AMERICAN METEOROLOGICAL SOCIETY Bulletin of the American Meteorological Society
EARLY ONLINE RELEASE This is a preliminary PDF of the author-produced manuscript that has been peer-reviewed and accepted for publication. Since it is being posted so soon after acceptance, it has not yet been copyedited, formatted, or processed by AMS Publications. This preliminary version of the manuscript may be downloaded, distributed, and cited, but please be aware that there will be visual differences and possibly some content differences between this version and the final published version. The DOI for this manuscript is doi: 10.1175/BAMS-D-15-00247.1 The final published version of this manuscript will replace the preliminary version at the above DOI once it is available. If you would like to cite this EOR in a separate work, please use the following full citation: Gourley, J., Z. Flamig, H. Vergara, P. Kirstetter, R. Clark III, E. Argyle, A. Arthur, S. Martinaitis, G. Terti, J. Erlingis, Y. Hong, and K. Howard, 2016: The Flooded Locations And Simulated Hydrographs (FLASH) project: improving the tools for flash flood monitoring and prediction across the United States. Bull. Amer. Meteor. Soc. doi:10.1175/BAMS-D-15-00247.1, in press. © 2016 American Meteorological Society
Manuscript (non-LaTeX)
Click here to download Manuscript (non-LaTeX) flash_bams_v2.docx
1 2
The Flooded Locations And Simulated Hydrographs (FLASH) project: improving
3
the tools for flash flood monitoring and prediction across the United States
4 5
Jonathan J. Gourley1, Zachary L. Flamig2,3,1, Humberto Vergara2,3,1, Pierre-Emmanuel
6
Kirstetter3,1, Robert A. Clark III2,3,1, Elizabeth Argyle2,3,1, Ami Arthur2,1, Steven
7
Martinaitis2,1, Galateia Terti2,1,4, Jessica M. Erlingis2,3,1, Yang Hong3, and Kenneth W.
8
Howard1
9 10
1 – NOAA/National Severe Storms Laboratory, Norman, OK
11
2 – Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma,
12
Norman, OK
13
3– Advanced Radar Research Center, University of Oklahoma, Norman, OK
14
4- Universite Grenoble, Grenoble, France
15 16
Revised Draft
17
Submitted as a full article to: Bulletin of the American Meteorological Society
18
6/29/2016
19 20
Address for Correspondence: Jonathan J. Gourley, National Weather Center, 120 David
21
L. Boren Blvd, Norman, OK 73072-7303
22
Tel: 405 325-6472
23
Email:
[email protected]
24
25
Abstract
26
This study introduces the Flooded Locations And Simulated Hydrographs (FLASH)
27
project. FLASH is the first system to generate a suite of hydrometeorological products at
28
flash flood scale in real-time across the conterminous United States including rainfall
29
average recurrence intervals, rainfall-to-flash flood guidance ratios, and distributed
30
hydrologic model-based discharge forecasts. The key aspects of the system are 1)
31
precipitation forcing from the NSSL’s Multi-Radar Multi-Sensor (MRMS) system, 2) a
32
computationally-efficient distributed hydrologic modeling framework with sufficient
33
representation of physical processes for flood prediction, 3) capability to provide
34
forecasts at all grid points covered by radars without the requirement of model
35
calibration, and 4) an open-access development platform, product display, and
36
verification system for testing new ideas in a real-time demonstration environment and
37
for fostering collaborations.
38 39
This study assesses the FLASH system’s ability to accurately simulate unit peak
40
discharges over a 7-year period in 1643 unregulated, gauged basins. The evaluation
41
indicates that FLASH’s unit peak discharges had a linear and rank correlation of 0.64 and
42
0.79, respectively, and the timing of the peak discharges has errors less than two hours.
43
The critical success index with FLASH was 0.38 for flood events that exceeded action
44
stage. FLASH performance is demonstrated and evaluated for case studies including the
45
2013 deadly flash flooding case in Oklahoma City as well as the 2015 event in Houston,
46
both of which occurred on Memorial Day weekends.
47
2
48
Capsule
49
The Flooded Locations And Simulated Hydrographs (FLASH) project is described and
50
evaluated. FLASH advances the state-of-the-science in operational flash flood monitoring
51
and prediction in the U.S. National Weather Service.
52
3
53
Body Text
54
Flash flooding remains a significant threat to those who live in the U.S. and
55
beyond. From 01 October 2007 to 01 October 2015, the NWS reported a total of 28,826
56
flash flooding events in the U.S. yielding an average of 3603 per year1. Ten percent of
57
these flash flooding events resulted in combined crop and property damages exceeding
58
$100,000 USD per event. A total of 278 individuals lost their lives due to flash floods in
59
the U.S. during this 8-yr period. Fatalities resulting from floods and flash floods show no
60
clear trend in recent decades. A brief point regarding the subdivision of floods into faster-
61
responding flash floods is required here as this segregation impacts some of the statistics
62
reported in the literature. While there is no real physical basis for separating floods and
63
flash floods, it is oftentimes necessary to divide them based on scale due to differing
64
operational responsibilities within agencies including the NWS. According to the NWS
65
Glossary (NWS 2012), flash floods are rapid rises of water in “a stream or creek above a
66
predetermined flood level, beginning within six hours of the causative event.” Flash
67
floods fall within the responsibility of local NWS Weather Forecast Offices (WFOs)
68
distributed throughout the U.S., while the 13 regional River Forecast Center (RFCs)
69
handle larger-scale river flood events. The tools and product displays utilized within the
70
WFOs differ from what is used for river flood warnings at the RFCs. The primary focus
71
hereafter is on flash floods, while some of the statistics reported below apply to larger-
72
scale river floods.
73
Špitalar et al. (2014) studied flash flood fatalities and injuries from 2006 to 2012
74
in the U.S. and revealed no apparent trend in either. An interesting result from this study 1
Source: National Weather Service Performance Management Team (http://verification.nws.noaa.gov). 4
75
was the finding that most human-impacting events occur in rural settings. However, when
76
a flash flood occurs in an urban center, there are many more human impacts per event.
77
Ashley and Ashley (2008) analyzed flood fatalities from 1959 to 2005; they found a
78
median value of flood fatalities at 81 yr-1 with no statistically significant trend. Several
79
studies cite the role of vehicles as a significant factor in the cause of death during flash
80
floods in the U.S., accounting for more than half of the fatalities (French et al. 1983;
81
Ashley and Ashley 2008; Kellar and Schmidlin 2012; Sharif et al. 2014; Terti et al.
82
2016). Despite the lack of increases in flash flood events or fatalities over this short time
83
period, they will likely increase in frequency and magnitude in coming decades. First, the
84
U.S. population continues to urbanize (United Nations 2006); urbanized basins respond
85
quickly to rainfall due to reduced infiltration and faster conveyance of water in
86
channelized canals. This urbanization process places an increasing number of individuals
87
and their vehicles in exposed environments (Montz and Gruntfest, 2002). Second, climate
88
change studies have projected an intensifying hydrologic cycle under future emission
89
scenarios resulting in more intense rainfall events and flash floods (Trenberth et al. 2003;
90
Milly et al. 2005, 2008). Advancements in the flash flood forecasting and monitoring are
91
needed in order to keep pace with or to reduce the trend of flash flood fatalities in the
92
U.S. and beyond.
