dashed line) and burn scar simulation (red line) at (a) 439
Arroyo Seco NR Greenfield, CA (ID 11151870), (b) Arroyo Seco NR Soledad, CA (ID 11152000), 440
and (c) Arroyo Seco BL Reliz C NR Soledad, CA (ID 11152050). NSE scores for baseline (purple) 441
and burn scar simulations (red) are shown at top left. 442
443
5 Results 444
5.1 Hydrologic response due to burn scar incorporation 445
The pre-fire baseline simulation fails to capture the hydrologic behavior observed at the USGS 446
gages located within the burn scar (Fig. 6). Incorporation of burn scar characteristics substantially 447
alters the hydrologic response of the model and provides much higher fidelity streamflow 448
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
18
simulations (Fig. 6). Observed hydrographs are characterized by two early streamflow peaks 449
related to two precipitation bursts on January 27th and 28th. Our burn scar simulation captures this 450
behavior, while the baseline simulation streamflow peaks just once, with a lower magnitude and 451
an ~3-day lag after peak precipitation (Fig. 6). The steep rising limbs and high magnitude discharge 452
peaks of the burn scar hydrograph are indicative of flash flooding. Compared with the pre-fire 453
baseline scenario, the burn scar’s barren land and low infiltration rate substantially accelerate 454
drainage rates and increase discharge volume into stream channels. 455
456
5.2 Hydrologic response at four debris flow sites 457
We identified locations and extent of four debris flows from remote sensing data and field work 458
(Fig. 3& Fig. B1). rdNDVI shows vegetation loss caused by debris flows. Mill Creek, Big Creek, 459
and Nacimiento were relatively large debris flows with runout lengths between ~2–5 km. Rat 460
Creek occurred in a smaller catchment and had a runout length of ~300 m. The difference in runout 461
length and debris flow size is primarily controlled by upstream catchment size. Due to its low 462
stream order (1st Strahler stream order), Rat Creek is the only debris flow site modeled entirely as 463
overland flow in our WRF-Hydro simulations. 464
At the four debris flow sites, we use three metrics to characterize hydrologic anomalies: (1) 465
accumulated runoff volume, (2) peak discharge, and (3) time to peak discharge. Fig. 7 depicts 466
accumulated channelized discharge volume (blue shading) and accumulated overland discharge 467
volume (yellow-red shading) from January 27th 00:00 to 28th 12:00 near the four debris flow sites 468
in the burn scar simulation. Accumulation time period is chosen such that it covers the first two 469
runoff surges in the simulated hydrographs which are likely associated with debris flows (Fig. 8) 470
given that nearly concurrent peak rainfall intensity and peak discharge is a signature characteristic 471
of debris flows (Kean et al., 2011). Runoff volume is on the order of 104 m3 at Rat Creek and 106
472
m3 at the other three sites. 473
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
19
474
Fig. 7| WRF-Hydro simulated overland flow and streamflow in the burn scar simulation. (a)–(d) 475
Total volume of accumulated overland flow (yellow-red shading) and streamflow (blue shading) 476
on log10 scale between January 27th 00:00 and 28th 12:00 at four debris flow sites. Black rectangles 477
correspond to domains in Fig. 3a–d. Black circles and triangles indicate debris flow source areas 478
and deposits, respectively. 479
480
481
Dramatic hydrographic changes after inclusion of burn scar characteristics are simulated at debris 482
flow source areas (Fig. B6 and Table B4) and deposition sites (Fig. 8 and Table 2). WRF-Hydro 483
facilitates investigation of the hydrologic response at triggering and deposition locations and along 484
the runout path. Here, to emphasize the downstream hazards, our analysis is focused on debris 485
flow deposits. At Rat Creek, where a section of CA1 collapsed, the magnitude of discharge 486
substantially increases, and overland flow surges are concurrent with rainfall bursts (Fig. 8a). Total 487
discharge accumulated during the AR event increases approximately eight-fold (791%), and peak 488
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
20
discharge more than triples compared to the baseline simulation (Fig. 8a and Table 2). At Mill 489
Creek, Big Creek, and Nacimiento, baseline hydrographs are characterized by less variability, 490
muted responses to two early precipitation bursts, and a delayed third discharge peak that does not 491
occur until ~3 days after AR passage (Fig. 8b–d). Maximum discharge peaks in the baseline 492
hydrographs lag those in the burn scar simulation by ~2 days (Fig. 8b–d; Table 2). In the burn scar 493
simulation, total volume substantially increases at the three channelized sites – total volume 494
increases ~650% at Mill Creek, ~891% at Big Creek, and ~829% at Nacimiento (Fig. 8b–d and 495
Table 2), and the absolute increase in volume is on the order of 106 m3 (Table 2). Peak discharge 496
more than triples at Mill Creek and Big Creek and more than quadruples at Nacimiento. 497
Additionally, response times of the peak in discharge to the peak in precipitation decrease to less 498
than an hour, highlighting the simulated flashiness of the burned catchments. 499
500
501
502
503
504
Fig. 8| WRF-Hydro simulated discharge time-series at four debris flow deposition locations. (a)–505
(d) MRMS precipitation (green bars) and simulated discharge time-series for January 26th 00:00 506
to 31st 23:00 at (a) Rat Creek, (b) Mill Creek, (c) Big Creek, and (d) Nacimiento deposition 507
locations (black triangles in Fig. 7a–d) in baseline simulation (purple dashed line) and burn scar 508
simulation (red line). 509
510
511
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
21
Table 2 512
The total runoff volume, peak discharge, and peak timing at debris-flow deposits 513
Site name
Baseline simulation Burn scar simulation
Total
volume
(m3)
Peak
discharge
(m3 s-1)
Highest
peak
timing
Total
volume
(m3)
Peak
discharge
(m3 s-1)
1st Peak
timing
2nd Peak
timing
Rat Creek 6,897 0.54 28th 05:00 61,425
(+791%)
1.73
(+220%) 27th 09:00 28th 05:00
Mill Creek 312,925 13.10 29th 08:00 2,347,457
(+650%)
45.21
(+245%) 27th 13:00 27th 23:00
Big Creek 842,808 46.10 29th 16:00 8,354,095
(+891%)
154.10
(+234%) 27th 10:00 28th 05:00
Nacimiento 743,531 33.15 29th 16:00 6,904,706
(+829%)
135.41
(+308%) 27th 14:00 28th 00:00
514
Table 2| The total runoff volume, peak discharge, and peak timing in the baseline and burn scar 515
simulations from January 27th 00:00 to 31st 23:00 at deposition sites of Rat Creek, Mill Creek, Big 516
Creek, and Nacimiento debris flows (black triangles in Fig. 7a–d). The peak timing shown in the 517
baseline simulation is for the highest peak. The percent change of the total volume and peak 518
discharge in the burn scar simulation relative to the baseline simulation are shown in parentheses. 519
520
521
5.3 Debris flow hazard assessment for the Dolan burn scar 522
Since high magnitude runoff is often the cause and precursor of runoff-generated debris flows in 523
burned areas (Cannon et al., 2003, 2008; Rengers et al., 2016), we use simulated accumulated 524
volume of overland flow and streamflow to assess runoff-generated debris flow hazard potential 525
under pre-fire (i.e., baseline; Fig. 9a&d) and postfire (i.e., burn scar simulation; Fig. 9b&e) 526
conditions. We assess changes at both stream and catchment levels and use the difference between 527
burn scar and baseline simulations to assess added debris flow hazard potential (Fig. 9c&f). 528
