Abstract

The proximity to the Gulf of Mexico and local topography makes central Texas particularly prone to heavy precipitation and deadly flood events. Specifically, the Balcones Escarpment, located in central Texas, creates extremely favorable hydrologic characteristics for damaging floods. Urban centers such as San Antonio and Austin, Texas, are located along this terrain feature and have suffered at times, even with mitigation strategies, catastrophic flood damage. While the hydrologic effects of the Balcones Escarpment are well known, the meteorological impacts are uncertain. The purpose of this study is to evaluate the effect of the Balcones Escarpment in three cases of extreme precipitation in which the rainfall was maximized near the escarpment. Numerical simulations for each event were run at convection-allowing grid spacing using the Weather Research and Forecasting (WRF) Model and were used as control runs. Then, the Balcones Escarpment was removed by moving the associated terrain gradient to the north and west. The removal of the Balcones Escarpment did not change the overall characteristics of any of the three rainfall events, with the spatial pattern and magnitude of precipitation similar between the control and terrain-modified simulations. However, the location of the maximum precipitation was slightly, but consistently, shifted to the north and west. These results show that the overall atmospheric conditions are much more important for determining the intensity and occurrence of extreme rainfall in central Texas than the local topography, but the Balcones Escarpment can cause subtle hydrologically important changes in the location of the maximum accumulation.

1. Introduction

In many parts of the world, including the United States, flooding continues to pose a serious threat to human life, property, and infrastructure. The occurrence of floods and flash floods in any particular region are not solely dependent upon the rainfall amount or rate. The local- and large-scale topography, regional land use, and preceding soil moisture characteristics all play an important role in the evolution of flooding scenarios (e.g., Funk 2006; Collier 2007). Furthermore, storm motion characteristics play a central role in influencing the rainfall accumulation totals. The forecasting of floods is a complicated interplay between hydrologic and meteorological forcing, where not only the occurrence of an event, but also the magnitude is essential to evaluate potential impacts (e.g., Doswell et al. 1996).

During the time period from 1959 to 2005, Ashley and Ashley (2008) found that 4586 flood-related fatalities occurred in the continental United States with the most deadly events being flash flood or tropical cyclone related. The study also did not find any significant decrease in flood-related fatalities or risk over the 47-yr period, despite modern advancements in communication and flood mitigation. Compared to other weather-related threats over this time period, only flood-related fatalities failed to decrease even with the number of flood events per year staying relatively constant. This underscores the danger that floods, especially flash floods, still pose to people across the United States.

It has been well established that more fatalities and injuries due to flooding have occurred in Texas than any other state in the contiguous United States (CONUS) (e.g., Smith et al. 2000; Ashley and Ashley 2008; Sharif et al. 2010). Floods in Texas are characterized by frequent events that result in less than 20 deaths per occurrence with large individual events yielding more fatalities. In these events, over 50% of the fatalities are associated with flash floods, and 77% of these are motor vehicle related. Population-normalized annual flood-related death rates show a steady decline; however, from 1959 to 2008, Texas was the only state to record a flood-related fatality in each year (Sharif et al. 2015). This highlights the vulnerability of populations in Texas to flood-related dangers. Vulnerability can be thought of, in the case of flooding, as the potential loss of life and property due to the impact of a flood on a specific population with certain mitigation measures, or lack thereof, in place (Cutter 1996). In places such as Texas, the knowledge of this vulnerability has led to the implementation of flood mitigation strategies; however, because of increasing populations in flood-prone areas, loss of life and property from flooding events continue to be nonnegligible despite these measures (e.g., Pielke and Downton 2000; Burby 2001).

Within the state itself, a maximum of flood-related fatalities is located in central Texas, coincident with a topographic feature known as the Balcones Escarpment (Figs. 1a,c, as traced by identified cities) (Ashley and Ashley 2008; Sharif et al. 2015). The Balcones Escarpment is a region of steeply sloped terrain that separates the mainly limestone formations of the Edwards Plateau (i.e., the Texas Hill Country) from the flat clay- and sand-based coastal plain (Baker 1975). Several heavily populated urban areas, including San Antonio, New Braunfels, Austin, and Dallas, Texas, are all located along this terrain feature. Injury and death due to flash flooding occur more frequently in rural areas; however, if a flash flood does occur in an urban area, more people per event are injured or killed (Špitalar et al. 2014). Because of this increased risk, the region, which stretches in an arc from San Antonio to Dallas, is colloquially known as “Flash Flood Alley.” While there is no recognized definition or well-known origin of this term, it is used locally by meteorologists, newscasters, and civil officials (Sharif et al. 2015). The frequent use of this term highlights that the elevated flooding risk in central Texas is popular knowledge. Additionally, the increased flood danger has been formally discussed and evaluated within civil bodies such as National Weather Service (NWS) and local emergency management (e.g., NWS 1999; Martin and Edwards 1995). Furthermore, Stevenson and Schumacher (2014) showed a local clustering of rainfall events exceeding the 50-yr recurrence interval at various accumulation periods in central Texas over the 10 years from 2002 to 2011, with many of the events not limited to a single season.

Fig. 1.

(a),(b) Model terrain elevation plan view (m) and (c),(d) cross sections for the (a),(c) control simulations and (b),(d) the terrain-modification simulations over the entire ARW domain. Red line in (a),(b) represents the location of cross section in (c),(d). Location labels correspond to cities (specifically airports) located on or near the Balcones Escarpment to identify its approximate location in central Texas (KDRT is Del Rio, KSAT is San Antonio, KAUS is Austin, KTPL is Temple, and KDFW is Dallas–Fort Worth).

Fig. 1.

(a),(b) Model terrain elevation plan view (m) and (c),(d) cross sections for the (a),(c) control simulations and (b),(d) the terrain-modification simulations over the entire ARW domain. Red line in (a),(b) represents the location of cross section in (c),(d). Location labels correspond to cities (specifically airports) located on or near the Balcones Escarpment to identify its approximate location in central Texas (KDRT is Del Rio, KSAT is San Antonio, KAUS is Austin, KTPL is Temple, and KDFW is Dallas–Fort Worth).

The effect of the Balcones Escarpment on flooding can be broken into hydrologic and meteorological forcing factors. The hydrologic effects are static and better understood than the meteorological forcing. Steep limestone slopes with narrow river valleys and little vegetation characterize the transition from the Edwards Plateau to the coastal plain along the Balcones Escarpment. This limestone bedrock, combined with urbanization, increases rainfall–runoff and stream discharge (e.g., Baker 1975; Caran and Baker 1986). Consequently, measured stream discharges in this region typically exceed those observed for similar sized catchments in the rest of the United States. The narrow stream channels, combined with large flow discharges, lead to great flow depths, in some cases up to 15 m, and extensive flooding (Patton and Baker 1976; Baker 1977; Costa 1987). Furthermore, floods in this region reach peak stream discharge much closer to the time of maximum precipitation, compared to the rest of the Texas coastal plain (Smith et al. 2000). All of these factors lead to the Flash Flood Alley region along the Balcones Escarpment being more vulnerable to flash floods than any other region in the continental United States based on hydrologic rainfall–runoff characteristics alone (O’Connor and Costa 2004).

A significant amount of research investigates the effects or orography on precipitation initiation and maintenance throughout the United States and elsewhere, especially in Europe. Orography has been shown to initiate convection if instability is present, without any required synoptic to mesoscale forcing due to upslope mechanical lift. Furthermore, the topographic feature can serve as an elevated heat source that can further enhance upslope flow (e.g., Blumen 1990). The convective boundary layer over topography extends higher into the atmospheric column and can serve to reduce convective inhibition. Additionally, strong boundary layer ascent associated with topographic features can further increase the probability for convection initiation (e.g., Banta 1984; Kirshbaum 2013). Topographic features have also been found to maintain quasi-stationary convective systems that enhance the threat for flash flooding (e.g., Doswell et al. 1996). Sustained, convectively unstable low-level flow perpendicular to the terrain feature combined with weak upper-level flow scenarios have been found to cause quasi-stationary convection in significant flooding cases, such as the Rapid City flood of 1972, Big Thompson Canyon flood of 1976 (Maddox et al. 1978), Madison County flood of 1995 (Pontrelli et al. 1999), and the Fort Collins flood of 1997 (Petersen et al. 1999). In similar situations where strong flow confined to the lower levels impinges on a barrier, the upstream cold pool propagation can become balanced by the lower-level flow, which can further aide in quasi-stationary convective maintenance (e.g., Ducrocq et al. 2008). Both of these mechanisms can maintain quasi-stationary systems as long as the low-level moisture inflow is persistent, for example, from the presence of a moist low-level jet (LLJ). It, however, has also been shown that weak vertical shear cases with along-barrier winds can also produce quasi-stationary convective events (e.g., Soderholm et al. 2014). Many case studies of the orographic effects of specific topographic features have been carried out in many different regions around the globe. Some of these features include the Massif Central and other features in the Mediterranean (e.g., Ducrocq et al. 2008; Bresson et al. 2012), the Black Hills (Soderholm et al. 2014), the Rhine River valley (Weckwerth et al. 2014), the Blue Ridge in Virginia (Pontrelli et al. 1999), and southwestern England (Lean et al. 2009). Most of these features, except the last, deal with topographic features that have a larger vertical extent and gradients than the Balcones Escarpment; however, this does not necessarily mean similar mechanisms do not apply.