93
Meteorological hazards such as lightning and tornadoes have resulted in fewer
94
and fewer fatalities in recent decades due to improvements in the public alerting and
95
detection systems, such as weather radar and mobile warnings (López and Holle, 1996;
5
96
Brooks and Doswell 2002) and wireless emergency alerts sent through mobile carriers2.
97
Flash flood forecasts, on the other hand, have not witnessed the same benefit from these
98
technological advances in observing and warning. Unlike severe weather and tornadoes,
99
they depend on a meteorological hazard (typically rainfall) compounded by a hydrologic
100
and human behavioral setting. There can be instances in which an intense rainfall event
101
yields no flash flooding in an unpopulated area with dry, sandy soils. Conversely,
102
seemingly innocuous rainfall amounts occurring over urban areas with impervious
103
surfaces can have devastating consequences, like rainfall upstream of a steep canyon with
104
hikers in it or rainfall near a low-water crossing at night. The present tools used by NWS
105
forecasters for forecasting flash floods do not explicitly consider these additional factors,
106
or only do so in a rudimentary way. These forecasters rely primarily on basin-specific
107
flash flood guidance (FFG) products displayed within the flash flood monitoring and
108
prediction system (FFMP) to provide warnings of impending flash floods (RFC
109
Development Management Team 2003). The research community needs to improve
110
precipitation estimates and short-term NWP forecasts, understand and represent the
111
underlying hydrological fluxes and storages, and incorporate human vulnerabilities in a
112
multi-disciplinary flash flood prediction system in order to accomplish the ultimate goal
113
of reducing impacts to lives and property.
114
The Flooded Locations and Simulated Hydrographs (FLASH) project
115
encompasses a suite of products to advance the state-of-the-science in flash flood
116
prediction. The FLASH project is an outgrowth from the Multi-Radar Multi-Sensor
117
(MRMS)
system
that
generates
a
suite
2
of
severe
Source: National Weather Service Weather-Ready Nation (http://www.nws.noaa.gov/com/weatherreadynation/wea.html). 6
weather,
aviation,
and
118
hydrometeorological products for the NWS across the conterminous U.S. (CONUS) and
119
southern Canada (Zhang et al. 2015). MRMS provides precipitation estimates by
120
mosaicking data from approximately 180 weather radars on a grid with a horizontal
121
spacing of one km updated every two min. The unprecedented spatiotemporal resolution
122
of the rainfall rates is essential for flash flood forecasting, especially in steep headwater
123
basins and in urban environments that respond to the causative rainfall on the order of
124
minutes. The FLASH system uses the MRMS rainfall rates to generate products that fall
125
into the following categories: 1) comparison of MRMS rainfall estimates to static
126
thresholds, 2) comparison of MRMS rainfall estimates to dynamic thresholds, and 3) as
127
forcing to a distributed hydrologic modeling framework.
128 129
Project description
130
Researchers and developers designed and optimized the FLASH system to
131
improve NWS forecasters’ ability to forecast flash flooding at WFOs. The FLASH
132
product suite is designed to be all-inclusive of the variable hydrometeorological
133
processes that lead to flash flooding across the CONUS. Additionally, in recognition of
134
historical NWS forecaster training and experience, components of the legacy FFG system
135
are included. FLASH encompasses the comparison of MRMS rainfall estimates to static
136
and dynamic thresholds as well as forecasts from distributed hydrologic models that use
137
MRMS rainfall as forcing. The FLASH system uses the same 1-km resolution grid
138
containing the MRMS rainfall estimates, yielding over 10.8 million grid points across the
139
domain. The update frequency of the products varies but can be as high as two min. An
140
underlying prerequisite of all FLASH products is the capability to resolve
7
141
hydrometeorological processes at or finer than the flash flood scale, as defined by the
142
NWS, so that they will be applicable to forecasting operations across the entire CONUS
143
and beyond3. All products described below are summarized in Table 1.
144
First, the FLASH system compares MRMS radar-only rainfall estimates to static
145
precipitation frequency values that have been published in NOAA Atlas 14 (Perica et al.
146
2013). The computations of precipitation frequency values use annual maximum rainfall
147
time series at rain gauge locations across the U.S. (with the exception of TX, WA, OR,
148
ID, MT, and WY). Fitting a distribution to the data enables the computation of rainfall
149
average recurrence intervals (ARIs), also referred to as return periods, for a given
150
duration of precipitation ranging from 5 min up to 60 days. The FLASH product
151
compares the real-time estimated MRMS radar-only accumulations (in mm) to the static
152
frequency values for durations from 30 min up to 24 h in order to compute the closest
153
ARI (in yr) at each grid point with updates every two min. Output products include
154
estimated rainfall ARIs corresponding to 30 min, 1, 3, 6, 12, and 24 h, and then the
155
maximum ARI for any duration. These products provide an estimate of the rarity of the
156
rainfall, which forecasters can use directly for flash flood forecasting or in conjunction
157
with other FLASH products that consider the underlying hydrology.
158
Clark et al. (2014) describe the history, evolution, and current state of the NWS
159
FFMP program across the U.S. Moreover, they provide a systematic evaluation of FFG
160
using observations of flash flooding. Several methods for producing FFG have been
161
developed at regional RFCs. The FLASH products rely on the standard FFG grids
162
produced by the RFCs issued every 6, 12, or 24 h, or as necessary during a flash flooding 3
Users can access real-time experimental and operational FLASH products at http://flash.ou.edu. 8
163
event. The NCEP Weather Prediction Center mosaics the most recently issued RFC grids
164
together to create a national grid on an hourly basis. The ratio product simply compares
165
the latest MRMS radar-only rainfall estimate to its collocated FFG value. The MRMS
166
rainfall grids have an update frequency of two min, which also applies to the ratio
167
product despite the relatively infrequent updates with the FFG issuance. Ratios greater
168
than 1.0 (or 100%) suggest that rainfall amounts have exceeded the threshold amount to
169
cause bank-full conditions on small streams.
170
In recent years, advances in high-performance massively parallel computing and
171
remote sensing of the Earth’s atmosphere, surface, and subsurface have led to the advent
172
of regional, continental and even global models that forecast Earth’s water and energy
173
cycles (Wood et al. 2011). The resolution of the forcing datasets and digital elevation
174
models motivates accompanying horizontal grid cell model resolutions on the order of
175
one km, while physical process representation through parameterization at sub-grid scale
176
requires careful consideration (Beven and Cloke 2012). Nonetheless, the European
177
Commission in partnership with the ECMWF has implemented the Global Flood
178
Awareness System that provides flood information on a daily time scale basis using
179
quantitative precipitation forecasts (Alfieri et al. 2013). Remote sensing of precipitation
180
by active and passive microwave instruments provides inputs to the Global Flood
181
Monitoring System operated by NASA and the University of Maryland (Wu et al. 2014).