Consistent with the increasing erosive and entrainment power associated with increasing discharge, 529
our debris flow hazard increases as the accumulated discharge volume increases. To reduce the 530
effects of catchment size on the volume-based hazard levels, we normalize a catchment’s discharge 531
volume by the area of the catchment (Santi et al., 2012; Fig. 9d–f). Non-normalized catchment 532
hazard maps are also provided (Fig. B7). 533
534
In the pre-fire baseline simulation, the AR-induced precipitation produces lower debris flow 535
hazard over most of the domain, but elevated hazards along stream channels (Fig. 9a). We note no 536
substantial differences between areas in or out of the burn scar. In the burn scar simulation, debris 537
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
22
flow hazard levels increase across the Dolan burn scar and along channels outside but downstream 538
of the burn scar (Fig. 9b–c). The discharge volume increases by an order of magnitude near Rat 539
Creek, Big Creek, Mill Creek, and Nacimiento. Within the burn scar, hazards along major stream 540
channels, such as the Nacimiento River and San Antonio River increase. Outside the burn scar, 541
hazard levels along river channels downstream of the burn scar, such as the Arroyo Seco River, 542
also increase (Fig. 9c). 543
544
At the catchment level, debris flow hazards are assessed using accumulated discharge volumes 545
normalized by catchment areas (Fig. 9d–f). Accumulated discharge volumes are assessed at the 546
outlet of each catchment between January 27th 00:00 to 28th 12:00. In the baseline simulation, the 547
majority of catchments are subject to relatively low debris flow hazards compared to the burn scar 548
simulation with total normalized discharge volume less than 103 m3 km-2 (Fig. 9d). In the burn scar 549
simulation, over half of catchments within the Dolan burn scar have normalized discharge volume 550
greater than 105 m3 km-2, while over 1/4 of basins reach 106 m3 km-2 (Fig. 9e). The additional 551
debris flow hazard brought about by the inclusion of wildfire burn scar characteristics is substantial 552
(Fig. 9f). 553
To summarize changes in debris flow hazards as a result of including burn scar characteristics in 554
WRF-Hydro simulations, we create distributions of pre-fire baseline and burn scar catchment-area 555
normalized discharge from the 404 catchments located within the Dolan burn scar perimeter (Fig. 556
10). After incorporating burn scar characteristics, the full distribution shifts to the right, indicating 557
increased hazard levels – a shift considered robust by a Student’s t-test (p value: 4.6E-45). A 558
quantitative assessment of this shift indicates that the mean catchment area normalized discharge 559
volume increases by ~1300% (from ~380k to 5.5M m3 km-2) while the standard deviation increases 560
~1400% (from ~1.6M to 23.0M m3 km-2). We also assess shifts at a range of distribution 561
percentiles: 5P: 148% (~0.6k to ~1.5k m3 km-2), 25P: 725% (~3.7k to ~30.7k m3 km-2), 50P: 924% 562
(~13k to ~135k m3 km-2), 75P: 980% (~120k to ~1.3M m3 km-2), and 95P: 1300% (~2.1M to 563
~29.1M m3 km-2). In the burn scar simulation, more than half of catchments have normalized 564
volumes > 105 m3 km-2 and more than 1/4 of catchments have volumes > 106 m3 km-2 – values that 565
correspond to the 75P and 90P of the baseline simulation, respectively. Disproportionate shifting 566
of the right tail of the distribution suggests that extreme debris flow hazards increase non-linearly 567
under simulated burn scar conditions. 568
Our catchment-area normalized discharge volume-based hazard assessment also indicates that the 569
catchments containing Mill Creek, Big Creek, and Nacimiento had elevated hazard potential (Fig. 570
9d–f), consistent with our (limited) debris flow observations. Other areas with elevated hazards 571
include catchments containing the Arroyo Seco and San Antonio Rivers. Beyond the burn scar 572
perimeter, effects of fire expand to adjacent and downstream catchments, and the drainage basins 573
of the Arroyo Seco and Nacimiento Rivers are simulated to have potentially hazardous conditions, 574
i.e., normalized discharge volumes in excess of 106 m3 km-2 (Fig. 9e&f). 575
576
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
23
577
578
Fig. 9| Discharge volume-based runoff-generated debris flow hazards. Debris flow hazards at 579
individual stream level for the (a) baseline, (b) burn scar, and (c) difference between burn scar and 580
baseline simulations. Hazard is estimated as total discharge volume from January 27th 00:00 to 581
28th 12:00. (d)–(f) Normalized debris flow hazards by catchment area at catchment level. For each 582
catchment, the hazard is determined by total discharge volume at the catchment outlet from 583
January 27th 00:00 to 28th 12:00 divided by catchment area. 584
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
24
585
Fig. 10| Distributions of accumulated discharge volumes at the outlet of the 404 catchments 586
normalized by upstream catchment areas within Dolan burn scar in the baseline simulation (purple 587
line) and in the burn scar simulation (red line). Dashed vertical lines indicate median values. 588
5.4 Debris flow hazard assessment at regional scales 589
While the results we present above primarily focus on hazards in the Dolan burn scar, our WRF-590
Hydro domain includes a number of additional 2020 wildfire burn scar sites (Fig. 1a). Given the 591
long filament-like structure of western U.S. landfalling ARs, the heterogeneous nature of 592
landfalling trajectories, and the potential for systems to interact with diverse topographic terrains, 593
the development of tools capable of regional hazard assessments under high-gradient precipitation 594
events is crucial – particularly in a wildfire-prone region like California. To demonstrate the 595
potential utility of WRF-Hydro in regional applications, we assess hazards over our full domain 596
(Fig. 11). We find that hazard potential, from both channelized and overland flows, is greatest 597
within the burn scar sites, with maximum hazards found in the Dolan burn scar, consistent with 598
the location of elevated precipitation along the Coast Ranges – where more than 300 mm of rain 599
fell over three days (Fig. 11). Other high hazard-elevated precipitation regions within our domain 600
include the western edge of the Sierra Nevada and areas north of Monterey Bay, which collocate 601
with the Mineral and Del Puerto burn scars, respectively. Similar to our Dolan burn scar focused 602
analysis, areas within and downstream of these burn scar sites have elevated streamflow discharge 603
volumes compared to the non-burned areas (Fig. 11b). Likewise, areas of heightened accumulated 604
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
25
overland flow are elevated in burn scar regions, but also demonstrate a strong correspondence to 605
the spatial distribution of precipitation (Fig. 11a & c). 606
607
Fig. 11| MRMS accumulated precipitation and regional debris flow hazard assessment. (a) MRMS 608
accumulated precipitation during January 27th 00:00 to 29th 23:00 over the model domain 609
(shading; mm). Names of burn scars are labeled in black. (b) Accumulated streamflow (yellow-to-610
red shading; m3) and (c) accumulated overland flow from 27th 00:00 to 28th 12:00 over the model 611
domain (yellow-to-red shading; m3). Wildfire perimeters of 2020 wildfire season are outlined in 612