The meteorological influence of the Balcones Escarpment on extreme precipitation in central Texas is not completely understood. Orographic ascent along the escarpment has been hypothesized to be an important forcing mechanism for the extreme rainfall that is observed in central Texas (e.g., Baker 1975; Caran and Baker 1986). The abrupt elevation rise is believed to help initiate convection when warm moist air from the Gulf of Mexico ascends the Balcones Escarpment. Caracena and Fritsch (1983) suggested the escarpment could serve to preserve air parcel saturation through orographic lift and potentially stall a northward-moving airmass boundary. However, in the cases discussed, Caracena and Fritsch (1983) were not able to discern the importance of the Balcones Escarpment in the precipitation processes relative to other meteorological factors. Additionally, Nielsen-Gammon et al. (2005) examined in detail a flooding event that occurred in central Texas in 2002 along the Balcones Escarpment. They found that the main forcing for the extreme rainfall accumulations was from a stationary upper-level trough that remained in place for almost a week due to complex interactions of many meteorological contributions. However, smaller-scale processes, including mechanical lifting along the Balcones Escarpment, were thought to serve to localize heavy precipitation along the terrain feature. The importance of the Balcones Escarpment on the location and intensity of the observed rainfall is not well known and has been identified as a topic warranting more investigation through the aforementioned and other studies (e.g., Smith et al. 2000).

This study’s goal is to determine the influence of the Balcones Escarpment on three recent flooding events that occurred in central Texas. In each of these three cases, radar imagery indicated regions of repeated convective development near the Balcones Escarpment and the largest precipitation accumulations occurred near the escarpment. This led the authors to suspect that orographic effects were important in enhancing or focusing the precipitation in these extreme rainfall events. Each case is examined using simulations from a numerical weather prediction (NWP) atmospheric model. A control run representing the model’s interpretation of the event is compared to a terrain-modified run in which the terrain gradient associated with the Balcones Escarpment will be altered. The focus of the results is on precipitation differences between the control and terrain-modified runs. Section 2 provides a synoptic overview and reported observations for each event, while section 3 provides a summary of the NWP model configuration. The results of the experiments, discussion of the results, and an overall summary are presented in sections 4, 5, and 6, respectively.

2. Description of rainfall events

Events were chosen based upon recent flooding that occurred on the Balcones Escarpment in central Texas. Two of the events were associated with mesoscale convective vortices (MCVs)—one displaying unusually slow movement—while the third formed along a prefrontal trough. These represent large-scale environments typical of strongly forced extreme-rain-producing convective systems in this part of the country. A concise meteorological overview of each rainfall event is presented in this section and is supplemented by synoptic maps valid during the period of most intense rainfall taken from the North American Regional Reanalysis (NARR; Mesinger et al. 2006; NCEP/ESRL 2015). Additionally, an observed sounding from Corpus Christi, Texas (CRP), and the National Centers for Environmental Prediction (NCEP) stage-IV gridded precipitation analysis (Lin and Mitchell 2005) for each event are presented. The precipitation accumulation of the control run is evaluated by comparison to the NCEP stage-IV gridded precipitation analysis. The NCEP stage IV uses a multisensor approach that includes both rain gauge and radar accumulation data to create the gridded analysis (Lin and Mitchell 2005). This precipitation analysis was chosen because of the high spatial (4 km) and temporal (hourly) coverage of the product, which would not be possible from a reanalysis (e.g., the NARR). While issues with the accuracy of stage-IV analysis arise in complex terrain and regions with sparse gauge coverage, central Texas is heavily populated with many gauges and contains relatively simple terrain (when compared to the Rocky Mountains, for instance) that causes little radar interference and beam blockage, which reduces these possible inaccuracies. Although as with any gridded precipitation analysis, there are uncertainties in the exact quantitative precipitation estimate; however, the stage-IV analysis should create a representative depiction of the accumulated precipitation in these specific cases.

a. 9 June 2010

The intense rainfall during the early morning hours on 9 June 2010 was associated with an MCV that migrated through Texas (Fig. 2a) and into Arkansas over an unusually long period of time, from 3 to 11 June 2010. This event has been previously covered in great detail by Schumacher et al. (2013) and Schumacher and Clark (2014). Isolated rainfall associated with the northward-moving MCV and pronounced warm-air advection (Fig. 2b) initiated to the northeast of the San Antonio (KSAT) area around 0000 UTC 9 June. Heavy rain continued to fall, fed by moist inflow, at both the surface and 850 hPa, from the Gulf of Mexico (Figs. 2c,d) with small storm motions in and around central Texas until 1800 UTC on the same day (Fig. 3a), at which point the MCV began moving out of the region to the northeast. Precipitable water values from the 1200 UTC CRP sounding (Fig. 2d) were above the 90th climatological percentile for that day (NOAA 2015b). A maximum rainfall amount of 287 mm (11.3 in.) for the event was observed by multiple stations to the northwest of New Braunfels, Texas, by the Community Collaborative Rain, Hail, and Snow Network (CoCoRaHs; Cifelli et al. 2005). The flash flooding caused property damage totaling $10 million and one fatality, with the heaviest damage in the Guadalupe River basin. The MCV associated with this event went on to produce a deadly flash flood in Arkansas that resulted in 20 fatalities (NOAA 2015a).

Fig. 2.

(a)–(c) NARR analyses at 0900 UTC 9 Jun 2010, and (d) skew T–logp diagram showing sounding from CRP at 1200 UTC 9 Jun 2010. (a) Absolute vorticity at 500 hPa (×10−5 s−1), shaded every 3 × 10−5 s−1 above −9 × 10−5 s−1; 500-hPa geopotential height (contoured every 60 m); and 500-hPa wind barbs (half barb = 5, full barb = 10, pennant = 50 kt, 1 kt = 0.5144 m s−1). (b) 850-hPa geopotential height (contoured every 25 m), 850-hPa wind barbs, and 850-hPa temperature (shaded every 5° from −20° to 35°C). (c) Precipitable water (shaded contours every 5 mm for values from 10 to 50 mm), 10-m wind barbs, and mean sea level pressure (MSLP) (contoured every 3 hPa). Red dashed line in (d) shows the environmental virtual temperature curve, and dashed black line in (d) shows the virtual temperature of a lifted parcel with the mean characteristics of the lowest 500 m.

Fig. 2.

(a)–(c) NARR analyses at 0900 UTC 9 Jun 2010, and (d) skew T–logp diagram showing sounding from CRP at 1200 UTC 9 Jun 2010. (a) Absolute vorticity at 500 hPa (×10−5 s−1), shaded every 3 × 10−5 s−1 above −9 × 10−5 s−1; 500-hPa geopotential height (contoured every 60 m); and 500-hPa wind barbs (half barb = 5, full barb = 10, pennant = 50 kt, 1 kt = 0.5144 m s−1). (b) 850-hPa geopotential height (contoured every 25 m), 850-hPa wind barbs, and 850-hPa temperature (shaded every 5° from −20° to 35°C). (c) Precipitable water (shaded contours every 5 mm for values from 10 to 50 mm), 10-m wind barbs, and mean sea level pressure (MSLP) (contoured every 3 hPa). Red dashed line in (d) shows the environmental virtual temperature curve, and dashed black line in (d) shows the virtual temperature of a lifted parcel with the mean characteristics of the lowest 500 m.

Fig. 3.

NCEP stage-IV precipitation analysis accumulation for each flooding event valid (a) the 18-h period ending 1800 UTC 9 Jun 2010, (b) the 12-h period ending 1200 UTC 31 Oct 2013, and (c) the 12-h period ending 1800 UTC 25 May 2013. Location labels refer to the same cities as described in Fig. 1 with the addition of KUVA.