182
The Ensemble Framework For Flash Flood Forecasting, or EF5, is a distributed
183
hydrologic modeling system that also applies globally; the implementation discussed
184
hereafter runs at continental scale provided the MRMS rainfall forcing.
9
185
EF5 is an open-source framework that encompasses the relevant processes for
186
flash flood modeling (Fig. 1); it is the hydrologic modeling engine central to the FLASH
187
project. The precipitation forcing normally comes from the MRMS rainfall estimates,
188
however the Hydrometeorological Testbed-Hydrologic (HMT) experiment (Martinaitis et
189
al. 2016) incorporated forcing from short-term, radar-based extrapolations (referred to as
190
“ADSTAT QPF” in Fig. 1) and quantitative precipitation forecasts supplied by the High
191
Resolution Rapid Refresh model (referred to as “HRRR QPF” in Fig. 1). The water
192
balance
193
potential evapotranspiration estimates (Koren et al. 1998). We prefer a simple handling of
194
evapotranspiration rates in lieu of implementing a sophisticated energy balance in a land
195
surface model because the modeling objective is rainfall-driven flash flood forecasting.
196
The model simulations need to provide forecasts out to 6 h within 10 min of clock time
197
and thus demand computational efficiency. Additional model physics and two-way
198
coupling would need to be incorporated for modeling systems designed to handle
199
seasonal water balance for water resources management, snowpack dynamics and
200
subsequent melting, coastal flooding exacerbated by storm surges, and surface water and
201
groundwater interactions in karstic aquifers. The WRF-Hydro modeling framework
202
(Gochis et al. 2014) is the hydrologic modeling core of NOAA’s new National Water
203
Model that is being designed to address these multiple hydrologic applications.
components
of
EF5
utilize
climatological
mean
monthly
204
The “ensemble” part of EF5 refers to the multiple water balance concepts
205
presently supported in the framework including the Sacramento soil moisture accounting
206
(SAC-SMA) model (Burnash et al. 1973), the Coupled Routing and Excess Storage
207
(CREST) model (Wang et al. 2011), and a “hydrophobic” model that prohibits water
10
208
from infiltrating the underlying soils. CREST differs from SAC-SMA in that it uses a
209
“percent imperviousness” parameter to describe the low infiltration rates over urban
210
areas. Water flows laterally in the unsaturated zone using linear reservoirs and on the
211
surface using either linear reservoirs (in the original CREST implementation) or through
212
the kinematic wave approximation to the 1-D Saint-Venant equation. The linear reservoir
213
approach is conceptually simpler but requires parameterization of the reservoir responses,
214
whereas the kinematic wave approximation is based on physical principles. The
215
kinematic wave routing scheme also requires parameters. Vergara et al. (2016) describe a
216
regionalization technique for estimating the routing parameters in the model channels,
217
which comprise approximately 25% of the total 10.8 million grid cells. First, they directly
218
estimate the parameters at USGS stream gauge locations and then regionalize them using
219
multi-dimensional statistical modeling guided by spatially distributed maps of variables
220
describing the geomorphology, hydroclimatology, and land use. Surface roughness and
221
slope dictate the routing parameters in the overland grid cells. This results in a-priori
222
routing parameters at all grid cells across the CONUS.
223
The EF5 products include soil saturation (in %), discharge (in m3 s-1), and
224
discharge normalized by the cell’s upstream drainage area (referred to as unit discharge,
225
in m3 s-1 km-2). The soil saturation state is the water content of the top-layer soils in
226
relation to their maximum water capacity. Soil saturation provides antecedent conditions
227
that can qualitatively inform forecasters on the potential hydrologic response to an
228
impending rain system. The unit discharge product is most directly applicable to flash
229
flood forecasting. The normalization of discharge by basin area helps to focus the
230
products on those specific locations that are most likely experiencing anomalous flows,
11
231
rather than merely identifying large discharges that occur regularly in major river systems
232
like the Mississippi River. Moreover, there is a basis for computing the observed unit
233
peak discharges for catastrophic flash floods throughout the world and establishing
234
“envelope curves”; these curves vary regionally but they provide the maximum expected
235
unit peak discharge for a given drainage area (Gaume et al. 2009). A hydrologic model
236
designed to forecast flash floods must therefore be able to accurately simulate unit peak
237
discharges.
238 239
Unit peak discharge evaluation
240
The EF5 products differ from the traditional rainfall-based flash flood guidance in
241
that they directly forecast the surface hydrologic conditions. Here we evaluate them in a
242
robust, statistically sound manner. A reanalysis project at NSSL produced a decadal
243
archive of MRMS radar-only precipitation rates at 1-km spatial resolution with 5-min
244
update frequency from the raw level-II NEXRAD radar data archive. These precipitation
245
rates resemble the real-time estimates, but do not get the benefits of dual-polarization
246
radar variables for improved quality control. Next, we identified 1643 USGS-gauged
247
basins that met the following criteria: 1) there must be no known upstream regulation or
248
diversion of river discharge, 2) at least 80% of the basin must fall within an area where
249
the MRMS radar beam height is 1 km AGL or less, 3) snowfall must contribute less than
250
30% of the annual precipitation, and 4) the contributing drainage area must be less than
251
1000 km2. The goal of the objective evaluation was to assess the EF5’s ability to simulate
252
flash floods independent of errors from factors such as regulation, snowmelt, and biased
12
253
MRMS inputs. This objective evaluation concentrates on EF5’s implementation of the
254
CREST water balance and the kinematic wave routing scheme.
255
After running EF5 forced by the MRMS radar-only rainfall dataset over the 1643
256
filtered basins between 01 March 2004 to 31 December 2011, we developed an
257
automated procedure to compare the simulated time series to the observed time series.
258
This is accomplished by examining the observed discharge time series record, identifying
259
individual events, extracting the peak discharges, computing the unit peak discharges,
260
and then comparing these to the equivalent simulated values from EF5. These event-
261
based characteristics comprise the publicly available unified flash flood database
262
described in Gourley et al. (2013). There were a total of 33,726 events in the analysis.
263
The basin areas range from 1.1 km2 to 999.7 km2 with a median of 115.8 km2, first
264
quartile of 37.8 km2 and third quartile of 305.6 km2. Figure 2a shows a density-colored
265
scatter plot of the simulated vs. observed unit peak discharges for the entire dataset. Peak
266
flows simulated by the CREST water balance module with kinematic wave routing
267
correlate well with observed discharges as indicated by a Pearson (linear) correlation of
268
0.64 and Spearman (rank) correlation of 0.79.
269
Peak timing errors decrease with smaller basin areas due to shorter concentration
270
times. We estimated the mean concentration time for each basin using an empirical
271
relationship for basin lag time developed by Mockus (1961), which is based on the basin
272
slope, stream length and the Natural Resources Conservation Service curve number. The
273
normalized peak timing error was computed as the peak timing error (in hours) divided
274
by the estimated mean concentration time (in hours) and is plotted as a function of basin
275
area in Fig. 2b. There is very slight tendency to simulate peak discharges too early. The
13
276
magnitude of the peak discharge is controlled more by the CREST water balance module
277
whereas the timing is dictated by the kinematic wave routing function. The a-priori
278
estimates of the kinematic wave parameters are data-driven and regionalized as
279
previously described whereas the CREST water balance parameters are related to
280
observable features of the land surface (e.g., soil types, land use/cover, topographic
281
derivatives).