black in (a), and in blue in (b) and (c). The coastline of California is in grey. 613
614
6 Discussion 615
Given the historic and growing frequency of wildfires in the western U.S. (Swain 2021; Williams 616
et al., 2019; Goss et al., 2020) and globally (Jolly et al., 2015; Flannigan et al., 2013), developing 617
tools to investigate, better understand, and potentially predict changes in burn scar hydrology and 618
natural hazards at regional scales is critical. Here, we demonstrate the first use of WRF-Hydro to 619
simulate the surface hydrologic response over a burn scar during a landfalling AR. We augmented 620
the default version of WRF-Hydro to output overland flow and to replicate burn scar behavior by 621
adjusting vegetation type and infiltration rate parameters. WRF-Hydro simulations were validated 622
against PSL soil moisture and USGS streamflow observations before we used simulated 623
streamflow and overland flow volumes to characterize debris flow hazard potential. 624
625
A comparison between baseline and burn scar simulations demonstrated that changes in hydraulic 626
properties of burned areas causes drastic changes in surface flows, including faster discharge 627
response times, greater discharge volumes, and overall flashier hydrologic behavior in surface 628
flows. As a result of including bur scar characteristics in WRF-Hydro simulations, median 629
catchment-area normalized discharge volume increases nine-fold, while 95P volume increases 13-630
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
26
fold. The magnitude of our simulated changes is consistent with findings from previous postfire 631
hydrology studies (Anderson et al., 1976; Scott, 1993; Meixner & Wohlgemuth, 2003; Kinoshita 632
& Hogue, 2015; Kean et al., 2011). At Rat Creek, where a debris flow destroyed CA1, our model 633
simulation predicted an eight-fold increase in accumulated overland flow and a tripling in peak 634
discharge when compared to the baseline simulation. At Mill Creek, Big Creek, and Nacimiento, 635
the increase of runoff volume from the baseline to the burn scar simulation is on the order of 106 636
m3. Our hazard assessments based on catchment-area normalized discharge volumes indicated that 637
Mill Creek, Big Creek, and Nacimiento were under elevated debris flow hazards, corresponding 638
well with identified debris flow occurrences. 639
640
Despite methodological differences, our debris flow hazard assessment for this AR event is 641
generally consistent with the USGS’ postfire, pre-AR, design-storm-based preliminary hazard 642
assessment (USGS, 2020). As described above, USGS preliminary hazard assessments use logistic 643
regression models to estimate the likelihood of debris flow occurrence and multivariate linear 644
regression models to estimate debris flow volumes. This empirical approach is trained on historical 645
western U.S. debris flow occurrence and magnitude data and incorporates estimated burn scar soil 646
erodibility and burn severity data (Cannon et al., 2010; Gartner et al., 2014; Staley et al., 2016). 647
For precipitation, the USGS assessment utilizes a design storm approach that assumes 1–5 year 648
return interval magnitude precipitation falls uniformly over a region/burn scar (USGS, 2020). For 649
the Dolan burn scar, both assessments find that large stream channels had relatively higher hazard 650
levels than small streams or overland areas. However, close comparison of hazard maps reveals 651
differences in spatial distribution of high-hazard catchments. In the USGS assessment, higher 652
hazard levels are predicted north and southeast of the burn scar, whereas in our assessment the 653
highest hazards occur along major stream channels. We hypothesize that USGS-assessed areas of 654
higher hazard potential are related to their use of design-storm precipitation (see Fig. 2 for MRMS 655
precipitation footprint) and burn severity data (Burned Area Emergency Response, 2020). 656
Comparison with the USGS assessment framework suggests room for improvement in WRF-657
Hydro-based assessments (i.e., inclusion of burn severity and soil erodibility data), but also 658
highlights the potential utility of working with spatially-distributed and time-varying precipitation. 659
However, this also means the accuracy of WRF-Hydro predictions depends on the accuracy of 660
precipitation forcing, and in our hindcast application, MRMS precipitation data (Appendix A). 661
Accordingly, our WRF-Hydro-based hazard assessment could benefit from precipitation products 662
mosaiced from various sources to constrain precipitation-based uncertainties (e.g., gauge-663
corrected and/or Mountain Mapper MRMS), although the long processing time of these datasets 664
inhibits timely post-event assessments. 665
666
As a water-only model, WRF-Hydro is currently restricted to simulating the hydrologic ingredients 667
of debris flows. While water-only models have been widely used to investigate and better 668
understand debris flow dynamics (Arattano & Savage, 1994; Arattano & Franzi, 2010; Rengers et 669
al., 2016; McGuire & Youberg, 2020; Di Cristo et al., 2021), sediment supply, soil erodibility, and 670
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
27
other sedimentological factors also play important roles in determining the potential for and 671
severity of mass failure events (McGuire et al., 2017). Developing a debris flow model that couples 672
hydrologic and sediment erosion and transport processes would represent a significant advance 673
and be of great practical use (Banihabib et al., 2020; Shen et al., 2021). At a minimum, soil grain 674
size maps and domain-specific rainfall intensity-duration curves can provide guidance to define 675
transitions from water floods to debris flows if historical debris flow data is available in the study 676
domain (McGuire & Youberg, 2020; Tognacca et al., 2000; Gregoretti & Fontana, 2008; Cannon 677
et al., 2007). 678
679
7 Conclusion 680
681
Use of WRF-Hydro to simulate runoff-generated debris flow hazards in burn scar settings 682
represents a novel application. It is notable that in this application we have balanced the 683
computational cost of a regional domain with our choice of resolved spatial resolution for terrain 684
routing and overland flow calculations (100 m). However, WRF-Hydro has previously been 685
applied to smaller domains at higher terrain routing resolutions (~30 m). Future work could assess 686
the use of the model to study burn scar hydrology at finer spatial scales, should the application 687
warrant and should underlying data at sufficient resolution exist. Other potential applications of 688
our modified model framework include alpine areas and steep hillslopes with sparse vegetation 689
where runoff-generated debris flows dominate over landslide-initiated ones (Davies et al., 1992; 690
Coe et al., 2003, 2008). 691
692
Further, our burn scar parameter changes are performed to Noah-MP, which is the core land 693
surface component of the National Centers for Environmental Prediction Global Forecast System 694
(GFS) and Climate Forecast System (CFS), thus the findings presented herein, are likely to prove 695
useful in the broader worlds of forecast meteorology and climate science. In addition, here WRF-696
Hydro is driven by historical precipitation and meteorological data, i.e., in hindcast mode. We see 697
no reason why this modeling framework could not also be employed to project hazards under 698
future climatic conditions (e.g., Huang et al., 2020), or given its relatively low computational 699