Fig. 3.

NCEP stage-IV precipitation analysis accumulation for each flooding event valid (a) the 18-h period ending 1800 UTC 9 Jun 2010, (b) the 12-h period ending 1200 UTC 31 Oct 2013, and (c) the 12-h period ending 1800 UTC 25 May 2013. Location labels refer to the same cities as described in Fig. 1 with the addition of KUVA.

b. 31 October 2013

An upper-level trough centered over northeastern New Mexico (Fig. 4a) was responsible for widespread intense rainfall that occurred in south-central and eastern Texas on 31 October 2013. A developing surface front, combined with a prefrontal lower-level trough and warm-air advection (Figs. 4b,c), provided sustained lift over the period from 0000 to 1800 UTC that same day. Precipitable water values (Figs. 4c,d) above the 90th percentile for the daily average contributed to the strength and intensity of the rainfall (NOAA 2015b) . Furthermore, storm motions resulted in continued “echo training” of intense rainfall over the same regions. Unlike the previously discussed event, the rainfall on 31 October occurred in an environment with strong synoptic-scale forcing for ascent. The largest rainfall totals were observed in a southwest-to-northeast line stretching from north of KSAT through Austin (KAUS) to Temple, Texas (Fig. 3b). The heavy rain that fell in this corridor resulted from convection that repeatedly formed along and ahead of the surface front, and also along the eastern edge of the Balcones Escarpment. Within this belt of large rainfall accumulations, a maximum CoCoRaHS observation of 305 mm (12.0 in.) was recorded. The most intense flash flooding occurred in watersheds near the Austin area, where 100 million dollars’ worth of damage occurred and over 100 homes were destroyed. Four fatalities were reported in the KAUS area due to rapidly rising floodwaters (NOAA 2015a).

Fig. 4.

As in Fig. 2, but (a)–(c) NARR analyses at 0600 UTC 31 Oct 2013, and (d) skew T–logp diagram showing sounding from CRP at 0000 UTC 31 Oct 2013.

Fig. 4.

As in Fig. 2, but (a)–(c) NARR analyses at 0600 UTC 31 Oct 2013, and (d) skew T–logp diagram showing sounding from CRP at 0000 UTC 31 Oct 2013.

c. 25 May 2013

On 25 May 2013, convection associated with a quasi-stationary, preexisting MCV (Fig. 5a) produced large amounts of precipitation in the KSAT metro area and other nearby regions of south-central Texas (Fig. 3c). The MCV was formed as a result of convection that developed in the Texas Panhandle on 23 May 2013. Unlike the June 2010 case, the MCV moved into the KSAT area from a more northerly direction, compared to the 9 June 2010 event. Rain began to fall around 0600 UTC 25 May and continued until 1800 UTC, at which point convection ceased at the center of the system. During this period, the MCV stayed relatively stationary with warm moist flow from the southeast providing the moisture supply (Figs. 5b,c). This southeasterly moist flow off the Gulf of Mexico extended over a deep layer, from the surface to around 600 hPa in height (Figs. 5d). Further, the 1200 UTC 25 May 2013 CPR sounding (Fig. 5d) measured precipitable water values over the 90th climatological percentile for that day (NOAA 2015b). The San Antonio International Airport recorded 250 mm (9.87 in.) of precipitation between 0800 and 1700 UTC 25 May with the heaviest rain rates occurring near 1200 UTC. A U.S. Geological Survey rain gauge in San Antonio recorded a 1-h rainfall accumulation of 156 mm (6.13 in.) with a 24-h accumulation of 432 mm (17.0 in.). Once again, the heaviest precipitation occurred near the southeastern edge of the Balcones Escarpment. Because of the high rainfall rates, flash flooding occurred in creeks, streams, and rivers, leading to many road closures. Three fatalities were recorded when flood waters swept cars and pedestrians off of roadways (NOAA 2015a).

Fig. 5.

As in Fig. 2, but (a)–(c) NARR analyses at 1200 UTC 25 May 2013, and (d) skew T–logp diagram showing sounding from CRP at 1200 UTC 25 May 2013.

Fig. 5.

As in Fig. 2, but (a)–(c) NARR analyses at 1200 UTC 25 May 2013, and (d) skew T–logp diagram showing sounding from CRP at 1200 UTC 25 May 2013.

3. Methods

To test the influence of the Balcones Escarpment on the intensity and distribution of precipitation in central Texas, version 3.6 of the Advanced Research core of the Weather Research and Forecasting (WRF) Model (ARW; Skamarock et al. 2008) is used to conduct numerical simulations of these three cases. All simulations use 4-km horizontal grid spacing and 50 vertical levels on a stretched grid ( m near surface; m aloft) with a model top of 50 hPa. Grid spacing on the order of 4 km has been shown to adequately resolve convective systems but not the motions of individual convective cells (e.g., Bryan et al. 2003); thus, model configurations such as these are commonly referred to as “convection allowing” rather than truly “convection resolving.” Convection-allowing models are now commonly used in operational predictions, such as the high-resolution Rapid Refresh (RAP) and North American Mesoscale Forecast System (NAM) CONUS nest models run by the National Centers for Environmental Prediction.

For each of the cases described in section 2, a control and a terrain-modified model run, in which the Balcones Escarpment is removed, are performed. Identical lateral boundary conditions (LBCs) and model physics (Table 1) are used across both the control and terrain-modified simulations, and the initial conditions (ICs) are identical except for the minor changes related to the terrain modification discussed below. To remove the Balcones Escarpment and not introduce model physical imbalances, the terrain modification is done prior to the execution of the ARW preprocessing for the October and May 2013 cases. The terrain was manually removed after this step for the June 2010 case, since the initial conditions were derived from model runs already completed from another study on this event (see next paragraph). The underlying model terrain is specified at a 10-arc-min resolution taken from the standard ARW geography files to create a slightly smoothed terrain field for modification. The Balcones Escarpment is removed in the model by extending the Texas coastal plain westward from Houston to Del Rio, Texas, and northward to the Red River at the Texas–Oklahoma border (Fig. 1b). A slight terrain gradient approximately equal to the slope from the Gulf of Mexico to the Balcones Escarpment is maintained through the entire region of terrain modification to ensure the proper representation of an extension of the Texas coastal plain (Fig. 1d). At the grid points where the model terrain is lowered, atmospheric information must be supplemented from the parent model initial conditions to “fill in” the atmosphere that was previously below ground. This is accomplished by obtaining pressure-level information that was previously below ground from the parent model, which contains information on all pressure levels even if below the terrain surface. The initial thermodynamic data on these pressure levels in the model are obtained from the U.S. Standard Atmosphere, 1976 (COESA 1976), and the wind at the lowest pressure level above ground is applied at all levels that were previously below ground but now above. Similar terrain modification procedures have been used previously in Alcott and Steenburgh (2013), Schumacher et al. (2015), and Morales et al. (2015). Figure 1 depicts the prescribed model elevation used in the control and terrain-modified runs, as well as the extent of the model domain.

Table 1.

ARW model configuration.

ARW model configuration.
ARW model configuration.

The ICs and LBCs for the aforementioned events are chosen based upon the parent model’s depiction of the precipitation field in order to obtain a control run that reasonably simulated the observed precipitation. The ICs and LBCs for the 25 May 2013 and 31 October 2013 cases were obtained from members of the National Oceanic and Atmospheric Administration’s (NOAA’s) second-generation global medium-range ensemble reforecast dataset (Table 2) (Reforecast-2) (Hamill et al. 2013). The ICs and LBCs for the 9 June 2010 case were obtained from a member of one of the ensembles described by Schumacher and Clark (2014)—namely, member 20 of the “single_24hr” configuration (Table 2). This member was found to have the heavy precipitation in approximately the same location as it was observed, and was used because none of the Reforecast-2 ensemble members provided an accurate precipitation forecast at lead times of interest to this study. Each model simulation is initialized at 0000 UTC (Table 2) and run for either 48 (9 June 2010; 31 October 2013 cases) or 36 h (25 May 2013 case). The heaviest precipitation associated with each event began at least 6 h into the numerical simulation, which allowed the model to come into balance with the new topography.

Table 2.

