water

Technical Note Integrating Reservoir Operations and Flood Modeling with HEC-RAS 2D

Matthew Garcia * , Andrew Juan and Philip Bedient Department of Civil and Environmental Engineering, Rice University, , TX 77005, USA; [email protected] (A.J.); [email protected] (P.B.) * Correspondence: [email protected]; Tel.: +1-713-348-4221

 Received: 4 July 2020; Accepted: 10 August 2020; Published: 12 August 2020 

Abstract: Current free to use models developed by the United States Army Corps of Engineers (USACE) perform unique functions (e.g., hydrology, hydraulics, reservoir operations, and flood impact analysis) that are widely used in numerous studies and applications. These models are commonly set up in a framework that is limited to point source connections, which is problematic in regions with flat topography and complex hydrodynamics. The separate models need to be integrally linked and jointly considered for accurate risk communication and decision-making, especially during major storm events. Recently, (2017) exposed the shortcomings of the existing framework in West Harris County, TX, where an insufficient understanding of potential flood risk and impacts contributed to the extensive flood damages sustained in the region. This work illustrates the possibility of using a single hydraulic model, HEC-RAS 2D, to perform all hydrologic, hydraulic, and reservoir operations modeling necessary for accurate flood impact assessments. Implications of this study include a simplification of the entire flood impact analysis, which could help future flood risk communication and emergency planning.

Keywords: urban reservoir operations; urban flood modeling; flood warning; reservoir modeling

1. Introduction The U.S. Army Corps of Engineers’ (USACE) reservoir operations (e.g., flood storage, water supply, and flow regulation) are typically simulated using the Hydrologic Engineering Center—Reservoir System Simulation (HEC-ResSim) software [1]. ResSim is often used in conjunction with other HEC software packages within the Corps Water Management System (CWMS) platform. CWMS and its publicly available equivalent, watershed analysis tool (WAT), are designed to integrate multiple model processes including hydrology (HMS), reservoir operations (ResSim), hydraulics (RAS), and flood impact (FIA). While useful, setting up and running CWMS or WAT properly is challenging because of a seamless integration across all linked models being required. Moreover, CWMS and WAT are limited to point source connections, which is problematic in flat-slopped regions with frequent high-intensity storms and complex hydrodynamics. Although other modeling software exists that integrate these computations (e.g., MIKE, TUFLOW, InfoWorks ICM, etc.), the prohibitively high costs of these systems, along with the requirement of most U.S. municipalities to use the official HEC software, make their use infeasible for many regions. Harris County, , encountered the limitations of the current HEC system when the region was devastated by Hurricane Harvey in August 2017 [2]. Within the region, the Addicks and Barker reservoir system and their vicinity were among the most severely impacted. Because of the record-breaking rainfall, the reservoirs were forced to release massive outlet flows accompanied by uncontrolled spillway runoff that caused extensive flood damage [3]. Until recently, the potential flood extent and local impact caused by emergency reservoir releases in major storms such as Harvey were poorly understood. Updating the CWMS models preemptively for major

Water 2020, 12, 2259; doi:10.3390/w12082259 www.mdpi.com/journal/water Water 2020, 12, x FOR PEER REVIEW 2 of 8

Waterwere2020 poorly, 12, 2259 understood. Updating the CWMS models preemptively for major flood events 2 is of a 8 massive undertaking due to the problems mentioned above. To address this issue while still floodmaintaining events isthe a massive use of HEC undertaking software, due all to the the necessary problems flood mentioned depth above. assessment To address processes this issuewere whileintegrated still maintaining in a single the model. use of This HEC study software, demonstrates all the necessary that complex flood depth hydrodynamics, assessment processes reservoir wereoperations, integrated and in a flood single impact model. This modeling study demonstrates can be computed that complex while hydrodynamics, accounting for reservoir their operations,interdependencies and flood with impact a single modeling hydraulic can model, be computed HEC-RAS while 2D. accounting This model for integration their interdependencies will simplify withthe preemptive a single hydraulic analysis model, of the HEC-RAS flood impacts 2D. This from model reservoir integration operation will changes simplify and the mitigate preemptive the analysisdevastating of the effects flood seen impacts in the from and operation Barker r changesegion from and Hurricane mitigate the Harvey devastating in Houston effects seenand inother the regions Addicks in and the Barkerfuture. region from Hurricane Harvey in Houston and other regions in the future.