expense, in operational forecast mode. Indeed, modern ensemble-based meteorological forecasting 700
could provide high spatiotemporal forcing data with which disaster preparedness managers could 701
probabilistically assess debris flow hazard potential, and issue advanced life and property saving 702
warnings. 703
704
705
706
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
28
Appendix A 707
Text A1. Multi-Radar/Multi-Sensor System (MRMS) radar-only precipitation estimate and 708
uncertainty 709
MRMS is a precipitation product that covers the contiguous United States (CONUS) on 1-km grids. 710
It combines precipitation estimates from sensors and observational networks (Zhang et al., 711
2011, 2014, 2016), and is produced at the National Centers for Environmental Prediction (NCEP) 712
and distributed to National Weather Service forecast offices and other agencies. Input datasets 713
used to produce MRMS include the U.S. Weather Surveillance Radar-1988 Doppler (WSR-88D) 714
network and Canadian radar network, Parameter-elevation Regressions on Independent Slopes 715
Model (PRISM; Daly et al. 1994, 2017), Hydrometeorological Automated Data System (HADS) 716
gauge data with quality control (Qi et al., 2016), and outputs from numerical weather prediction 717
models. There are four different MRMS quantitative precipitation estimates (QPE) products 718
incorporating different input data or combinations: radar only, gauge only, gauge-adjusted radar, 719
and Mountain Mapper. For our study period (i.e., January 1–31, 2021), only the radar-only QPE 720
is currently available. 721
722
We acknowledge that precipitation data has uncertainties. Use of different precipitation products 723
may produce different results. A study comparing different gridded precipitation datasets including 724
satellite-based precipitation data, gauge dataset, and multi-sensor products revealed large 725
uncertainties in precipitation intensity (Bytheway et al., 2020). However, comparing different 726
precipitation datasets to characterize uncertainties is beyond the scope of this study. MRMS 727
provides gridded precipitation at high temporal (hourly) and spatial (1-km) resolutions, making it 728
a useful tool to demonstrate the utility of WRF-Hydro in post-wildfire debris flow hazard 729
assessments. 730
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
29
Appendix B 731
732
733
Fig. B1 Optical- and SAR-based remote sensing data of four debris flows. Optical data from 734
Sentinel-2 show pre- and post-debris flow imagery in real color. rdNDVI calculated from the 735
Sentinel-2 data show a decrease in vegetation corresponding to debris flow locations. Sentinel-1 736
backscatter change shows the change in ground surface properties determined by calculating the 737
log ratio of pre- and post-event SAR images. The pre-event, post-event satellite images, Sentinel-738
1 Backscatter, and Sentinel-2 rdNDVI change at (a) Rat Creek, (b) Mill Creek, (c) Big Creek, and 739
(d) Nacimiento. 740
741
742
743
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
30
744
Fig. B2 Schematic trapezoidal shape and related parameters of channels in WRF-Hydro. Bw is 745
the channel bottom width (m), z is the channel side slope (m), and H is water elevation (m). The 746
cross-sectional area of flow is calculated as (𝐵+ 𝐻 𝑧)𝐻. 747
748
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
31
Table B1 Parameters of trapezoidal channels in WRF-Hydro. 749
Stream order
Channel bottom
width
Bw (m)
Channel side slope
z (m)
Manning’s roughness
coefficient n
1
1.5
3
0.33
2
3
1
0.21
3
5
0.5
0.09
4
10
0.18
0.06
5
20
0.05
0.04
6
40
0.05
0.03
7
60
0.05
0.02
8
70
0.05
0.02
9
80
0.05
0.01
10
100
0.05
0.01
750
Table B1 Parameters of the trapezoidal channels in WRF-Hydro including channel bottom width 751
Bw (m), channel side slope z (m), and Manning’s roughness coefficient n. 752
753
754
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
32
755
Fig. B3 (a) Stream order defined by the USGS 30-m DEM in our WRF-Hydro model domain 756
and (b) the channel bottom width (m) which is a function of stream order (Table B1). 757
758
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
33
Table B2 759
MODIS IGBP 20-category land cover type and properties in Noah-MP LSM 760
Land
cover
code
Land cover type
Canopy
height
(m)
Max carboxylation
rate at 25°C
(
𝝁𝒎𝒐𝒍
𝑪𝑶
𝟐
/
(
𝒎
𝟐
∙
𝒔
)
)
Overland
flow
roughness
1 Evergreen Needleleaf Forest 20 50 0.2
2 Evergreen Broadleaf Forest 20 60 0.2
3 Deciduous Needleleaf Forest 18 60 0.2
4 Deciduous Broadleaf Forest 16 60 0.2
5 Mixed Forests 16 55 0.2
6 Closed Shrublands 1.1 40 0.055
7 Open Shrublands 1.1 40 0.055
8 Woody Savannas 13 40 0.055
9 Savannas 10 40 0.055
10 Grasslands 1 40 0.055
11 Permanent wetlands 5 50 0.07
12 Croplands 2 80 0.035
13 Urban and Built-Up 15 0 0.025
14 Cropland/natural vegetation
mosaic 1.5 60 0.035
15 Snow and Ice 0 0 0.01
16 Barren or Sparsely Vegetated 0 0 0.035
17 Water 0 0 0.005
18 Wooded Tundra 4 50 0.055
19 Mixed Tundra 2 50 0.055
20 Barren Tundra 0.5 50 0.055
761
Table B2 MODIS IGBP 20-category land cover type and properties in Noah-MP LSM. 762
763
764
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
34
Fig. B4 MODIS IGBP 20-category land cover type in the model domain. Red polylines are 2020 765
wildfire burn scar perimeters. 766
767
768
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
35
Fig. B5 1-km STATSGO data with 16 soil texture types. Red polylines are 2020 wildfire burn 769
scar perimeters. 770
771
772
773
774
775
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
36
Table B3 776
Default and calibrated soil parameters in WRF-Hydro 777
778
Soil type
Default After calibration
Grain size
distribution
index
Porosity
Saturated
hydraulic
conductivity
(m s-1)
Grain size
distribution
index
Porosity
Saturated
hydraulic
conductivity
(m s-1)
Sand 2.79 0.339 4.66E-5 2.51 0.315
1.5 x 10-7 m s-1
for all the burn
scars, and
original values
elsewhere.
Loamy sand 4.26 0.421 1.41E-5 3.83 0.392
Sandy loam 4.74 0.434 5.23E-6 4.27 0.404
Silt loam 5.33 0.476 2.81E-6 4.80 0.442
Silt 3.86 0.484 2.18E-6 3.47 0.450
Loam 5.25 0.439 3.38E-6 4.73 0.408
Sandy clay loam
6.77 0.404 4.45E-6 6.09 0.376
Silty clay loam 8.72 0.464 2.03E-6 7.85 0.432
Clay loam 8.17 0.465 2.45E-6 7.35 0.432
Sandy clay 10.73 0.406 7.22E-6 9.66 0.378
Silty clay 10.39 0.468 1.34E-6 9.35 0.435
Clay 11.55 0.468 9.74E-7 10.40 0.435
Organic material
5.25 0.439 3.38E-6 4.73 0.408
Water 0.00 1.00 0.00 0.00 1.00
Bedrock 2.79 0.200 1.41E-4 2.51 0.186
Other 4.26 0.421 1.41E-5 3.83 0.392
Playa 11.55 0.468 9.74E-7 10.40 0.435
Lava 2.79 0.200 1.41E-4 2.51 0.186
White sand 2.79 0.339 4.66E-5 2.51 0.315
779
Table B3 Soil parameters in default and calibrated WRF-Hydro. Default soil parameters in WRF-780
Hydro are adapted from the soil analysis by Cosby et al. (1984). Grain size distribution index and 781
soil porosity are altered from default values during the global soil moisture calibration. Saturated 782
hydraulic conductivity is altered from default values during the streamflow calibration. 783
784
785
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
37
786
787
Fig. B6 WRF-Hydro simulated discharge time-series at four debris flow source areas. (a)–(c) 788
MRMS precipitation (green bars) and simulated discharge time-series for January 26th 00:00 to 789
31st 23:00 at Mill Creek, Big Creek, and Nacimiento debris flow source areas (black circles in Fig. 790
7b–d) in baseline (purple dashed line) and burn scar simulation (red line). 791
792
793
794
795
796
797