Model initialization times and boundary conditions

Model initialization times and boundary conditions
Model initialization times and boundary conditions

To rule out the possibility that changes in precipitation were associated with random chance, experiments were carried out on the 25 May 2013 case in which the perturbations to the initial atmospheric conditions were reduced in magnitude. The LBCs and ICs for this case were obtained from member 9 of the Reforecast-2 ensemble, which allowed for the determination of the IC and LBC perturbation off the control for this member. The original IC and LBC perturbation was scaled by 0.5 and 0.75 to create two modified sets of ICs and LBCs for the 25 May 2013 case. These ICs were used to perform two additional experiments, as described above, containing a control run and a terrain-modified run, where the Balcones Escarpment was removed. This in total, for the 25 May 2013 case, created three numerical experiments based on scaled perturbation versions of the same Reforecast-2 ensemble member ICs. Any precipitation shifts associated with the removal of the Balcones Escarpment in these three runs for the 25 May 2013 case were then analyzed, along with a closer look at relevant meteorological differences. Furthermore, the entire Reforecast-2 ensemble (all 11 members) for this day was run in a similar ARW configuration to quantify the atmospheric variability associated with this event.

4. Results

The development of heavy precipitation along the Balcones Escarpment in all of the cases allows for the exploration of the effects of removing the terrain feature. Each case will be discussed individually below. The differences in precipitation, and other possible meteorological reasons for those differences between the control and terrain-modified simulations are examined in the following sections for each of the identified events.

a. 9 June 2010

In the control simulation, the areal coverage of heavy precipitation (>50 mm) (Fig. 6a) is smaller than that shown in the precipitation analysis (Fig. 3a), and the largest observed accumulations are underpredicted. On the other hand, there is a broader northwest extent of precipitation in the simulation. A slower MCV motion was produced in the control run compared to the observations, which is one possible reason for the difference in accumulated rainfall. Despite these errors in the precipitation placement, the simulation of substantial precipitation along the Balcones Escarpment still provides a modeled environment that is adequate to perform this experiment.

Fig. 6.

Modeled precipitation output for 9 Jun 2010 flooding event valid for the 18-h period ending 1800 UTC 9 Jun 2010. Color scale and time period correspond to NCEP stage-IV observations shown in Fig. 3a. Precipitation simulated by (a) the control run and (b) the terrain-modified run, and (c) the control run minus terrain-modified run precipitation difference where cold (warm) colors represent large values in the control (terrain modified) run. Boxed area in (c) represents the area over which the precipitation was averaged to create the Hovmöller plots to follow.

Fig. 6.

Modeled precipitation output for 9 Jun 2010 flooding event valid for the 18-h period ending 1800 UTC 9 Jun 2010. Color scale and time period correspond to NCEP stage-IV observations shown in Fig. 3a. Precipitation simulated by (a) the control run and (b) the terrain-modified run, and (c) the control run minus terrain-modified run precipitation difference where cold (warm) colors represent large values in the control (terrain modified) run. Boxed area in (c) represents the area over which the precipitation was averaged to create the Hovmöller plots to follow.

Both simulations produced heavy precipitation in the same area, with the magnitude of the largest accumulations being similar. There were two primary differences between them, however: the development of a region of organized precipitation accumulation to the south of KSAT stretching to Uvalde, Texas (KUVA), in the terrain-modified simulation that is not present in the control run (cf. Figs. 6a,b), and more precipitation to the southeast of KSAT in the control run compared to the terrain-modified run (Fig. 6c). Hovmöller diagrams (Hovmöller 1949) (Fig. 7) reveal that the averaged difference between the two runs develops initially around 0000 UTC 9 June 2010. The precipitation in the control run remains more stationary (Fig. 7a) in the KSAT area, while the precipitation in the terrain-modified run has a persistent westward motion with time (Fig. 7b). On the other hand, the band of heaviest precipitation in the terrain-modified run is better organized, explaining the larger accumulations within that band. The precipitation in the control run initiates slightly earlier, farther to the east (Fig. 7c) and south (Fig. 7d) than the terrain-modified run without the Balcones Escarpment. While the location of the precipitation is altered with the removal of the Balcones Escarpment, the occurrence and relative magnitude (Table 3) were not affected. In particular, the result that a better-organized MCS emerged in the terrain-modified run, and was in a location parallel to the escarpment (except with the escarpment removed), illustrates that this terrain feature is not necessarily the primary mechanism supporting heavy-rain-producing MCSs in this area.

Fig. 7.

Hovmöller (time–longitude) diagrams averaged over the box in Fig. 6c valid for the 24-h period beginning 1800 UTC 8 Oct 2010. (a) The control run, (b) the terrain-modified run, and (c) the control minus terrain-modified run difference, where cold (warm) colors represent larger values in the control (terrain-modified) run. (d) As in (c), but the area averaging has been done over latitude. The approximate longitude of KSAT is noted and will be on future figures. Black arrows on right side of plots denote time period over which precipitation accumulations are plotted in Figs. 3a and 6.

Fig. 7.

Hovmöller (time–longitude) diagrams averaged over the box in Fig. 6c valid for the 24-h period beginning 1800 UTC 8 Oct 2010. (a) The control run, (b) the terrain-modified run, and (c) the control minus terrain-modified run difference, where cold (warm) colors represent larger values in the control (terrain-modified) run. (d) As in (c), but the area averaging has been done over latitude. The approximate longitude of KSAT is noted and will be on future figures. Black arrows on right side of plots denote time period over which precipitation accumulations are plotted in Figs. 3a and 6.

Table 3.

Area-averaged precipitation for the control and terrain-modified run in each case analyzed. Differences calculated off the control run. Region over which average is taken corresponds to geographic spatial extent shown in all precipitation plots (e.g., Figs. 6, 8, 10 etc.) bounded by roughly 25°–36.5°N, 91°–107°W.

Area-averaged precipitation for the control and terrain-modified run in each case analyzed. Differences calculated off the control run. Region over which average is taken corresponds to geographic spatial extent shown in all precipitation plots (e.g., Figs. 6, 8, 10 etc.) bounded by roughly 25°–36.5°N, 91°–107°W.
Area-averaged precipitation for the control and terrain-modified run in each case analyzed. Differences calculated off the control run. Region over which average is taken corresponds to geographic spatial extent shown in all precipitation plots (e.g., Figs. 6, 8, 10 etc.) bounded by roughly 25°–36.5°N, 91°–107°W.

b. 31 October 2013

The 31 October 2013 case differs from the other two cases presented here in that it occurred when strong synoptic-scale (rather than mesoscale) forcing for ascent was present. This can be seen in the large area of precipitation accumulation over the 12-h period ending 1200 UTC 31 October 2013 (Fig. 3b). The control run for this case produced a fairly representative precipitation pattern over this same period (Fig. 8a) but did not reproduce the magnitude of extreme precipitation observed near KAUS (Fig. 3b). Furthermore, the heaviest precipitation swath in the control run did not reach as far south as in the observations. Similar to the other control runs in this study, the placement of precipitation along the Balcones Escarpment is sufficient to test the influence of the terrain feature.

Fig. 8.

As in Fig. 6, but in reference to the 31 Oct 2013 flooding event. Precipitation accumulation is for the 12-h period ending 1200 UTC 31 Oct 2013.

Fig. 8.

As in Fig. 6, but in reference to the 31 Oct 2013 flooding event. Precipitation accumulation is for the 12-h period ending 1200 UTC 31 Oct 2013.

The overall precipitation pattern between the control run (Fig. 8a) and the terrain-modified run is fairly similar, except for spatial shifts in the main bands of precipitation as seen in the accumulation difference plot presented in Fig. 8c. The terrain-modified run produces precipitation in the same southwest–northeast line as the control run but is offset to the northwest (Fig. 8c). In fact, this pattern is not limited to the region associated with the Balcones Escarpment. Hovmöller diagrams show that the most noticeable difference occurs in the eastern part of Texas between 0200 and 0400 UTC 31 October 2013 (Fig. 9). The terrain-modified run has a larger amount of precipitation farther to the east around 0300 UTC (Figs. 8, 9c) compared to the control run. More broadly, this signal is associated with precipitation differences in northeast Texas (Fig. 8c) near the terrain-modified run maximum, while the control run has a maximum near Temple, Texas (KTPL). This difference is far removed from the terrain modification, however. Near the region of terrain modification, Fig. 9c shows precipitation accumulation being slightly higher for the terrain-modified run and shifted to the west compared to the control run at similar time periods (e.g., 0600 and 1000 UTC). This illustrates over time the eastward (westward) shift in the precipitation maximum when the Balcones Escarpment is included (not included) that is seen in Fig. 8c. The precipitation accumulation swaths in both runs do still move from west to east, which is not surprising given the strong upper-level winds associated with this case. Similar to the previous case, a shift in the precipitation patterns to the northwest is observed near the region of terrain modification; however, the overall characteristics and precipitation (Table 3) of the event remained comparable.