282
The spatial variation of the errors in peak discharge simulation using the CREST
283
water balance model with kinematic wave routing is shown in Fig. 3. Underestimation in
284
the magnitude of peak discharge simulation is evident in Colorado with additional
285
variations in skill in the Intermountain West, whereas there are no regional biases in the
286
eastern two-thirds of the CONUS. The peak timing errors in Fig. 3b reveal no regional
287
dependencies. A vast majority of the event peaks are simulated with errors less than two
288
h. Next, we computed contingency table statistics for flood peaks that exceeded action
289
stage (typically related to bank-full stage) for a subset of the USGS basins that had
290
defined flood stages and met the aforementioned criteria for anthropogenic effects, radar
291
coverage, snowmelt contribution, and basin scale; these computations were computed for
292
all three water balance modules. Table 2 reveals the CREST water balance module has
293
the best overall skill in detecting flood peaks that exceed action stage. The critical
294
success index (CSI) with CREST is 0.38, which can be compared to the benchmark skill
295
established for the flash flood guidance tool used operationally in the NWS (Clark et al.
296
2014). Although Clark et al. (2014) used a different sample in terms of time period and
297
basins, the CSI with FFG was 0.20. The differences in CSI and Heidke skill score
298
amongst the different water balance components are not considerable, highlighting the
14
299
importance of accurate rainfall forcing and possibly routing. The MRMS precipitation
300
product has been comprehensively evaluated across the CONUS in Chen et al. (2013). A
301
relevant finding from that study was the dependence of the bias with the radar-only
302
precipitation product on the radar quality index product; underestimation was more
303
prevalent in regions with poor low-level radar coverage.
304
The performance of EF5 unit peak discharge and timing of peak discharge shown
305
in Figs. 2-3 and Table 2 is what can be expected in basins that are well covered by radars,
306
have no significant contribution from snowmelt, and are unregulated by dams and other
307
anthropogenic hydraulic structures. The following examples demonstrate the forecast
308
capability of the FLASH products for two well-observed urban flash flooding events.
309 310
Case studies
311
During the afternoon of 31 May 2013, a supercell thunderstorm produced a
312
tornado approximately 40 km (25 mi) west of downtown Oklahoma City, OK (OKC) in
313
El Reno, OK. This was only ten days following the devastating EF5 tornado that struck
314
Moore, OK, 20 km (12 mi) to the south of OKC, killing 24 people and causing $2 billion
315
in property damage. While public concern and media interest remained on yet another
316
tornado threatening the OKC area, the supercell thunderstorm took on more of an E-W
317
orientation and became quasi-stationary. This placed the core of the thunderstorm directly
318
over OKC for several hours. Storm total rainfall accumulations from MRMS approached
319
250 mm (10”) with the greatest accumulations directly over the OKC metro and a second
320
maximum closer to the tornado to the NW (see Fig. 4a). The City of Oklahoma City
321
Emergency Management provided very detailed flash flooding reports from their first
15
322
responders that included times and locations of high-water rescues, water in homes, street
323
closures, 13 fatalities, and damages estimated at $16 million4. This flash flooding event
324
was the deadliest in OKC’s history and the second deadliest statewide behind the 1984
325
Tulsa event, which also incidentally occurred on Memorial Day weekend.
326
The 3-h MRMS-to-FFG ratio product revealed rainfall estimates exceeding
327
critical rainfall thresholds to cause bank-full conditions for an expansive region including
328
the OKC metro (Fig. 4b). The greatest ratios were collocated with the highest ARI values
329
closer to the tornado but displaced to the NW from where most of the impacts occurred.
330
The 3-h rainfall ARI product valid from 2200 UTC 31 May 2013 to 1200 UTC 01 June
331
2013 shows the heaviest precipitation was occurring NW of the reported impacts in the
332
OKC metro (see blue-filled circles in Fig. 4c). The ARI product indicated a 0.5% annual
333
exceedance probability (200 yr) event for grid cells closer to the tornado to the NW, but
334
the ARIs near the reported flash flooding impacts were closer to 20% (5 yr). The same
335
MRMS rainfall forced the EF5 unit peak discharge product shown in Fig. 4d. Unit peak
336
discharges exceeded 10 m3 s-1 km-2 directly over OKC with the highest concentration of
337
impacts. The EF5 distributed hydrologic model product offers improvements over the
338
other rainfall-based FLASH products in this event because it considers low infiltration
339
rates over the OKC metropolitan area and routes the excess water downstream.
340
Moreover, it provides a 6-h forecast of the discharges resulting from the MRMS-
341
estimated rainfall inputs.
342
Following weeks of heavy rainfall, a slow moving storm system with a history of
343
producing flash flooding approached Houston, TX during the evening hours of 25 May 4
Twelve of the people who lost their lives were non-native English speakers and took shelter in storm drain structures (culverts) underground in fear of the tornado. 16
344
2015, again on Memorial Day5. The NWS in Houston/Galveston issued its first-ever flash
345
flood emergency for Harris County at 0452 UTC 26 May 2015. The maximum gauge-
346
measured rainfall accumulation occurred at Brays Bayou and Beltway 8 with 280 mm
347
(11”) in 12 hours and 254 mm (10”) in 6 hours (Lindner and Fitzgerald, 2015). City and
348
county emergency response units responded to hundreds of calls for rescues, mainly from
349
motorists trapped on flooded roadways. After the floodwaters receded, more than 1000
350
stranded vehicles littered the highways and were towed. Floodwaters impacted more than
351
1000 structures. The urban flash flood resulted in eight vehicle-related fatalities. Of these,
352
rescuers were able to reach three elderly victims by boat, but the rescue boat capsized in
353
the floodwaters causing them to lose their lives6.
354
The 24-h MRMS radar-only rainfall product for the event shows a small region
355
with 200-250 mm (8-10”) of rainfall over the Houston metropolitan region (Fig. 5a). The
356
precipitation frequency estimates from the NOAA Atlas 14 series do not yet exist for the
357
state of TX, so there are no FLASH ARI products available for this event. However, the
358
Harris County Flood Control District computed the annual exceedance probability of
359
rainfall in the 2-6 h accumulation period to be between 1% (100 yr) and 0.2% (500 yr)
360
over Brays and Buffalo Bayous (Lindner and Fitzgerald, 2015). The MRMS rainfall-to-
361
FFG maximum ratio product valid at 0400 UTC 26 May 2015, 52 minutes prior to the
362
NWS issuance of the flash flood emergency, indicates about 25% of Harris County
5
The Blanco River in nearby San Marcos, TX flooded from 330 mm (13”) of rain in its headwaters on 23-24 May resulting in 11 fatalities and damage to hundreds of homes. 6 The rescuers were unable to restart the gas-powered boat motor using the hand rope pull. A similar difficulty in restarting a gas motor on the rescue boat occurred with the July 1997 flash flood event in Ft. Collins, CO. 17
363
exceeded flash flooding thresholds (Fig. 5b)7. Not surprisingly, most of the largest ratios
364
are collocated with the heaviest rainfall in the western part of Harris County. There are
365
indications that the eastern parts of the county flooded as well according to the USGS
366
discharge observations, despite these areas having MRMS rainfall-to-FFG ratios less than
367
1.0.