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
38
Table B4 798
The total runoff volume, peak discharge, and peak timing at debris-flow source areas 799
Site name
Baseline simulation Burn scar simulation
Total
volume
(m3)
Peak
discharge
(m3 s-1)
Peak
timing
Total
volume
(m3)
Peak
discharge
(m3 s-1)
Peak
timing
Mill Creek 10,023 0.23 27th 23:00 83,853
(+737%)
1.24
(+439%) 27th 13:00
Big Creek 11,611 0.71 28th 05:00 128,879
(+1010%)
2.81
(+296%) 28th 05:00
Nacimiento 3,031 0.05 27th 13:00 49,792
(+1542%)
0.76
(+1420%) 27th 13:00
800
Table B4 The total runoff volume, peak discharge, and peak timing in the baseline and burn scar 801
simulations from January 27th 00:00 to 31st 23:00 at source areas of Rat Creek, Mill Creek, Big 802
Creek, and Nacimiento debris flows (black circles in Fig. 7b–d). The percent change of the total 803
volume and peak discharge in the burn scar simulation relative to the baseline simulation are shown 804
in parentheses. 805
806
807
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
39
808
809
Fig. B7 Discharge volume-based runoff-generated debris flow hazard at catchment level in the (a) 810
baseline simulation, (b) burn scar simulation, and (c) the difference between the burn scar and 811
baseline simulations. For each catchment, the hazard is assessed by computing the total discharge 812
volume at the catchment outlet from January 27th 00:00 to 28th 12:00. 813
814
815
Data availability statement 816
The NLDAS-2 reanalysis forcing data is publicly available at NASA GES DISC: 817
https://disc.gsfc.nasa.gov/datasets?keywords=NLDAS. A detailed description can be found at 818
https://ldas.gsfc.nasa.gov/nldas/v2/forcing. The MRMS radar-only precipitation estimate is 819
publicly available at: https://mtarchive.geol.iastate.edu/. A description can be found at 820
https://www.nssl.noaa.gov/projects/mrms/. The PSL in-situ soil moisture data is publicly available 821
at: https://psl.noaa.gov/data/obs/datadisplay/. The USGS streamflow is publicly available at: 822
https://waterdata.usgs.gov/nwis/. The remote sensing data used in this manuscript were provided 823
by the European Space Agency (ESA) Copernicus program and accessed on Google Earth Engine 824
(https://code.earthengine.google.com). All processed data required to reproduce the results of this 825
study are archived on Zenodo at http://doi.org/10.5281/zenodo.5544083. 826
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
40
Code availability statement 827
The modified WRF-Hydro Fortran code and instructions to output the overland flow at terrain 828
routing grid can be downloaded at https://github.com/NU-CCRG/Modified-WRF-Hydro. 829
HazMapper v1.0 is available at https://hazmapper.org/. The SAR backscatter change method code 830
is available at https://github.com/MongHanHuang/GEE_SAR_landslide_detection. 831
Author contribution 832
Conceptualization: CL, ALH, & DEH; Simulation and model analysis: CL; JW & WY model 833
methodological development. Remote sensing analysis: ALH; Field Observations: NJF; GIS 834
assistance: YX; Funding acquisition: GB & DH; CL wrote the original draft and all authors 835
reviewed and edited the manuscript. 836
Competing interests 837
The authors declare that they have no conflict of interest. 838
Acknowledgments 839
C.L., A.L.H., J.W., X.L., G.B., and D.E.H. acknowledge support from NSF PREEVENTS 840
#1848683. We acknowledge high-performance computing support from Cheyenne 841
(doi:10.5065/D6RX99HX) provided by NCAR's Computational and Information Systems 842
Laboratory, sponsored by the National Science Foundation. 843
844
845
References 846
Anderson, H. W., Hoover, M. D., & Reinhart, K. G. (1976). Forests and water: effects of forest management on 847
floods, sedimentation, and water supply (Vol. 18): Department of Agriculture, Forest Service, Pacific 848
Southwest Forest and Range Experiment Station, Berkeley, CA. 849
Andersson, J. C. M., Arheimer, B., Traoré, F., Gustafsson, D., & Ali, A. (2017). Process refinements improve a 850
hydrological model concept applied to the Niger River basin. Hydrological Processes, 31(25), 4540-851
4554. doi:https://doi.org/10.1002/hyp.11376 852
Arattano, M., & Franzi, L. (2010). On the application of kinematic models to simulate the diffusive processes of 853
debris flows. Nat. Hazards Earth Syst. Sci., 10(8), 1689-1695. doi:10.5194/nhess-10-1689-2010 854
Arattano, M., & Savage, W. Z. J. B. o. t. I. A. o. E. G.-B. d. l. A. I. d. G. d. l. I. (1994). Modelling debris flows 855
as kinematic waves. 49, 3-13. 856
Banihabib, M. E., Jurik, L., Kazemi, M. S., Soltani, J., & Tanhapour, M. (2020). A Hybrid Intelligence Model 857
for the Prediction of the Peak Flow of Debris Floods. Water, 12(8). doi:10.3390/w12082246 858
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
41
Bart, R. (2016). A regional estimate of postfire streamflow change in California. Water Resources Research, 52, 859
n/a-n/a. doi:10.1002/2014WR016553 860
Bart, R., & Hope, A. (2010). Streamflow response to fire in large catchments of a Mediterranean-climate region 861
using paired-catchment experiments. Journal of Hydrology, 388, 370-378. 862
doi:10.1016/j.jhydrol.2010.05.016 863
Bisson M, Favalli M, Fornaciai A, Mazzarini F, Isola I, Zanchetta G, Pareschi MT (2005) A rapid method to 864
assess fire-related debris flow hazard in the Mediterranean region: An example from Sicily (southern 865
Italy) International Journal of Applied Earth Observation and Geoinformation 7:217-231 866
doi:https://doi.org/10.1016/j.jag.2005.04.003 867
Bitew, M. M., & Gebremichael, M. (2011). Assessment of satellite rainfall products for streamflow simulation 868
in medium watersheds of the Ethiopian highlands. Hydrol. Earth Syst. Sci., 15(4), 1147-1155. 869
doi:10.5194/hess-15-1147-2011 870
Brown, E.K., Wang, J., & Feng, Y. (2020). U.S. wildfire potential: a historical view and future projection using 871
high-resolution climate data. Environmental Research Letters. 16, 034060 872
Bytheway, J. L., Hughes, M., Mahoney, K., & Cifelli, R. (2020). On the Uncertainty of High-Resolution Hourly 873
Quantitative Precipitation Estimates in California, Journal of Hydrometeorology, 21(5), 865-879. 874
Retrieved Oct 25, 2021, from https://journals.ametsoc.org/view/journals/hydr/21/5/jhm-d-19-875
0160.1.xml 876
Camera, C., Bruggeman, A., Zittis, G., Sofokleous, I., & Arnault, J. (2020). Simulation of extreme rainfall and 877
streamflow events in small Mediterranean watersheds with a one-way-coupled atmospheric-hydrologic 878
modelling system. Natural hazards and earth system sciences, 20(10), 2791-2810. doi:10.5194/nhess-879
20-2791-2020 880
Canfield, H., Goodrich, D., & Burns, I. S. (2005). Selection of Parameters Values to Model Postfire Runoff and 881
Sediment Transport at the Watershed Scale in Southwestern Forests. 882
Cannon, S., Gartner, J., C, P., & Parise, M. (2003). Widlfire-related debris flow generation through episodic 883
progressive sediment-bulking processes, western USA (Vol. 1). 884
Cannon, S., Gartner, J., Wilson, R., Bowers, J., & Laber, J. (2008). Storm Rainfall Conditions for Floods and 885
Debris Flows from Recently Burned Basins in Southwestern Colorado and Southern California. 886
Geomorphology, 96, 250-269. doi:10.1016/j.geomorph.2007.03.019 887
Cannon, S. H. (2001). Debris flow generation from recently burned watersheds. Environmental & Engineering 888
Geoscience, 7(4), 321-341. 889
Cannon, S. H., Boldt, E. M., Laber, J. L., Kean, J. W., & Staley, D. M. J. N. H. (2011). Rainfall intensity–890
duration thresholds for postfire debris flow emergency-response planning. 59(1), 209-236. 891