Fig. 9.

As in Figs. 7a–c, but for the 24-h period beginning 1800 UTC 30 Oct 2013.

Fig. 9.

As in Figs. 7a–c, but for the 24-h period beginning 1800 UTC 30 Oct 2013.

c. 25 May 2013

The control simulation for the 25 May 2013 case produces a broad region of intense rainfall, with localized totals exceeding 300 mm (Fig. 10a). The control run precipitation shield is shifted to the northwest in comparison to the analyzed precipitation (Fig. 3c), with further precipitation production in west Texas. Numerical simulations, including almost all members of the Reforecast-2 ensemble, did not move the parent MCV far enough south to replicate the observed precipitation pattern.

Fig. 10.

Modeled precipitation output for the three experiments performed for the 25 May 2013 flooding case valid for the 12-h period ending 1800 UTC 25 May 2013. (a)–(c) Full perturbation simulation associated with the ICs of member 9 of the Reforecast-2 ensemble, (d)–(f) three-quarters of the full perturbation, and (g)–(i) one-half of the full perturbation. (a),(d),(g) Precipitation from the control; (b),(e),(h) precipitation from the terrain-modified run; and (c),(f),(i) control minus terrain-modified difference, where cold (warm) colors represent larger values in the control (terrain modified) run. The boxed region in (c) depicts the area over which precipitation was averaged to create the Hovmöller plot for the control simulation in Fig. 11.

Fig. 10.

Modeled precipitation output for the three experiments performed for the 25 May 2013 flooding case valid for the 12-h period ending 1800 UTC 25 May 2013. (a)–(c) Full perturbation simulation associated with the ICs of member 9 of the Reforecast-2 ensemble, (d)–(f) three-quarters of the full perturbation, and (g)–(i) one-half of the full perturbation. (a),(d),(g) Precipitation from the control; (b),(e),(h) precipitation from the terrain-modified run; and (c),(f),(i) control minus terrain-modified difference, where cold (warm) colors represent larger values in the control (terrain modified) run. The boxed region in (c) depicts the area over which precipitation was averaged to create the Hovmöller plot for the control simulation in Fig. 11.

The precipitation patterns in the control and terrain-modified simulations are quite similar (cf. Figs. 10a,b). Figure 10c illustrates that the precipitation maximum in the control run is located southeast of that in the terrain-modified run; or in other words, it implies that the removal of the Balcones Escarpment would serve to shift the maximum precipitation to the northwest. However, similar to the other runs, there is not a substantial change in total precipitation when the Balcones Escarpment is removed (Table 3).

The control and the terrain-modified runs initially both produce a nearly stationary region of precipitation to the west of KSAT (Figs. 11a,b). However, a general shift in precipitation to the west in the terrain-modified run throughout the analyzed period can be seen. This signal is clearest during the time of heaviest rainfall around 1200 UTC, as seen by the dipole in Fig. 11c. The terrain-modified run produces a swath of higher area-averaged precipitation (Fig. 11b) compared to the control run (Fig. 11a) at the time of heaviest precipitation (~1200 UTC) from 100.5° to 99.0°W. However, the control run produces a locally more intense region of precipitation around 99.0°W and farther east near 98.75°W (Fig. 11a). Although precipitation shifts to the north and west and local changes in accumulation maxima are seen when the Balcones Escarpment is removed, both simulations produce a heavy rain event of similar magnitude and spatial distribution.

Fig. 11.

As in Figs. 7a–c, but for the 23-h period beginning 0100 UTC 25 May 2013).

Fig. 11.

As in Figs. 7a–c, but for the 23-h period beginning 0100 UTC 25 May 2013).

The evolution of the precipitation and near-surface boundaries in the control and terrain-modified runs is broadly consistent with the observations from the event. A large north–south-oriented squall line (not shown) passed through the region and decayed early on 25 May 2013. This MCS left a broad cold pool over the area (Fig. 12a) that influenced the initiation and evolution of new convection associated with the 25 May heavy rain event. The convection responsible for the main flooding event originally initiated and organized between 0500 and 0800 UTC along the preexisting temperature gradient near the Balcones Escarpment (cf. Figs. 12b,e). As convection continued to initiate, cold pools reinforced the thermal gradient on the southern flank of the precipitation (Fig. 12e). In the control run (Figs. 13a–c), the preexisting temperature gradient is similarly located along the Balcones Escarpment (Fig. 13a) but is lacking a tongue of high- air to the southwest and is located slightly west of the analysis. While convection initiates slightly earlier in the model (around 0400 UTC; Fig. 13a) compared to observations, the subsequent evolution of the temperature gradient and cold pool is similar to what was observed. When the Balcones Escarpment is removed, the preexisting thermal gradient is located in the same area as in the control run, although the values of are different owing to the change in elevation (Fig. 13d). In the terrain-modified run, convection initiates at nearly the same time as the control run, and the evolution of the convectively generated cold pools are very similar despite the removal of the Balcones Escarpment (cf. Figs. 13b,c and Figs. 13c,f). Although the near-surface temperature gradients set up along the Balcones Escarpment, these results show that they are not caused by the presence of the terrain feature. Thus, in this case, the mesoscale forcing mechanisms and preexisting conditions have a larger impact in determining where the convection initiates than does the Balcones Escarpment.

Fig. 12.

Surface observations with virtual potential temperature (K) in black and dewpoint temperature (°C) in brown valid at (a) 0400, (b) 0800, and (c) 1100 UTC 25 May 2013. Base radar reflectivity (dBZ) from the Austin/San Antonio NWS radar (EWX) is overlaid valid at (a) 0400, (b) 0804, and (d) 1102 UTC 25 May 2013. Features discussed in the text are marked (e.g., developing/decaying systems and remnant cool air). Virtual potential temperature (contoured every 1 K) from the RAP hourly analysis valid at (d) 0400, (e) 0800, and (f) 1100 UTC 25 May 2013 with wind barbs from 10 m AGL overlaid.

Fig. 12.

Surface observations with virtual potential temperature (K) in black and dewpoint temperature (°C) in brown valid at (a) 0400, (b) 0800, and (c) 1100 UTC 25 May 2013. Base radar reflectivity (dBZ) from the Austin/San Antonio NWS radar (EWX) is overlaid valid at (a) 0400, (b) 0804, and (d) 1102 UTC 25 May 2013. Features discussed in the text are marked (e.g., developing/decaying systems and remnant cool air). Virtual potential temperature (contoured every 1 K) from the RAP hourly analysis valid at (d) 0400, (e) 0800, and (f) 1100 UTC 25 May 2013 with wind barbs from 10 m AGL overlaid.

Fig. 13.

Contoured (fill colors) virtual potential temperature on the second-lowest terrain-following model level, wind (vectors) at the same level, and vertical velocity at 3 km MSL (contoured every 50 cm s−1 starting at 50 cm s−1, green) for the (a)–(c) control and (d)–(f) terrain-modified runs. Terrain is contoured every 250 m in gray. (a),(d) Valid at 0400 UTC 25 May 2013; (b),(e) valid 4 h later at 0800 UTC; and (c),(f) valid at 1100 UTC the same day.

Fig. 13.

Contoured (fill colors) virtual potential temperature on the second-lowest terrain-following model level, wind (vectors) at the same level, and vertical velocity at 3 km MSL (contoured every 50 cm s−1 starting at 50 cm s−1, green) for the (a)–(c) control and (d)–(f) terrain-modified runs. Terrain is contoured every 250 m in gray. (a),(d) Valid at 0400 UTC 25 May 2013; (b),(e) valid 4 h later at 0800 UTC; and (c),(f) valid at 1100 UTC the same day.