368
The EF5 unit peak discharge product valid at 0400 UTC 26 May 2015
369
corresponds much more closely with the USGS flood observations (Fig. 5c). In
370
particular, values in the 2-6 m3 s-1 km-2 range extend in a SW-NE oriented band displaced
371
to the NW of the primary maximum, which aligns well with gauges reaching flood
372
conditions. Furthermore, these high unit discharge values extend into the eastern sections
373
of the county, which did not receive as much heavy rainfall but still flooded according to
374
the USGS observations. Values in the primary maximum exceeded 10 m3 s-1 km-2, which
375
are similar in magnitude to those that occurred with the OKC urban flash flood. The EF5
376
maximum discharge product highlights the major tributaries draining into the Gulf of
377
Mexico; these values are not necessarily rare or unusual (Fig. 5d). However, it reveals a
378
broad, contiguous swath of values in the 1-50 m3 s-1 range centered over Houston. This
379
indicates the CREST water balance model produced ponded conditions where the rainfall
380
rates exceeded the infiltration capabilities of the underlying urban surfaces.
381
The hydrographs of EF5-simulated discharge at Buffalo Bayou (USGS 08074000,
382
drainage area of 870 km2) indicate peak discharges exceeding major flood stage
383
according to all three water balance modules (Fig. 6). The simulated peaks overestimated
7
The plots presented in Figs. 4 were extracted from the online products archives at http://mrms.ou.edu and http://flash.ou.edu. The same websites also host the real-time product displays. 18
384
the observed peak discharge and reached maximum values approximately 2 h too early.
385
However, all EF5 model configurations forecasted a major flood, which falls well within
386
the modeling expectations given that the water balance parameters require no discharge-
387
based calibration and thus apply equally to ungauged grid points. The modeling objective
388
of EF5 in the FLASH context is focused directly on flood stage and timing simulation at
389
all grid points in the modeling domain.
390
envisaged with improved physical representation of baseflow conditions. In this
391
particular event, Fig. 6 reveals that the simulated baseflow conditions were too high and
392
improvements in these initial states would have yielded more accurate peak flow
393
simulations.
Improvements in peakflow simulation are
394 395
Summary and conclusions
396
The MRMS system provides precipitation rate estimates across the CONUS at 1-
397
km resolution with updates every two min. Now that MRMS has been transitioned to the
398
NWS for operational use, there is a great opportunity to capitalize on these rainfall rates
399
operating at flash flood scale for hydrologic applications. The FLASH system runs on the
400
back end of MRMS and yields a suite of rainfall and forecast stream discharge products
401
that are designed to advance operational flash flood monitoring and prediction to the
402
state-of-the-science. The primary products comprising FLASH are comparisons of
403
MRMS rainfall to static thresholds to compute average recurrence intervals, comparisons
404
of MRMS rainfall to dynamic flash flood guidance thresholds to compute rainfall-to-flash
405
flood guidance ratio, peak discharge, unit peak discharge, and soil saturation. A
406
distributed hydrologic modeling infrastructure called EF5 produces 0-6 h forecasts of the
19
407
discharge and soil saturation products using three different water balance components
408
and a kinematic wave routing scheme. All FLASH products reside on the same 1-km grid
409
as MRMS products with update frequencies of 2-10 min.
410
We evaluated unit peak discharges from EF5 for 33,726 events across the U.S.
411
and found that they correlate well with observed discharges (linear and rank correlation =
412
0.64 and 0.79, respectively). The water balance components of EF5 rely on parameters
413
based on physical properties of the Earth and are not adjusted using streamflow
414
observations. This means we can expect the EF5 maximum unit discharge forecasts to
415
have similar skill in ungauged basins, which comprises 99.9% of FLASH’s 10.8 million
416
grid cells. Errors in the timing of peak discharges predominantly fall within two h. The
417
urban flash flooding cases in Oklahoma City, OK and Houston, TX on Memorial Day
418
weekends of 2013 and 2015, respectively, showcased the FLASH product suite, while the
419
discharge-based objective evaluation focused on the EF5 distributed hydrologic model
420
outputs.
421
FLASH encompasses a variety of rainfall and forecast discharge products because
422
each of them provides valuable information about the magnitude and spatial extent of
423
flash flooding [see Martinaitis et al. (2016) for details]. We are using the ensemble of
424
products to develop flash flood recommenders available in the Hazard Services software
425
(Argyle et al. 2016). Guided by NWS forecaster inputs, we will optimize these
426
recommenders so that they take full advantage of the products that work best for their
427
areas of responsibility. NWS forecasters remain a critical component in testing new
428
concepts such as flash flood recommenders using the HMT-Hydro experiment, a vital
429
research-to-operations conduit and training mechanism.
20
430
The FLASH system provides new tools in the NWS forecaster toolbox for
431
warning on dangerous flash floods, but there are limitations. All products depend on the
432
accuracy of the MRMS radar-only rainfall rate estimates. The quality declines in the
433
Intermountain West where intervening mountains inhibit low-level radar coverage. Also,
434
the EF5 distributed hydrologic model products provide forecasts out to 6 h, but they are
435
normally based on observed rather than forecast rainfall. Forecasters should also note that
436
flash-flooding impacts can take place at scales that are typically unresolvable by
437
continental-scale models including EF5. Examples include flooding from clogged storm
438
drains, low-water crossings, and dam and levee breaches. The hydrologic modeling
439
concepts employed in EF5 do not necessarily apply to flood prediction in all
440
geomorphological settings. For instance, the kinematic wave routing scheme has limited
441
applicability in flat areas, including the low-lying Gulf Coast, Mississippi Delta, the
442
southern part of Florida, and the Central Valley of California [see Vergara et al. (2016)
443
for details]. Finally, we have designed the discharge simulations in EF5 very specifically
444
for flood detection, magnitude and timing estimation; the model outputs may not be
445
applicable to all facets of hydrologic forecasting such as baseflow conditions or flood
446
recessions.
447
The implementation of FLASH represents a first step toward modernizing flash
448
flooding products and services in the NWS. Our ultimate goal is to advance the state of
449
flash flood forecasting to realize quantifiable reductions in losses of life and property.