Cannon, S. H., & DeGraff, J. (2009). The increasing wildfire and postfire debris flow threat in western USA, 892
and implications for consequences of climate change. In Landslides–disaster risk reduction (pp. 177-893
190): Springer. 894
Cannon, S. H., Gartner, J. E., Rupert, M. G., Michael, J. A., Rea, A. H., & Parrett, C. (2010). Predicting the 895
probability and volume of postwildfire debris flows in the intermountain western United States. GSA 896
Bulletin, 122(1-2), 127-144. doi:10.1130/B26459.1 897
Cerdà, A. (1998). Changes in overland flow and infiltration after a rangeland fire in a Mediterranean scrubland. 898
Hydrological Processes, 12(7), 1031-1042. doi:https://doi.org/10.1002/(SICI)1099-899
1085(19980615)12:7<1031::AID-HYP636>3.0.CO;2-V 900
Chen, F., & Dudhia, J. (2001). Coupling an Advanced Land Surface–Hydrology Model with the Penn State–901
NCAR MM5 Modeling System. Part I: Model Implementation and Sensitivity, Monthly Weather 902
Review, 129(4), 569-585. Retrieved Oct 24, 2021, from 903
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
42
https://journals.ametsoc.org/view/journals/mwre/129/4/1520-904
0493_2001_129_0569_caalsh_2.0.co_2.xml 905
Chen, L., Berli, M., & Chief, K. (2013). Examining Modeling Approaches for the Rainfall-Runoff Process in 906
Wildfire-Affected Watersheds: Using San Dimas Experimental Forest. JAWRA Journal of the 907
American Water Resources Association, 49(4), 851-866. doi:https://doi.org/10.1111/jawr.12043 908
Reynolds, C., (2021, February 25). Highway 1 washout near Big Sur expected to be fixed by summer, Los 909
Angeles Times. https://www.latimes.com/travel/story/2021-02-25/highway-1-to-big-sur-will-reopen-910
by-summer-caltrans-says 911
Coe, J., Godt, J., Parise, M., & Moscariello, A. (2003). Estimating debris flow probability using fan stratigraphy, 912
historic records, and drainage-basin morphology, Interstate 70 highway corridor, central Colorado, USA. 913
Paper presented at the Debris flow Hazards Mitigation: Mechanics, Prediction, and Assessment, edited 914
by: Rickenmann, D. and Cheng, Ch., Proceedings 3rd International DFHM Conference, Davos, 915
Switzerland. 916
Coe, J. A., Kinner, D. A., & Godt, J. W. J. G. (2008). Initiation conditions for debris flows generated by runoff 917
at Chalk Cliffs, central Colorado. 96(3-4), 270-297. 918
Cosby, B. J., Hornberger, G. M., Clapp, R. B., & Ginn, T. R. (1984). A Statistical Exploration of the 919
Relationships of Soil Moisture Characteristics to the Physical Properties of Soils. Water Resources 920
Research, 20(6), 682-690. doi:https://doi.org/10.1029/WR020i006p00682 921
Cydzik, K., & Hogue, T. S. J. J. J. o. t. A. W. R. A. (2009). Modeling postfire response and recovery using the 922
hydrologic engineering center hydrologic modeling system (HEC‐HMS) 1. 45(3), 702-714. 923
Daly, C., R. P. Neilson, & D. L. Phillips. (1994). A statistical-topographic model for mapping climatological 924
precipitation over mountainous terrain. J. Appl. Meteor., 33, 140–158, https://doi.org/10.1175/1520-925
0450(1994)033<0140:ASTMFM>2.0.CO;2. 926
Daly, C., M. E. Slater, J. A. Roberti, S. H. Laseter, & L. W. Swift Jr.. (2017). High-resolution precipitation 927
mapping in a mountainous watershed; Ground truth for evaluating uncertainty in a national precipitation 928
dataset. Int. J. Climatol., 37, 124–137, https://doi.org/10.1002/joc.4986. 929
Davies, T., Phillips, C., Pearce, A., & Zhang, X. J. I. P. (1992). Debris flow behaviour–an integrated overview. 930
209(21), 225. 931
Dekker, L. W., & Ritsema, C. J. (1994). How water moves in a water repellent sandy soil: 1. Potential and actual 932
water repellency. Water Resources Research, 30(9), 2507-2517. 933
doi:https://doi.org/10.1029/94WR00749 934
Di Cristo, C., Iervolino, M., Moramarco, T., & Vacca, A. (2021). Applicability of Diffusive model for mud-935
flows: An unsteady analysis. Journal of Hydrology, 600, 126512. 936
doi:https://doi.org/10.1016/j.jhydrol.2021.126512 937
Doerr, S. H., & Thomas, A. (2000). The role of soil moisture in controlling water repellency: New evidence 938
from forest soils in Portugal. Journal of Hydrology, 231-232, 134-147. doi:10.1016/S0022-939
1694(00)00190-6 940
Friedel, M.J. (2011a) A data-driven approach for modeling post-fire debris-flow volumes and their uncertainty 941
Environmental Modelling & Software 26:1583-1598 doi:https://doi.org/10.1016/j.envsoft.2011.07.014 942
Friedel, M.J. (2011b) Modeling hydrologic and geomorphic hazards across post-fire landscapes using a self-943
organizing map approach Environmental Modelling & Software 26:1660-1674 944
doi:https://doi.org/10.1016/j.envsoft.2011.07.001 945
Gartner, J. E., Cannon, S. H., & Santi, P. M. (2014). Empirical models for predicting volumes of sediment 946
deposited by debris flows and sediment-laden floods in the transverse ranges of southern California. 947
Engineering Geology, 176, 45-56. doi:https://doi.org/10.1016/j.enggeo.2014.04.008 948
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
43
George, D. L., & Iverson, R. M. (2014). A depth-averaged debris flow model that includes the effects of evolving 949
dilatancy. II. Numerical predictions and experimental tests. Proceedings of the Royal Society A: 950
Mathematical, Physical and Engineering Sciences, 470(2170), 20130820. doi:10.1098/rspa.2013.0820 951
Gochis, D. J., & Chen, F. (2003). Hydrological Enhancements to the Community Noah Land Surface Model (No. 952
NCAR/TN-454+STR). University Corporation for Atmospheric Research. doi:10.5065/D60P0X00 953
Goodrich, D., Burns, I., Unkrich, C., Semmens, D., Guertin, D., Hernandez, M., . . . Levick, L. (2012). 954
KINEROS2/AGWA: model use, calibration, and validation. Transactions of the ASABE, 55, 1561-955
1574. doi:10.13031/2013.42264 956
Goss, M., Swain, D. L., Abatzoglou, J. T., Sarhadi, A., Kolden, C. A., Williams, A. P., & Diffenbaugh, N. S. 957
(2020). Climate change is increasing the likelihood of extreme autumn wildfire conditions across 958
California. Environmental Research Letters, 15(9), 094016. doi:10.1088/1748-9326/ab83a7 959
Gregoretti, C., & Fontana, G. D. (2008). The triggering of debris flow due to channel-bed failure in some alpine 960
headwater basins of the Dolomites: analyses of critical runoff. Hydrological Processes, 22(13), 2248-961
2263. doi:https://doi.org/10.1002/hyp.6821 962
Gupta, H. V., Kling, H., Yilmaz, K. K., & Martinez, G. F. (2009). Decomposition of the mean squared error and 963
NSE performance criteria: Implications for improving hydrological modelling. Journal of Hydrology, 964
377(1), 80-91. doi:https://doi.org/10.1016/j.jhydrol.2009.08.003 965
Handwerger, A. L., Jones, S. Y., Amatya, P., Kerner, H. R., Kirschbaum, D. B., and Huang, M.-H. (in review): 966
Strategies for landslide detection using open-access synthetic aperture radar backscatter change in 967
Google Earth Engine, Nat. Hazards Earth Syst. Sci. Discuss. 968
Huang, X., Swain, D. L., & Hall, A. D. (2020). Future precipitation increase from very high resolution ensemble 969
downscaling of extreme atmospheric river storms in California. 6(29), eaba1323. 970
doi:10.1126/sciadv.aba1323 %J Science Advances 971
Ice, G. G., Neary, D. G., & Adams, P. W. (2004). Effects of Wildfire on Soils and Watershed Processes. Journal 972
of Forestry, 102(6), 16-20. doi:10.1093/jof/102.6.16 973
Iverson, R. M., & George, D. L. (2014). A depth-averaged debris flow model that includes the effects of evolving 974
dilatancy. I. Physical basis. Proceedings of the Royal Society A: Mathematical, Physical and 975