Vertical sections through the MCS during its mature stage reveal that the inflow region is characterized by a strong southerly LLJ and the gradual isentropic upglide associated with mature MCVs (e.g., Fritsch et al. 1994; Trier et al. 2000a,b; Schumacher and Johnson 2009). At 1100 UTC, the simulated reflectivity pattern is quite similar between the two runs, although a slight shift in the location of the convection can be seen (Figs. 14a,b). The meridional inflow in both the control and terrain-modified runs from 1 to 2 km MSL in height exceeds 20 m s−1 (Figs. 14c,d). Furthermore, the inflow in the control run is maximized well above the maximum terrain slope, and the wind speeds are similar between the two runs, suggesting that there is minimal influence of orographic lift on the inflow to the MCS. The inflow in the terrain-modified run is at approximately the same height above ground as the control run, and a near-surface cold pool of similar magnitude exists in both runs (Figs. 14c,d). The spatial pattern and extent of the cold pools differ slightly; however, this may be because negatively buoyant downdrafts have less distance to travel to reach the ground in the control run versus the terrain-modified run. If the downdrafts have less vertical distance to travel, all else being equal, the time available for evaporative cooling is reduced and more time is available for the cold pool to spread out. The cross sections (Figs. 14c,d) show a shift in the highest simulated reflectivities to the north by about 10–20 km, consistent with the plan view precipitation maps shown earlier (Fig. 10). In general, the structure and evolution of the atmospheric forcing for convection is nearly identical whether or not the Balcones Escarpment is included, with its main influence being a small shift in the location of the maximum precipitation.

Fig. 14.

Simulated radar view and cross sections valid at 1100 UTC 25 May 2013 for (left) control and (right) terrain-modified runs. (a),(b) Simulated radar reflectivity at 1.5 km MSL contoured (fill colors), terrain elevation (gray contours ever 250 m from 250 to 2500 m), potential vorticity at 600 hPa valid 0600 UTC same day around the time of convection initiation (PVU; 1 PVU = 10−6 K kg−1 m2 s−1) (blue contours at 1 and 1.5 PVU), and black line (A–A′) represents the horizontal depiction of cross section. (c),(d) Cross sections along line in (a),(b) of meridional wind speed (shaded according to color scale), potential temperature surfaces in black contours every 2 K, and radar reflectivity in blue (contoured every 10 dBZ from 20 to 50 dBZ).

Fig. 14.

Simulated radar view and cross sections valid at 1100 UTC 25 May 2013 for (left) control and (right) terrain-modified runs. (a),(b) Simulated radar reflectivity at 1.5 km MSL contoured (fill colors), terrain elevation (gray contours ever 250 m from 250 to 2500 m), potential vorticity at 600 hPa valid 0600 UTC same day around the time of convection initiation (PVU; 1 PVU = 10−6 K kg−1 m2 s−1) (blue contours at 1 and 1.5 PVU), and black line (A–A′) represents the horizontal depiction of cross section. (c),(d) Cross sections along line in (a),(b) of meridional wind speed (shaded according to color scale), potential temperature surfaces in black contours every 2 K, and radar reflectivity in blue (contoured every 10 dBZ from 20 to 50 dBZ).

To further investigate the robustness of the northward and westward shift in precipitation discussed above, the perturbation used to create the ICs and LBCs for member 9 of the Reforecast-2 ensemble was isolated and scaled by 0.75 and 0.5, respectively, to create two new numerical simulations for the 25 May 2013 case. These additional two runs create a mini-ensemble for this case that allows for further examination of the effects of the Balcones Escarpment. A more complete method would have been to run the entire Reforecast-2 ensemble with terrain modification; however, only member 9 produced any precipitation near the Balcones Escarpment. With no precipitation predicted, the effects of the terrain modification cannot be analyzed. Thus, the scaling of the ensemble perturbation associated with member 9 was undertaken as an alternative.

In both the three-quarter and half perturbation runs a very similar spatial precipitation pattern develops compared to that in the control run, consistent with a southward-moving MCV (Fig. 10). The heaviest precipitation in the full-terrain version of these simulations is displaced northwest of that in the control (Fig. 10), but the area-averaged precipitation accumulation is similar (Table 3). When the Balcones Escarpment is removed in both of the scaled perturbation runs, the precipitation accumulation shifts even farther to the north and west (Figs. 10f,i). There are varying magnitudes and orientations of the precipitation difference dipole in all three cases (Figs. 10c,f,i) for the 25 May 2013 event, but the general shift in the precipitation when the Balcones Escarpment is removed is seen in all three. This increases the confidence that the precipitation shift in all of the study simulations is not associated with chaotic convective dynamics but the terrain feature itself.

The simulations for all of the cases so far have also shown, to first order, very little difference in the pattern or magnitude of the precipitation between the control and terrain-modified runs, but that there are spatial shifts in precipitation pattern. This suggests that atmospheric processes have much more control over the distribution of precipitation than do the details of local topography. One way to put these spatial variations into context is to examine how they compare to the variations associated with uncertainties in the large-scale atmospheric pattern. This was done by comparing the differences between the control and terrain-modified simulations to the output of the full 11-member Reforecast-2 ensemble downscaled to 4-km grid spacing with WRF. Figure 15 shows that the 50-mm rainfall contours from the control and terrain-modified runs largely overlap, but there is much larger spread in the ensemble with varied ICs and LBCs. In other words, the spatial shift in precipitation associated with removing the Balcones Escarpment is much less than the spread in precipitation due to the atmospheric variability represented by the full ensemble. This corroborates the findings from the other simulations, that while the Balcones Escarpment does slightly shift the location and in some instances focus the precipitation, it is not alone responsible for the heavy precipitation nor does it determine the overall magnitude of the event.

Fig. 15.

Contours of 50-mm precipitation accumulation from ensemble output for the 25 May 2013 case valid for the 12-h period ending 1800 UTC 25 May 2013. Contoured in different hues of gray is the 50-mm precipitation accumulation for each member of a 4-km WRF ensemble created from the ICs and LBCs of the Reforecast-2 ensemble initialized 0000 UTC 25 May 2013. Blue contours correspond to the full perturbation control run for the 25 May 2013 case, and red contours correspond to the full perturbation terrain-modified run.

Fig. 15.

Contours of 50-mm precipitation accumulation from ensemble output for the 25 May 2013 case valid for the 12-h period ending 1800 UTC 25 May 2013. Contoured in different hues of gray is the 50-mm precipitation accumulation for each member of a 4-km WRF ensemble created from the ICs and LBCs of the Reforecast-2 ensemble initialized 0000 UTC 25 May 2013. Blue contours correspond to the full perturbation control run for the 25 May 2013 case, and red contours correspond to the full perturbation terrain-modified run.

5. Discussion

In an effort to evaluate the effect that the Balcones Escarpment has on the flow impinging on it, a Froude number analysis of the low-level flow in the three presented cases was undertaken. The mountain Froude number, , where U is the speed of the flow perpendicular to the obstacle, is the height of topographic feature, and N is the Brunt–Väisällä frequency is often used to determine whether flow blocking by terrain will occur (Markowski and Richardson 2010). If , then some depth of the flow will be blocked by the terrain feature. The for each case was calculated from the observed sounding presented in Figs. 2d, 4d, and 5d over the lowest 600 m of the atmosphere. The height of the Balcones Escarpment was approximated to be 400 m and the meridional component of the wind, which is approximately perpendicular to the western part of the Balcones Escarpment (Fig. 1a), averaged over the lowest three sounding observations was taken to represent U. The analysis resulted in Froude numbers from ~2.0 to ~2.4, indicating that the flow impinging on the Balcones Escarpment in these three cases is not blocked by the terrain feature. Considering that all three soundings show weakly stable low-level temperature profiles and strong low-level meridional winds, this situation is not conducive to blocking by small topographic variations. While this result is not surprising given the vertical extent of the Balcones Escarpment, it speaks to the nature of the orographic lift and flow interactions in these three extreme rainfall events in central Texas. Further, it illustrates one of the problems with comparing the orographic effects associated with the Balcones Escarpment to other topographic features with a larger vertical extent (Rocky Mountains, Massif Central, Blue Ridge, Black Hills, etc.). In similar environmental conditions, obstacles of only 1000 m would be needed to produce regimes where blocking of some depth of the flow is likely (i.e., ). Given these flow-blocking differences, it would be difficult to directly compare the results from this study to those focused on terrain features with a larger vertical extent. Furthermore, these aforementioned terrain features have a larger spatial expanse and are located closer to other large topographic features (e.g., the Alps in the case of the Massif Central) that create flow patterns that further complicate the orographic influences.

While the simulations show that the Balcones Escarpment is the apparent cause of the precipitation shift, the exact meteorological differences that cause the shift in precipitation when the terrain feature is removed are difficult to discern because the shift is so subtle. The specific mechanisms for the shift appear to be related to a combination of the differences in the spread of cold pools and possibly slight differences in the location where air parcels arrive at their level of free convection because of small reductions in ascent when the terrain is removed. Identification of the exact meteorological cause of the precipitation shift would be a good topic for going forward from an idealized modeling standpoint. Consequently, the authors are currently developing and running idealized simulations to attempt to address these questions in future work.