450
Several enhancements are underway. The MRMS precipitation forcing will improve with
451
more use of dual-polarization radar variables, probabilistic quantitative precipitation
452
estimates, and satellite-aided precipitation in regions with poor low-level coverage. We
21
453
will continue to evaluate the potential of using rainfall forcing from short-term radar-
454
based extrapolations and NWP forecasts. The hydrologic model forecasts will improve
455
through assimilation of surface water extent, depth, and velocity from spaceborne and
456
airborne radars, soil moisture from passive microwave radiometers, and stream discharge
457
from in-site gauges and remote-sensing platforms. Future enhancements to the model
458
physics include improved snow water equivalent estimation and snowpack modeling.
459
Finally, error models will be employed in order to identify and correct region-dependent
460
errors resulting from a combination of forcing errors and model physics limitations.
461
Being able to forecast overland and stream discharges at ungauged locations is a
462
necessary but insufficient condition to determine if there is an impending flash flood. We
463
are developing flood severity thresholds based on relationships between observed flood
464
stages and discharge, regionalized to ungauged grid cells using spatially distributed maps
465
of geomorphology, hydroclimatology, soil types, and land use/land cover. Further
466
refinements of these thresholds will occur by providing forecasts to local emergency
467
managers and receiving feedback about local “trouble spots”. EF5 discharge forecasts
468
will be compared to these thresholds to identify local flooding hazards. The next step is to
469
incorporate the hydrologic hazard forecasts with dynamic societal vulnerability factors in
470
order to anticipate specific impacts. For instance, a flash flood has a much greater
471
potential to cause a vehicle-related fatality when there is a flooded roadway at night
472
(Terti et al. 2016). We will use factors such as time of day, day of the week, road and
473
population density, proximity to a stream, etc., to produce impact-specific products.
474
Provided probabilistic NWP guidance and precipitation forcings [described in Kirstetter
22
475
et al. (2014)], the FLASH system will also yield probabilistic flash flood products in
476
coming years.
477
23
478
Acknowledgements
479
Funding for this research was provided by the Disaster Relief Appropriations Act of 2013
480
(P.L. 113-2), which provided support to the Cooperative Institute for Mesoscale
481
Meteorological Studies at the University of Oklahoma under research grant
482
NA14OAR4830100. The authors are greatly indebted to the leadership, advice, and
483
overall vision provided by Dr. Peter Lamb. He will be missed. The authors acknowledge
484
the constructive comments provided by two anonymous reviewers and Dr. Vincent
485
Fortin, Environment Canada, that led to an improved manuscript.
486
24
487
References
488 489
Alfieri, L., P. Burek, E. Dutra, B. Krzeminski, D. Muraro, J. Thielen, and F.
490
Pappenberger, 2013: GloFAS - global ensemble streamflow forecasting and flood
491
early warning, Hydrol. Earth Syst. Sci., 17, 1161–1175.
492 493
Argyle, E. M., J. J. Gourley, Z. L. Flamig, T. Hansen, and K. Manross, 2016: Towards a
494
user-centered design of a weather forecasting decision support tool, Bull. Amer.
495
Meteor. Soc., this issue.
496 497 498
Ashley, S., and W. Ashley, 2008: Flood fatalities in the United States. J. Appl. Meteor. Climatol., 47, 806–818.
499 500
Beven, K. J., and H. L. Cloke, 2012: Comment on ‘‘Hyperresolution global land surface
501
modeling: Meeting a grand challenge for monitoring Earth’s terrestrial water’’ by
502
Eric F. Wood et al., Water Resour. Res., 48, W01801, doi:10.1029/
503
2011WR010982.
504 505 506
Brooks, H. E., and C. A. Doswell, III, 2002: Deaths in the 3 May 1999 Oklahoma City tornado from a historical perspective. Wea. Forecasting, 17, 354–361.
507 508
Burnash, R. J. C., R. L. Ferral, and R. A. McGuire, 1973: A general streamflow
509
simulation system—Conceptual modeling for digital computers. Tech. Rep. to the
25
510
Joint Federal and State River Forecast Center, U.S. National Weather Service and
511
California Department of Water Resources, Sacramento, CA, 204 pp.
512 513
Chen, S., J. J. Gourley, Y. Hong, P. E. Kirstetter, J. Zhang, K. W. Howard, Z. L. Flamig,
514
J. Hu, and Y. Qi, 2013: Evaluation and uncertainty estimation of NOAA/NSSL
515
next-generation National Mosaic Quantitative Precipitation Estimation product
516
(Q2) over the Continental United States, J. Hydrometeor., 14, 1308-1322. doi:
517
10.1175/JHM-D-12-0150.1
518 519
Clark, R. A., J. J. Gourley, Z. L. Flamig, Y. Hong, and E. Clark, 2014: CONUS-wide
520
evaluation of National Weather Service flash flood guidance products, Wea.
521
Forecasting, 29, 377-392. doi:10.1175/WAF-D-12-00124.1.
522 523
Gaume, E., V. Bain, P. Bernardara, O. Newinger, M. Barbuc, A. Bateman, L.
524
Blaškovicˇová, G. Blöschl, M. Borga, A. Dumitrescu, I. Daliakopoulos, J. Garcia,
525
A. Irimescu, S. Kohnova, A. Koutroulis, L. Marchi, S. Matreata, V. Medina, E.
526
Preciso, D. Sempere-Torres, G. Stancalie, J. Szolgay, I. Tsanis, D. Velasco, and
527
A. Viglione, 2009: A compilation of data on European flash floods. J. Hydrol.,
528
367 (1–2), 70–78.
529 530
Gochis, D.J., W. Yu, D.N. Yates, 2014: The WRF-Hydro model technical description
531
and user’s guide, version 2.0. NCAR Technical Document. 120 pages. Available
532
online at: http://www.ral.ucar.edu/projects/wrf_hydro/.
26
533 534
Gourley, J. J., Y. Hong, Z. L. Flamig, A. Arthur, R. A. Clark, M. Calianno, I. Ruin, T.
535
Ortel, M. E. Wieczorek, E. Clark, P.-E. Kirstetter, and W. F. Krajewski, 2013: A
536
unified flash flood database across the United States, Bull. Amer. Meteor. Soc.,
537
94, 799-805. doi: http://dx.doi.org/10.1175/BAMS-D-12-00198.1
538 539 540
Kellar, D., and T. Schmidlin, 2012: Vehicle-related flood deaths in the United States, 1995–2005. J. Flood Risk Manage. 5, 153–163.
541 542
Kirstetter, P. E., J. J. Gourley, Y. Hong, J. Zhang, S. Moazamigoodarzi, C. Langston, and
543
A. Arthur, 2015: Probabilistic precipitation rate estimates with ground-based
544
radar networks. Wat. Resour. Res., 51, 1422-1442, doi:10.1002/2014WR015672.
545 546 547
Koren, V., J. Schaake, Q. Duan, M. Smith, and S. Cong, 1998: PET Upgrades to NWSRFS, Project Plan. In Unpublished Report.
548 549
Lindner, J., and S. Fitzgerald, 2015: Immediate Report – Final
May 25-27, 2015 Storm
550
and Flood Information [Memorandum]. Houston, TX: Harris County Flood
551
Control District.
552 553 554
López, R. E., and R. L. Holle, 1996: Fluctuations of lightning casualties in the United States: 1959–1990. J. Climate, 9, 608– 615.