Engineering Sciences, 470(2170), 20130819. doi:10.1098/rspa.2013.0819 976
Jolly, W. M., Cochrane, M. A., Freeborn, P. H., Holden, Z. A., Brown, T. J., Williamson, G. J., & Bowman, D. 977
M. J. S. (2015). Climate-induced variations in global wildfire danger from 1979 to 2013. Nature 978
Communications, 6(1), 7537. doi:10.1038/ncomms8537 979
Julien, P. Y., Saghafian, B., & Ogden, F. L. (1995). RASTER-BASED HYDROLOGIC MODELING OF 980
SPATIALLY-VARIED SURFACE RUNOFF1. JAWRA Journal of the American Water Resources 981
Association, 31(3), 523-536. doi:https://doi.org/10.1111/j.1752-1688.1995.tb04039.x 982
Kean, J. W., McCoy, S. W., Tucker, G. E., Staley, D. M., & Coe, J. A. (2013). Runoff-generated debris flows: 983
Observations and modeling of surge initiation, magnitude, and frequency. Journal of Geophysical 984
Research: Earth Surface, 118(4), 2190-2207. doi:https://doi.org/10.1002/jgrf.20148 985
Kean, J. W., & Staley, D. M. (2021). Forecasting the Frequency and Magnitude of Postfire Debris Flows Across 986
Southern California. Earth's Future, 9(3), e2020EF001735. doi:https://doi.org/10.1029/2020EF001735 987
Kean, J. W., Staley, D. M., & Cannon, S. H. (2011). In situ measurements of postfire debris flows in southern 988
California: Comparisons of the timing and magnitude of 24 debris flow events with rainfall and soil 989
moisture conditions. Journal of Geophysical Research F: Earth Surface, 116(4). 990
doi:10.1029/2011JF002005 991
Kinoshita, A. M., & Hogue, T. S. (2015). Increased dry season water yield in burned watersheds in Southern 992
California. Environmental Research Letters, 10(1), 014003. doi:10.1088/1748-9326/10/1/014003 993
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
44
Kling, H., Fuchs, M., & Paulin, M. (2012). Runoff conditions in the upper Danube basin under an ensemble of 994
climate change scenarios. Journal of Hydrology, 424-425, 264-277. 995
doi:https://doi.org/10.1016/j.jhydrol.2012.01.011 996
Larsen, I., MacDonald, L., Brown, E., Rough, D., Welsh, M., Pietraszek, J., . . . Schaffrath, K. (2009). Causes 997
of Postfire Runoff and Erosion: Water Repellency, Cover, or Soil Sealing? Soil Science Society of 998
America Journal - SSSAJ, 73. doi:10.2136/sssaj2007.0432 999
Lahmers, T. M., Gupta, H., Castro, C. L., Gochis, D. J., Yates, D., Dugger, A., . . . Hazenberg, P. J. J. o. H. 1000
(2019). Enhancing the structure of the WRF-Hydro hydrologic model for semiarid environments. 20(4), 1001
691-714. 1002
Lahmers, T. M., Castro, C. L., & Hazenberg, P. (2020). Effects of Lateral Flow on the Convective Environment 1003
in a Coupled Hydrometeorological Modeling System in a Semiarid Environment, Journal of 1004
Hydrometeorology, 21(4), 615-642. Retrieved Sep 29, 2021, from 1005
https://journals.ametsoc.org/view/journals/hydr/21/4/jhm-d-19-0100.1.xml 1006
MacDonald, L. H., & Huffman, E. L. (2004). Postfire soil water repellency: Persistence and soil moisture 1007
thresholds. Soil Science Society of America Journal, 68, 1729-1734. 1008
Martin, D., & Moody, J. (2001). Comparison of Soil Infiltration Rates in Burned and Unburned Mountainous 1009
Watersheds. Hydrological Processes, 15, 2893-2903. doi:10.1002/hyp.380 1010
McGuire, L. A., Rengers, F. K., Kean, J. W., & Staley, D. M. (2017). Debris flow initiation by runoff in a 1011
recently burned basin: Is grain-by-grain sediment bulking or en masse failure to blame? Geophysical 1012
Research Letters, 44(14), 7310-7319. doi:https://doi.org/10.1002/2017GL074243 1013
McGuire, L. A., & Youberg, A. M. (2020). What drives spatial variability in rainfall intensity-duration thresholds 1014
for post-wildfire debris flows? Insights from the 2018 Buzzard Fire, NM, USA. Landslides, 17(10), 1015
2385-2399. doi:10.1007/s10346-020-01470-y 1016
McMichael, C. E., & Hope, A. S. (2007). Predicting streamflow response to fire-induced landcover change: 1017
implications of parameter uncertainty in the MIKE SHE model. J Environ Manage, 84(3), 245-256. 1018
doi:10.1016/j.jenvman.2006.06.003 1019
Meixner, T., & Wohlgemuth, P. (2003). Climate Variability, Fire, Vegetation Recovery, and Watershed 1020
Hydrology. 1021
Meyer, G. A., & Wells, S. G. (1997). Fire-related sedimentation events on alluvial fans, Yellowstone National 1022
Park, U.S.A. Journal of Sedimentary Research, 67(5), 776-791. doi:10.1306/D426863A-2B26-11D7-1023
8648000102C1865D 1024
Mikoš, M., & Bezak, N. (2021). Debris Flow Modelling Using RAMMS Model in the Alpine Environment With 1025
Focus on the Model Parameters and Main Characteristics. 8(732). doi:10.3389/feart.2020.605061 1026
Miller, D. A., & White, R. A. (1998). A Conterminous United States Multilayer Soil Characteristics Dataset for 1027
Regional Climate and Hydrology Modeling, Earth Interactions, 2(2), 1-26. Retrieved Sep 27, 2021, from 1028
https://journals.ametsoc.org/view/journals/eint/2/2/1087-3562_1998_002_0001_acusms_2.3.co_2.xml 1029
Mitsopoulos, I. D., & Mironidis, D. (2006). Assessment of post fire debris flow potential in a Mediterranean 1030
type ecosystem. WIT Transactions on Ecology and the Environment, 90. 1031
Moody, J., & Ebel, B. (2012). Hyper-dry conditions provide new insights into the cause of extreme floods after 1032
wildfire. CATENA, 93, 58-63. doi:10.1016/j.catena.2012.01.006 1033
Moody, J., Shakesby, R., Robichaud, P., Cannon, S., & Martin, D. A. J. E.-S. R. (2013). Current research issues 1034
related to post-wildfire runoff and erosion processes. 122, 10-37. 1035
Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner, R. L., Harmel, R. D., & Veith, T. L. (2007). Model 1036
evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions 1037
of the ASABE, 50(3), 885-900. 1038
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
45
Nash, J. E., & Sutcliffe, J. V. (1970). River flow forecasting through conceptual models part I — A discussion 1039
of principles. Journal of Hydrology, 10(3), 282-290. doi:https://doi.org/10.1016/0022-1694(70)90255-1040
6 1041
Neary, D., Gottfried, G., & Ffolliott, P. (2003). Post-wildfire watershed flood responses. 1042
Nikolopoulos, E.I., Destro, E., Bhuiyan, M.A.E, Borga, M., & Anagnostou, E.N. (2018). Evaluation of predictive 1043
models for postfire debris flow occurrence in the western United States Nat Hazards Earth Syst Sci 1044
18:2331-2343 doi:10.5194/nhess-18-2331-2018 1045
Niu G-Y et al. (2011) The community Noah land surface model with multiparameterization options (Noah-MP): 1046
1. Model description and evaluation with local-scale measurements J Geophys Res Atm 116 1047
doi:https://doi.org/10.1029/2010JD015139 1048
Nyman, P., Smith, H. G., Sherwin, C. B., Langhans, C., Lane, P. N. J., & Sheridan, G. J. (2015). Predicting 1049
sediment delivery from debris flows after wildfire. Geomorphology, 250, 173-186. 1050
doi:https://doi.org/10.1016/j.geomorph.2015.08.023 1051
Oakley, N. S. (2021). A warming climate adds complexity to postfire hydrologic hazard planning. Earth's Future, 1052
9, e2021EF002149. https://doi.org/10.1029/2021EF002149 1053
Ogden, F. L. J. D. o. C., & Environmental Engineering, U. o. C., Storrs. (1997). CASC2D reference manual. 1054
Parise, M, Cannon, S.H. (2008) The effects of wildfires on erosion and debris-flow generation in Mediterranean 1055
climatic areas: a first database. Proceedings of 1st World Landslide Forum. Tokyo, Japan, pp 465–468 1056
Parise, M., & Cannon, S. H. (2009, April). A database on post-fire erosion rates and debris flows in 1057
Mediterranean-Basin watersheds. In EGU General Assembly Conference Abstracts (p. 1530). 1058
Parise, M., Cannon, S.H. (2012) Wildfire impacts on the processes that generate debris flows in burned 1059