6. Summary and conclusions

In this study, three different heavy precipitation and flash-flood events were examined to evaluate the effect that the Balcones Escarpment has on extreme precipitation in central Texas. Each event had different synoptic or mesoscale characteristics but all resulted in flooding damage and fatalities along the terrain feature. Numerical simulations were run using ARW at a convection-allowing grid spacing to ensure that a reasonable representation of the event was produced along the Balcones Escarpment. For each case, an experiment was conducted in which the elevation gradient associated with the Balcones Escarpment was removed.

To first order, the removal of the Balcones Escarpment did not change the precipitation characteristics of the events presented in this study. The occurrence and magnitude of the events were not significantly altered, with the overall spatial pattern and area-averaged precipitation showing little change when the terrain feature was removed. However, a shift in the precipitation to the north and west was found in simulations of all three cases. This shift in precipitation associated with removing the Balcones Escarpment, when compared to an ARW ensemble based on the parent Reforecast-2 ICs and LBCs, was much smaller than shifts associated with typical ensemble variability. Although hydrologic factors associated with the Balcones Escarpment make central Texas prone to flooding, it does not appear to cause the extreme precipitation events, of the scale analyzed in this study, that lead to flooding. Based on the results of this study, we conclude that when the synoptic to mesoscale ingredients for extreme convective precipitation are in place over the Balcones Escarpment, the terrain does not directly affect the occurrence or magnitude of precipitation in the region. However, it does appear to affect the spatial distribution of the accumulated rainfall in a small but consistent way—namely, by shifting the axis of heaviest precipitation slightly to the northwest. While the Balcones Escarpment did not seem to directly affect the occurrence and magnitudes of the extreme precipitation events studied here, the increased hydrological propensity for dangerous flash flooding associated with the terrain feature should not be ignored. Ongoing work will attempt to examine the effect of the terrain rise on smaller-scale atmospheric phenomena, such as isolated thunderstorms. Since there is a spectrum of convective processes that can produce locally heavy precipitation, it is important to fully understand the meteorological influences the Balcones Escarpment has in this flood-prone region.

Acknowledgments

The authors thank Gregory Herman, Robert Tournay, Stacey Hitchcock, John Peters, Don Conlee, editor Daniel Kirshbaum, and three anonymous reviewers for their helpful comments and discussion regarding this work. High-performance computing resources from Yellowstone (ark:/85065/d7wd3xhc) were provided by the National Center for Atmospheric Research’s (NCAR) Computational and Information Systems Laboratory, which is sponsored by the National Science Foundation. NARR analyses were obtained from the NOAA/ESRL/Physical Sciences Division, and stage-IV analyses were provided by NCAR. This research was supported by National Science Foundation Grant AGS-1157425; National Science Foundation Graduate Research Fellowship Grant DGE-1321845, Amendment 3; and the National Science Foundation Science and Technology Center for Multiscale Modeling of Atmospheric Processes (CMMAP), managed by Colorado State University under Cooperative Agreement AGS-04252473.