555
27
556
Martinaitis, S. M., J. J. Gourley, Z. L. Flamig, E. M. Argyle, R. A. Clark, A. Arthur, B.
557
R. Smith, J. M. Erlingis, S. Perfater, B. Albright, 2016 : The HMT Multi-Radar
558
Multi-Sensor Hydro Experiment, Bull. Amer. Meteor. Soc., this issue.
559 560
Milly, P., J. Betancourt, M. Flakenmark, R. Hirsch, Z. Kundzewicz, D. Lettenmaier, and
561
R. Stouffer RJ, 2008 : Climate change—stationarity is dead: whither water
562
management? Science, 319, 573–574.
563 564 565
Milly, P., K. Dunne, and A. Vecchia, 2005: Global pattern trends in streamflow and water availability in a changing climate. Nature, 438, 347–350.
566 567 568
Mockus, V. 1961. Watershed lag. U.S. Dept. of Agriculture, Soil Conservation Service, ES–1015, Washington, DC.
569 570 571
Montz B., and E. Gruntfest, 2002: Flash flood mitigation: Recommendations for research and applications. Env. Hazards, 4, 15-22.
572 573 574
NWS, cited 2012: National Weather Service Glossary. [Available online at http://w1.weather.gov/glossary/index.php.]
575 576
Perica, S., D. Martin, S. Pavlovic, I. Roy, M. St. Laurent, C. Trypaluk, D. Unruh, M.
577
Yekta, and G. Bonnin, 2013: Precipitation-Frequency Atlas of the United States,
578
Volume 9, Version 2.0. US Department of Commerce, NOAA, NWS. Available
28
579
online
at
580
4_Volume9.pdf.
http://www.nws.noaa.gov/oh/hdsc/PF_documents/Atlas1
581 582
RFC Development Management Team, 2003: Flash Flood Guidance Improvement
583
Team—Final report. River Forecast Center Development Management Team Rep.
584
to the Operations Subcommittee of the NWS Corporate Board, 47 pp. [Available
585
online at http://www.nws.noaa.gov/oh/rfcdev/docs/ffgitreport.pdf.]
586 587 588
Sharif, H., T. Jackson, M. Hossain, and D. Zane, 2014: Analysis of flood fatalities in Texas. Nat. Haz. Rev., 16, 04014016.
589 590
Špitalar, M., J. J. Gourley, C. Lutoff, P. E. Kirstetter, M. Brilly, and N. Carr, 2014:
591
Analysis of flash flood parameters and human impacts in the US from 2006 to
592
2012. J. Hydrol., 519, 863-870. doi: 10.1016/j.jhydrol.2014.07.004.
593 594 595
Terti, G., I. Ruin, S. Anquetin, and J. Gourley, 2016: A situation-based analysis of flash flood fatalities in the United States, Bull. Amer. Meteor. Soc., this issue.
596 597 598
Trenberth, K. E., A. Dai, R. M. Rasmussen, and D. B. Parsons, 2003: The changing character of precipitation, Bull. Amer. Meteor. Soc., 84, 1205–1217.
599
29
600
United Nations, Department of Economic and Social Affairs, Population Division, 2010:
601
World Population Prospects: The 2006 Revision, vol. III: Analytical Report,
602
United Nations publication, No. E.07.XIII.4.
603 604
Vergara, H., P. Kirstetter, J. Gourley, Z. Flamig, Y. Hong, A. Arthur, and R. Kolar, 2016:
605
Estimating a-priori kinematic wave model parameters based on regionalization
606
for flash flood forecasting in the Conterminous United States. J. Hydrol. (in
607
press), doi:10.1016/j.jhydrol.2016.06.011.
608 609
Wang, J., Y. Hong, L. Li, J. J. Gourley, S. I. Khan, K. K. Yilmaz, R. F. Adler, F. S.
610
Policelli, S. Habib, D. Irwin, A. S. Limaye, T. Korme, and L. Okello, 2011: The
611
coupled routing and excess storage (CREST) distributed hydrological model.
612
Hydrol. Sci. Journal, 56, 84-98, doi: 10.1080/02626667.2010.543087
613 614
Wood, E. F., et al., 2011: Hyperresolution global land surface modeling: Meeting a grand
615
challenge for monitoring Earth’s terrestrial water, Water Resour. Res., 47,
616
W05301, doi:10.1029/2010WR010090.
617 618
Wu, H., R. F. Adler, Y. Tian, G. J. Huffman, H. Li, and J. Wang, 2014: Real-time global
619
flood estimation using satellite-based precipitation and a coupled land surface and
620
routing model, Water Resour. Res., 50, 2693-2717, doi:10.1002/2013WR014710.
621
30
622
Zhang, J., K. Howard, C. Langston, B. Kaney, Y. Qi, L. Tang, H. Grams, Y. Wang, S.
623
Cocks, S. Martinaitis, A. Arthur, K. Cooper, J. Brogden, and D. Kitzmiller, 2015:
624
Multi-Radar Multi-Sensor (MRMS) Quantitative Precipitation Estimation: Initial
625
Operating Capabilities. Bull. Amer. Meteor. Soc. doi:10.1175/BAMS-D-14-
626
00174.1, in press.
627 628 629
31
630 631 632
List of Tables
633
that correspond to the initial implementation (v11) of FLASH into the National Weather
634
Service. All products are generated on the same 0.01x0.01-degree grid as the Multi-Radar
635
Multi-Sensor products across the conterminous United States.
TABLE 1. Description of the FLASH product suite. The products listed here are the ones
636 637
TABLE 2. Statistics for the three water balance components supported in the Ensemble
638
Framework for Flash Flood Forecasting (EF5). The Pearson (linear) correlation and
639
Spearman (rank) correlation correspond to the observed and simulated peak flow values.
640
Contingency table statistics are reported based on the number of hits, misses, false
641
alarms, and correct negatives to compute the probability of detection (POD), false alarm
642
ratio (FAR), critical success index (CSI), and Heidke skill score (HSS).
643 644 645
32
646
TABLE 1. Description of the FLASH product suite. The products listed here are the ones
647
that correspond to the initial implementation (v11) of FLASH into the National Weather
648
Service. All products are generated on the same 0.01x0.01-degree grid as the Multi-Radar
649
Multi-Sensor products across the conterminous United States.
650 Product
Status
QPE average recurrence interval (1, 3, 6, 12, 24-h and max intervals) Radar-only QPE-to-flash flood guidance ratio (1, 3, 6-h and max ratios) CREST unit discharge CREST discharge CREST soil saturation 651 652
33
v11
Range of Values Update Frequency 0-200 yr 2 min
v11
0-500%
2 min
v11 v11 v11
0-20 m3 s-1 km-2 0-100,000 m3 s-1 0-100%
10 min 10 min 10 min
653
TABLE 2. Statistics for the three water balance components supported in the Ensemble
654
Framework for Flash Flood Forecasting (EF5). The Pearson (linear) correlation and
655
Spearman (rank) correlation correspond to the observed and simulated peak flow values.