watersheds Nat Hazards 61:217-227 doi:10.1007/s11069-011-9769-9 1060
Polade, S.D., Gershunov, A., Cayan, D.R, Dettinger, MD, & Pierce, D.W. (2017) Precipitation in a warming 1061
world: Assessing projected hydro-climate changes in California and other Mediterranean climate 1062
regions Scientific Reports 7:10783 doi:10.1038/s41598-017-11285-y 1063
Qi, Y., S. Martinaitis, J. Zhang, & S. Cocks. (2016). A real-time automated quality control of hourly rain gauge 1064
data based on multiple sensors in MRMS system. J. Hydrometeor., 17, 1675–1691, 1065
https://doi.org/10.1175/JHM-D-15-0188.1. 1066
Rengers, F. K., McGuire, L. A., Kean, J. W., Staley, D. M., & Hobley, D. E. J. (2016). Model simulations of 1067
flood and debris flow timing in steep catchments after wildfire. Water Resources Research, 52(8), 6041-1068
6061. doi:https://doi.org/10.1002/2015WR018176 1069
Robichaud, P., Beyers, J., & Neary, D. (2000). Evaluating the effectiveness of postfire rehabilitation treatments. 1070
Rosso R, Rulli MC, Bocchiola D (2007) Transient catchment hydrology after wildfires in a Mediterranean basin: 1071
runoff, sediment and woody debris Hydrol Earth Syst Sci 11:125-140 doi:10.5194/hess-11-125-2007 1072
Rulli, M. C., & Rosso, R. (2007). Hydrologic response of upland catchments to wildfires. Advances in Water 1073
Resources, 30(10), 2072-2086. doi:https://doi.org/10.1016/j.advwatres.2006.10.012 1074
Santi, P. M., deWolfe, V. G., Higgins, J. D., Cannon, S. H., & Gartner, J. E. (2008). Sources of debris flow 1075
material in burned areas. Geomorphology, 96(3-4), 310-321. doi:10.1016/j.geomorph.2007.02.022 1076
Santi, P. M., & Morandi, L. (2013). Comparison of debris flow volumes from burned and unburned areas. 1077
Landslides, 10(6), 757-769. 1078
Schaefli, B., & Gupta, H. V. (2007). Do Nash values have value? Hydrological Processes, 21(15), 2075-2080. 1079
https://doi.org/10.1002/hyp.6825 1080
Scheip, C. M., & Wegmann, K. W. (2021). HazMapper: A global open-source natural hazard mapping 1081
application in Google Earth Engine. Natural Hazards and Earth System Sciences, 21(5), 1495-1511. 1082
Schönfelder, L., Bakken, T., Alfredsen, K., & Adera, A. (2017). Application of HYPE in Norway. 1083
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
46
Scott, D. F., & Van Wyk, D. B. (1990). The effects of wildfire on soil wettability and hydrological behaviour of 1084
an afforested catchment. 121, 239. doi:10.1016/0022-1694(90)90234-o 1085
Scott, D. J. J. o. H. (1993). The hydrological effects of fire in South African mountain catchments. 150, 409-432. 1086
Shakesby, R. A., & Doerr, S. H. (2006). Wildfire as a hydrological and geomorphological agent. Earth-Science 1087
Reviews, 74(3), 269-307. doi:https://doi.org/10.1016/j.earscirev.2005.10.006 1088
Shen H, Lynch, B., Poulsen, C.J., & Yanites, B.J. (2021) A modeling framework (WRF-Landlab) for simulating 1089
orogen-scale climate-erosion coupling Computers & Geosciences 146:104625 1090
doi:https://doi.org/10.1016/j.cageo.2020.104625 1091
Sidman, G., Guertin, D., Goodrich, D., Unkrich, C., & Burns, I. (2016). Risk assessment of post-wildfire 1092
hydrological response in semiarid basins: the effects of varying rainfall representations in the 1093
KINEROS2/AGWA model. International Journal of Wildland Fire, 25, 268-278. doi:10.1071/WF14071 1094
Staley, D. M., Negri, J. A., Kean, J. W., Laber, J. L., Tillery, A. C., & Youberg, A. M. (2016). Updated logistic 1095
regression equations for the calculation of postfire debris flow likelihood in the western United States 1096
(2016-1106). Retrieved from Reston, VA: http://pubs.er.usgs.gov/publication/ofr20161106 1097
Stoof, C. R., Vervoort, R. W., Iwema, J., van den Elsen, E., Ferreira, A. J. D., & Ritsema, C. J. (2012). 1098
Hydrological response of a small catchment burned by experimental fire. Hydrol. Earth Syst. Sci., 16(2), 1099
267-285. doi:10.5194/hess-16-267-2012 1100
Swain, D. L. (2021). A Shorter, Sharper Rainy Season Amplifies California Wildfire Risk. Geophysical Research 1101
Letters, 48(5), e2021GL092843. doi:https://doi.org/10.1029/2021GL092843 1102
Swain, D. L., Langenbrunner, B., Neelin, J. D., & Hall, A. (2018) Increasing precipitation volatility in twenty-1103
first-century California Nature Climate Change 8:427-433 doi:10.1038/s41558-018-0140-y 1104
Tognacca, C., Bezzola, G., & Minor, H.-E. (2000). Threshold criterion for debris flow initiation due to channel-1105
bed failure. Paper presented at the Debris flow hazards mitigation: Mechanics, prediction and 1106
assessment. 1107
U.S. Forest Service, Burned Area Emergency Response, Dolan Postfire BAER Soil Burn Severity Map Released, 1108
October 10, 2020, https://inciweb.nwcg.gov/photos/CALPF/2020-10-06-1821-Dolan-PostFire-1109
BAER/related_files/pict20200910-143346-0.pdf 1110
U.S. Forest Service, Burned Area Emergency Response, Dolan Fire Burned Area Report, October 13, 2020, 1111
https://inciweb.nwcg.gov/photos/CALPF/2020-10-06-1821-Dolan-PostFire-1112
BAER/related_files/pict20200927-132609-0.pdf 1113
U.S. Geological Survey, Postfire Debris flow Hazards: Dolan Fire (Los Padres National Forest, CA), Landslide 1114
Hazards Program, retrieved September 27, 2021, from 1115
https://landslides.usgs.gov/hazards/postfire_debrisflow/detail.php?objectid=312 1116
Vergopolan, N., Chaney, N. W., Beck, H. E., Pan, M., Sheffield, J., Chan, S., & Wood, E. F. (2020). Combining 1117
hyper-resolution land surface modeling with SMAP brightness temperatures to obtain 30-m soil 1118
moisture estimates. Remote Sensing of Environment, 242, 111740. 1119
doi:https://doi.org/10.1016/j.rse.2020.111740 1120
Wang, J., Wang, C., Rao, V., Orr, A., Yan, E., & Kotamarthi, R. (2019). A parallel workflow implementation 1121
for PEST version 13.6 in high-performance computing for WRF-Hydro version 5.0: a case study over 1122
the midwestern United States. Geoscientific Model Development, 12(8), 3523-3539. doi:10.5194/gmd-1123
12-3523-2019 1124
Williams, A. P., Abatzoglou, J. T., Gershunov, A., Guzman-Morales, J., Bishop, D. A., Balch, J. K., & 1125
Lettenmaier, D. P. (2019). Observed Impacts of Anthropogenic Climate Change on Wildfire in 1126
California. Earth's Future, 7(8), 892-910. doi:https://doi.org/10.1029/2019EF001210 1127
Xia, Y., Mitchell, K., Ek, M., Cosgrove, B., Sheffield, J., Luo, L., . . . Lohmann, D. (2012). Continental-scale 1128
water and energy flux analysis and validation for North American Land Data Assimilation System 1129
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
47
project phase 2 (NLDAS-2): 2. Validation of model-simulated streamflow. Journal of Geophysical 1130
Research: Atmospheres, 117(D3). doi:https://doi.org/10.1029/2011JD016051 1131
Zhang, J. et al. (2011). National Mosaic and Multi-Sensor QPE (NMQ) system: Description, results, and future 1132
plans. Bull. Amer. Meteor. Soc., 92, 1321–1338, https://doi.org/10.1175/2011BAMS-D-11-00047.1. 1133
Zhang, J., Y. Qi, C. Langston, B., Kaney, & K. Howard. (2014). A real-time algorithm for merging radar QPEs 1134
with rain gauge observations and orographic precipitation climatology. J. Hydrometeor., 15, 1794–1809, 1135
https://doi.org/10.1175/JHM-D-13-0163.1. 1136
Zhang, J. et al. (2016). Multi-Radar Multi-Sensor (MRMS) quantitative precipitation estimation: Initial 1137
operating capabilities. Bull. Amer. Meteor. Soc., 97, 621–638, https://doi.org/10.1175/BAMS-D-14-1138
00174.1. 1139
1140
1141
https://doi.org/10.5194/nhess-2021-345
Preprint. Discussion started: 3 December 2021
c
Author(s) 2021. CC BY 4.0 License.