REFERENCES

REFERENCES
Alcott
,
T. I.
, and
W. J.
Steenburgh
,
2013
:
Orographic influences on a Great Salt Lake–effect snowstorm
.
Mon. Wea. Rev.
,
141
,
2432
2450
, doi:.
Ashley
,
S. T.
, and
W. S.
Ashley
,
2008
:
Flood fatalities in the United States
.
J. Appl. Meteor. Climatol.
,
47
,
805
818
, doi:.
Baker
,
V. R.
,
1975
: Flood hazards along the Balcones Escarpment in central Texas alternative approaches to their recognition, mapping, and management. Texas Bureau of Economic Geology Geological Circular 75-5, 22 pp.
Baker
,
V. R.
,
1977
:
Stream-channel response to floods, with examples from central Texas
.
Geol. Soc. Amer. Bull.
,
88
,
1057
1071
, doi:.
Banta
,
R. M.
,
1984
:
Daytime boundary-layer evolution over mountainous terrain. Part I: Observations of the dry circulations
.
Mon. Wea. Rev.
,
112
,
340
356
, doi:.
Blumen
,
W.
,
1990
: Atmospheric Processes over Complex Terrain. Meteor. Monogr., No. 45, Amer. Meteor. Soc., 323 pp.
Bresson
,
E.
,
V.
Ducrocq
,
O.
Nuissier
,
D.
Ricard
, and
C.
de Saint-Aubin
,
2012
:
Idealized numerical simulations of quasi-stationary convective systems over the northwestern Mediterranean complex terrain
.
Quart. J. Roy. Meteor. Soc.
,
138
,
1751
1763
, doi:.
Bryan
,
G. H.
,
J. C.
Wyngaard
, and
J. M.
Fritsch
,
2003
:
Resolution requirements for the simulation of deep moist convection
.
Mon. Wea. Rev.
,
131
,
2394
2416
, doi:.
Burby
,
R. J.
,
2001
:
Flood insurance and floodplain management: The U.S. experience
.
Global Environ. Change
,
3B
,
111
122
, doi:.
Caracena
,
F.
, and
J.
Fritsch
,
1983
:
Focusing mechanisms in the Texas Hill Country flash floods of 1978
.
Mon. Wea. Rev.
,
111
,
2319
2332
, doi:.
Caran
,
S. C.
, and
V. R.
Baker
,
1986
: Flooding along the Balcones Escarpment, central Texas. The Balcones Escarpment: Geology, Hydrology, Ecology and Social Development in Central Texas, Geological Society of America, 1–14.
Chen
,
F.
, and
J.
Dudhia
,
2001
:
Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity
.
Mon. Wea. Rev.
,
129
,
569
585
, doi:.
Cifelli
,
R.
,
N.
Doesken
,
P.
Kennedy
,
L. D.
Carey
,
S. A.
Rutledge
,
C.
Gimmestad
, and
T.
Depue
,
2005
:
The Community Collaborative Rain, Hail, and Snow Network: Informal education for scientists and citizens
.
Bull. Amer. Meteor. Soc.
,
86
,
1069
1077
, doi:.
COESA
,
1976
: U.S. Standard Atmosphere, 1976. NOAA, 227 pp.
Collier
,
C. G.
,
2007
:
Flash flood forecasting: What are the limits of predictability?
.
Quart. J. Roy. Meteor. Soc.
,
133
,
3
23
, doi: .
Costa
,
J. E.
,
1987
:
Hydraulics and basin morphometry of the largest flash floods in the conterminous United States
.
J. Hydrol.
,
93
,
313
338
, doi:.
Cutter
,
S. L.
,
1996
:
Vulnerability to environmental hazards
.
Prog. Hum. Geogr.
,
20
,
529
539
, doi:.
Doswell
,
C. A.
, III
,
H. E.
Brooks
, and
R. A.
Maddox
,
1996
:
Flash flood forecasting: An ingredients-based methodology
.
Wea. Forecasting
,
11
,
560
581
, doi:.
Ducrocq
,
V.
,
O.
Nuissier
,
D.
Ricard
,
C.
Lebeaupin
, and
T.
Thouvenin
,
2008
:
A numerical study of three catastrophic precipitating events over southern France. II: Mesoscale triggering and stationarity factors
.
Quart. J. Roy. Meteor. Soc.
,
134
,
131
145
, doi:.
Fritsch
,
J. M.
,
J. D.
Murphy
, and
J. S.
Kain
,
1994
:
Warm-core vortex amplification over land
.
J. Atmos. Sci.
,
51
,
1780
1807
, doi:.
Funk
,
T.
,
2006
: Heavy convective rainfall forecasting: A look at elevated convection, propagation, and precipitation efficiency. Proc. 10th Severe Storm and Doppler Radar Conf., Des Moines, IA, National Weather Association.
Hamill
,
T. M.
,
G. T.
Bates
,
J. S.
Whitaker
,
D. R.
Murray
,
M.
Fiorino
,
T. J.
Galarneau
, Jr
,
Y.
Zhu
, and
W.
Lapenta
,
2013
:
NOAA’s second-generation global medium-range ensemble reforecast dataset
.
Bull. Amer. Meteor. Soc.
,
94
,
1553
1565
, doi:.
Hong
,
S.-Y.
,
Y.
Noh
, and
J.
Dudhia
,
2006
:
A new vertical diffusion package with an explicit treatment of entrainment processes
.
Mon. Wea. Rev.
,
134
,
2318
2341
, doi:.
Hovmöller
,
E.
,
1949
:
The trough-and-ridge diagram
.
Tellus
,
1A
,
62
66
, doi:.
Iacono
,
M. J.
,
J. S.
Delamere
,
E. J.
Mlawer
,
M. W.
Shephard
,
S. A.
Clough
, and
W. D.
Collins
,
2008
: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, doi:.
Kirshbaum
,
D. J.
,
2013
:
On thermally forced circulations over heated terrain
.
J. Atmos. Sci.
,
70
,
1690
1709
, doi:.
Lean
,
H. W.
,
N. M.
Roberts
,
P. A.
Clark
, and
C.
Morcrette
,
2009
:
The surprising role of orography in the initiation of an isolated thunderstorm in southern England
.
Mon. Wea. Rev.
,
137
,
3026
3046
, doi:.
Lin
,
Y.
, and
K. E.
Mitchell
,
2005
: The NCEP stage II/IV hourly precipitation analyses: Development and applications. 19th Conf. on Hydrology, San Diego, CA, Amer. Meteor. Soc., 1.2. [Available online at https://ams.confex.com/ams/Annual2005/techprogram/paper_83847.htm.]
Maddox
,
R. A.
,
L. R.
Hoxit
,
C. F.
Chappell
, and
F.
Caracena
,
1978
:
Comparison of meteorological aspects of the Big Thompson and Rapid City flash floods
.
Mon. Wea. Rev.
,
106
,
375
389
, doi:.
Markowski
,
P.
, and
Y.
Richardson
,
2010
: Mesoscale Meteorology in Midlatitudes. John Wiley & Sons, 407 pp.
Martin
,
P. L.
, and
R.
Edwards
,
1995
: A severe weather, hurricane, and flood climatology for the Austin/San Antonio WFO county warning area. NOAA Tech. Memo. NWS SR-167, 28 pp.
Mesinger
,
F.
, and Coauthors
,
2006
:
North American Regional Reanalysis
.
Bull. Amer. Meteor. Soc.
,
87
,
343
360
, doi:.
Morales
,
A.
,
R. S.
Schumacher
, and
S. M.
Kreidenweis
,
2015
:
Mesoscale vortex development during extreme precipitation: Colorado, September 2013
.
Mon. Wea. Rev.
, doi:, in press.
Morrison
,
H. G.
,
G.
Thompson
, and
V.
Tatarskii
,
2009
:
Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: Comparison of one- and two-moment schemes
.
Mon. Wea. Rev.
,
137
,
991
1007
, doi:.
NCEP/ESRL
,
2015
: North American Regional Reanalysis. Accessed 26 January 2015. [Available online at http://nomads.ncdc.noaa.gov/data.php?name=access#narr_datasets.]
Nielsen-Gammon
,
J. W.
,
F.
Zhang
,
A. M.
Odins
, and
B.
Myoung
,
2005
:
Extreme rainfall in Texas: Patterns and predictability
.
Phys. Geogr.
,
26
,
340
364
, doi:.
NOAA
,
2015a
: Storm events database. National Climatic Data Center. Subset used: Flash flood events from 01/1950 to 01/2015, accessed 29 January 2015. [Available online at http://www.ncdc.noaa.gov/stormevents/.]
NOAA
,
2015b
: Storm Prediction Center sounding climatology page. [Available online at http://www.spc.noaa.gov/exper/soundingclimo/.]
NWS
,
1999
: Service assessment: South Texas Floods October 17–22, 1998. National Weather Service, 26 pp. [Available online at http://www.nws.noaa.gov/om/assessments/pdfs/txflood.pdf.]
O’Connor
,
J. E.
, and
J. E.
Costa
,
2004
:
Spatial distribution of the largest rainfall-runoff floods from basins between 2.6 and 26,000 km2 in the United States and Puerto Rico
.
Water Resour. Res.
,
40
, W01107, doi:.
Patton
,
P. C.
, and
V. R.
Baker
,
1976
:
Morphometry and floods in small drainage basins subject to diverse hydrogeomorphic controls
.
Water Resour. Res.
,
12
,
941
952
, doi:.
Petersen
,
W. A.
, and Coauthors
,
1999
:
Mesoscale and radar observations of the Fort Collins flash flood of 28 July 1997
.
Bull. Amer. Meteor. Soc.
,
80
,
191
216
, doi:.
Pielke
,
R. A.
, Jr.
, and
M. W.
Downton
,
2000
:
Precipitation and damaging floods: Trends in the United States, 1932–97
.
J. Climate
,
13
,
3625
3637
, doi:.
Pontrelli
,
M. D.
,
G.
Bryan
, and
J.
Fritsch
,
1999
:
The Madison County, Virginia, flash flood of 27 June 1995
.
Wea. Forecasting
,
14
,
384
404
, doi:.
Schumacher
,
R. S.
, and
R. H.
Johnson
,
2009
:
Quasi-stationary, extreme-rain-producing convective systems associated with midlevel cyclonic circulations
.
Wea. Forecasting
,
24
,
555
574
, doi:.
Schumacher
,
R. S.
, and
A. J.
Clark
,
2014
:
Evaluation of ensemble configurations for the analysis and prediction of heavy-rain-producing mesoscale convective systems
.
Mon. Wea. Rev.
,
142
,
4108
4138
, doi:.
Schumacher
,
R. S.
,
A. J.
Clark
,
M.
Xue
, and
F.
Kong
,
2013
:
Factors influencing the development and maintenance of nocturnal heavy-rain-producing convective systems in a storm-scale ensemble
.
Mon. Wea. Rev.
,
141
,
2778
2801
, doi:.
Schumacher
,
R. S.
,
D. M.
Schultz
, and
J. A.
Knox
,
2015
:
Influence of terrain resolution on banded convection in the lee of the Rocky Mountains
.
Mon. Wea. Rev.
,
143
,
1399
1416
, doi:.
Sharif
,
H. O.
,
T. L.
Jackson
,
M. M.
Hossain
,
S.
Bin-Shafique
, and
D.
Zane
,
2010
:
Motor vehicle-related flood fatalities in Texas, 1959–2008
.
J. Transp. Saf. Secur.
,
2
,
325
335
, doi:.
Sharif
,
H. O.
,
T. L.
Jackson
,
M. M.
Hossain
, and
D.
Zane
,
2015
:
Analysis of flood fatalities in Texas
.
Nat. Hazards Rev.
,
16
, 04014016, doi:.
Skamarock
,
W.
, and Coauthors
.,
2008
: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www2.mmm.ucar.edu/wrf/users/docs/arw_v3.pdf.]
Smith
,
J. A.
,
M. L.
Baeck
,
J. E.
Morrison
, and
P.
Sturdevant-Rees
,
2000
:
Catastrophic rainfall and flooding in Texas
.
J. Hydrometeor.
,
1
,
5
25
, doi:.
Soderholm
,
B.
,
B.
Ronalds
, and
D. J.
Kirshbaum
,
2014
:
The evolution of convective storms initiated by an isolated mountain ridge
.
Mon. Wea. Rev.
,
142
,
1430
1451
, doi:.
Špitalar
,
M.
,
J. J.
Gourley
,
C.
Lutoff
,
P.-E.
Kirstetter
,
M.
Brilly
, and
N.
Carr
,
2014
: Analysis of flash flood parameters and human impacts in the US from 2006 to 2012. J. Hydrol., 519A, 863–870, doi:.
Stevenson
,
S. N.
, and
R. S.
Schumacher
,
2014
:
A 10-year survey of extreme rainfall events in the central and eastern United States using gridded multisensor precipitation analyses
.
Mon. Wea. Rev.
,
142
,
3147
3162
, doi:.
Trier
,
S. B.
,
C. A.
Davis
, and
J. D.
Tuttle
,
2000a
:
Long-lived mesoconvective vortices and their environment. Part I: Observations from the central United States during the 1998 warm season
.
Mon. Wea. Rev.
,
128
,
3376
3395
, doi:.
Trier
,
S. B.
,
C. A.
Davis
, and
W. C.
Skamarock
,
2000b
:
Long-lived mesoconvective vortices and their environment. Part II: Induced thermodynamic destabilization in idealized simulations
.
Mon. Wea. Rev.
,
128
,
3396
3412
, doi:.
Weckwerth
,
T. M.
,
L. J.
Bennett
,
L.
Jay Miller
,
J.
Van Baelen
,
P.
Di Girolamo
,
A. M.
Blyth
, and
T. J.
Hertneky
,
2014
:
An observational and modeling study of the processes leading to deep, moist convection in complex terrain
.
Mon. Wea. Rev.
,
142
,
2687
2708
, doi:.