656
Contingency table statistics are reported based on the number of hits, misses, false
657
alarms, and correct negatives to compute the probability of detection (POD), false alarm
658
ratio (FAR), critical success index (CSI), and Heidke skill score (HSS).
659 Water Balance Module CREST SAC-SMA Hydrophobic
Number of Events 12,771 18,934 14,573
Pearson Spearman Correlation Correlation 0.64 0.57 0.55
0.79 0.70 0.71
660 661 662
34
POD
FAR
CSI
HSS
0.54 0.49 0.93
0.43 0.48 0.67
0.38 0.34 0.32
0.41 0.37 0.37
663
List of Figures
664 665
FIG. 1. Flowchart describing the Ensemble Framework For Flash Flood Forecasting
666
(EF5) system configured for the Flooded Locations and Simulated Hydrographs
667
(FLASH) project. The red boxes indicate model forcings, blue boxes are model physics
668
modules, and the purple box contains output products. The years in parentheses
669
correspond to the date each component was incorporated in EF5.
670 671
FIG. 2. Density-colored scatter plot comparing (a) simulated peak unit discharge from the
672
CREST distributed hydrologic model with kinematic wave routing to USGS observations
673
at 1643 gauged basins for 33,726 events between 01 March 2004 to 31 December 2011.
674
The normalized peak timing error shown in (b) is computed as the peak timing error
675
divided by the estimated basin concentration time and is plotted as a function of drainage
676
area.
677 678
FIG. 3. (a) Spatial depiction of the peak magnitude error (in %) for the USGS gauged
679
basins used in the study; (b) spatial depiction of the peak timing error (in hours). The
680
gray-shaded regions correspond to areas that have radar coverage by the NEXRAD
681
network within 2 km of the ground.
35
682
FIG. 4. (a) Accumulated 24-hour rainfall estimates from the Multi Radar Multi Sensor
683
system ending at 12 UTC 01 June 2013; (b) maximum 3-hour rainfall-to-flash flood
684
guidance ratio from 22 UTC 31 May 2013 to 12 UTC 01 June 2013; (c) maximum 3-hour
685
average recurrence interval of rainfall for the same times as (b); (d) maximum unit
686
discharge forecasts from the distributed hydrologic model for the same times as in (b) and
687
(c). The blue dots correspond to known flooding reports collected from the City of
688
Oklahoma City, media, and social media. The reports include rescues, water in homes,
689
street closures, and 13 fatalities. Recent tornado tracks are shown in colors as indicated in
690
the legend.
691 692
FIG. 5. (a) Accumulated 24-hour rainfall estimates from the Multi-Radar Multi-Sensor
693
system ending at 17 UTC 26 May 2015; (b) rainfall-to-flash flood guidance ratio valid at
694
04 UTC 26 May 2015; (c) maximum unit discharge forecasts from the distributed
695
hydrologic model from 04 UTC 26 May 2015 to 10 UTC 26 May 2015; (d) maximum
696
discharge forecasts from the distributed hydrologic model for the same times as in (c).
697
The inverted triangles indicate USGS discharges that exceeded flood thresholds
698
(yellow=action, orange=minor, red=moderate, purple=major), while the brown circles
699
correspond to NWS local storm reports of flash flooding.
700
corresponds to the gauge at Brays Bayou and Beltway 8 with 280 mm (11”).
The filled green circle
701 702
FIG. 6. Hydrographs of simulated and observed discharge in Buffalo Bayou (USGS
703
08074000, drainage area of 870 km2) during the Houston flash flood event on Memorial
704
Day 2015. Action, minor, moderate, and major flood stages are shown as horizontal
36
705
dashed lines colored in yellow, orange, red, and dark red. The inset picture shows the
706
location of the stream gauge. All 3 distributed hydrologic model simulations correctly
707
forecast peak discharges to exceed major flood stage.
708 709
37
710 711
FIG. 1. Flowchart describing the Ensemble Framework For Flash Flood Forecasting
712
(EF5) system configured for the Flooded Locations and Simulated Hydrographs
713
(FLASH) project. The red boxes indicate model forcings, blue boxes are model physics
714
modules, and the purple box contains output products. The years in parentheses
715
correspond to the date each component was incorporated in EF5.
716
38
717 718
FIG. 2. Density-colored scatter plot comparing (a) simulated peak unit discharge from the
719
CREST distributed hydrologic model with kinematic wave routing to USGS observations
720
at 1643 gauged basins for 33,726 events between 01 March 2004 to 31 December 2011.
721
The normalized peak timing error shown in (b) is computed as the peak timing error
722
divided by the estimated basin concentration time and is plotted as a function of drainage
723
area.
724
39
725 726
FIG. 3. (a) Spatial depiction of the peak magnitude error (in %) for the USGS gauged
727
basins used in the study; (b) spatial depiction of the peak timing error (in hours). The
728
gray-shaded regions correspond to areas that have radar coverage by the NEXRAD
729
network within 2 km of the ground.
40
730 731
FIG. 4. (a) Accumulated 24-hour rainfall estimates from the Multi Radar Multi Sensor
732
system ending at 12 UTC 01 June 2013; (b) maximum 3-hour rainfall-to-flash flood
733
guidance ratio from 22 UTC 31 May 2013 to 12 UTC 01 June 2013; (c) maximum 3-hour
734
average recurrence interval of rainfall for the same times as (b); (d) maximum unit
735
discharge forecasts from the distributed hydrologic model for the same times as in (b) and
736
(c). The blue dots correspond to known flooding reports collected from the City of
737
Oklahoma City, media, and social media. The reports include rescues, water in homes,
738
street closures, and 13 fatalities. Recent tornado tracks are shown in colors as indicated in
739
the legend.
740
41
741 742
FIG. 5. (a) Accumulated 24-hour rainfall estimates from the Multi-Radar Multi-Sensor
743
system ending at 17 UTC 26 May 2015; (b) rainfall-to-flash flood guidance ratio valid at
744
04 UTC 26 May 2015; (c) maximum unit discharge forecasts from the distributed
745
hydrologic model from 04 UTC 26 May 2015 to 10 UTC 26 May 2015; (d) maximum
746
discharge forecasts from the distributed hydrologic model for the same times as in (c).
747
The inverted triangles indicate USGS discharges that exceeded flood thresholds
748
(yellow=action, orange=minor, red=moderate, purple=major), while the brown circles
749
correspond to NWS local storm reports of flash flooding.
750
corresponds to the gauge at Brays Bayou and Beltway 8 with 280 mm (11”).
751
42
The filled green circle
752
753 754
FIG. 6. Hydrographs of simulated and observed discharge in Buffalo Bayou (USGS
755
08074000, drainage area of 870 km2) during the Houston flash flood event on Memorial
756
Day 2015. Action, minor, moderate, and major flood stages are shown as horizontal
757
dashed lines colored in yellow, orange, red, and dark red. The inset picture shows the
758
location of the stream gauge. All 3 distributed hydrologic model simulations correctly
759
forecast peak discharges to exceed major flood stage.
760 761
43