2. Methodology

2.1. Study Area Harris County, County, TX TX has has an an average average ground ground slope slope of of0.02% 0.02% and and an anaverage average annual annual rainfall rainfall of 136 of 136cm since cm since 2001 2001 [4]. In [4]. the In past the past two twodecades, decades, rapid rapid urbanizat urbanizationion (sprawl (sprawl from from the Houston the Houston metro) metro) has hasexacerbated exacerbated the region’s the region’s flood flood vulnerability vulnerability as it asshifts it shifts away away from fromcostal costal prairie prairie and rice and fields. rice fields. This Thisstudy’s study’s area of area interest of interest is 1526 is km 15262 in km West2 in Harris West HarrisCounty, County, encompassing encompassing several several watersheds: watersheds: Upper UpperCypress Cypress Creek, Creek,Addicks Addicks and Barker and BarkerReservoirs, Reservoirs, Buffalo BuBayou,ffalo Bayou,and a portion and a portion of White of Oak White Bayou Oak Bayou(Figure (Figure1). During1). Duringmajor storms, major extensive storms, extensive out-of-bank out-of-bank inundation inundation alongside alongside inter-basin inter-basin overflows overflowsfrom Upper from Cypress Upper Creek Cypress into Creek Addicks into have Addicks beenhave known been to knownoccur. Addicks to occur. and Addicks Barker, and a Barker,pair of adry pair reservoirs of dry reservoirs consisting consisting of earthen of earthen levees levees with with concrete concrete spillways spillways at at their their ends, ends, discharge discharge into BuBuffaloffalo Bayou which runs eastward through downtown Houston where where it it meets White White Oak Oak Bayou. Bayou. While recent studies have evalu evaluatedated individual individual watersheds watersheds within within this this West West Houston Houston region region [5 [,56–,7],], this study study intends intends to to holistically holistically evaluate evaluate the the complex complex hydrodynamics, hydrodynamics, reservoir reservoir operations, operations, and flood and impactsflood impacts as one as system. one system. Another Another major major distinction distinction from from a typical a typical hydrodynamic hydrodynamic model mode runl ofrun this of regionthis region is the is the length length ofthe of the simulation. simulation. Since Since reservoir reservoir operations operations happen happen on on the thescale scale ofof weeksweeks to months rather than than days, days, the the hindcast hindcast of of Hurricane Hurricane Harvey Harvey was was run run from from 24 24August August 2017 2017 through through 20 20October October 2017. 2017.

Figure 1. Overview of West Houston, TX watersheds.watersheds.

2.2. Hydrodynamic and Reservoir Operation Modeling (HEC-RAS(HEC-RAS 2D) Overview and Setup This study uses the 2D component of HEC-RASHEC-RAS (available in version 5.0.0 and more recent versions) [8 []8] for for hydrodynamic, hydrodynamic, reservoir reservoir operation, operation, and floodplain and floodplain modeling. modeling. HEC-RAS HEC 2D- generatesRAS 2D

Water 2020, 12, x FOR PEER REVIEW 3 of 8 Water 2020, 12, 2259 3 of 8 generates a mesh with properties derived from terrain and land use/land cover (LULC) datasets to represent the study area. Stormwater is introduced into the model domain by setting boundary acondition mesh with inflows properties at specific derived locations, from terrain applying and land rainfall use /landover coverthe mesh, (LULC) or both. datasets HEC to-RAS represent 2D then the studycomputes area. the Stormwater flow, flood is depth, introduced water into surface the model elevation, domain and by velocity setting for boundary every cell condition as the water inflows is atconveyed specific throughout locations, applying the domain rainfall to the over exit the boundary mesh, or conditions. both. HEC-RAS Recent 2D versi thenons computes of HEC-RAS the flow,(i.e., floodversion depth, 5.0.5 water and surface newer) elevation, also include and limited velocity scripting for every capabilitiescell as the water (rule is based conveyed operations) throughout for thestructures domain in to 2D the domains, exit boundary allowing conditions. for certain flood Recent control versions procedures of HEC-RAS such as (i.e., sluice version gate operations 5.0.5 and newer)to be coded also includeinto thelimited model. scripting capabilities (rule based operations) for structures in 2D domains, allowingFigure for 2 certain shows flood the 2D control model procedures domain, suchwhich as sluiceis developed gate operations in HEC- toRAS be codedusing intoLIDAR the terrain model. data Figurefrom 20182 shows and theconsists 2D model of over domain, 181,000 which cells. is It developedhas a base inmesh HEC-RAS resolution using of LIDAR122 m ( terrain400 ft), datawith fromhigher 2018 resolution and consists cells of defined over 181,000 along cells. levees It has (61 a m base or mesh 200 ft) resolution and chann of 122els m (15 (400 m ft), or with 50 ft). higher The resolutionManning’s cells roughness defined values along forlevees the (61 domain m or 200 are ft) associated and channels with (15land m cover or 50 data ft). The from Manning’s the 2016 roughnessNational Land values Cover for the Database domain are(NLCD) associated [9], with with landUSACE cover records data from of the the reservoir 2016 National operations Land Cover (gate Databaseopening, pool (NLCD) elevation, [9], with rate USACE of rise, recordsetc.,) roughness of the reservoir values being operations adapted (gate from opening, Kalyanapu pool elevation,et al. 2009 rate[10]. of To rise, simulate etc.) roughness Harvey, values spatially being averaged adapted net from rainfall Kalyanapu data etfrom al. 2009 NOAA’s [10]. ToMulti simulate-Radar/Multi Harvey,- spatiallySensor System averaged (MRMS) net rainfall [11] raster data wasfrom applied NOAA’s to Multi-Radar the entire domain./Multi-Sensor Net rainfall System was (MRMS) applied [11 to] rasteraccount was for applied infiltration to the in entire the domain.model domain Net rainfall since was HEC applied-RAS 2D to account currently for does infiltration not compute in the modelinfiltration domain loss. since HEC-RAS 2D currently does not compute infiltration loss.

Figure 2. Figure 2. HEC-RASHEC-RAS 2D modelmodel domain, boundary conditions, and gaugegauge locations.locations. The Addicks and Barker reservoirs were each modeled with a weir and gated culvert matching the The Addicks and Barker reservoirs were each modeled with a weir and gated culvert matching existing structures’ geo-referenced location, cross-sectional area, and levee heights (Figure3). Since the the existing structures’ geo-referenced location, cross-sectional area, and levee heights (Figure 3). USACE records of the reservoir operations (e.g. gate opening, pool elevation, and rate of rise) in Harvey Since the USACE records of the reservoir operations (e.g. gate opening, pool elevation, and rate of significantly deviated from the Water Control Manual (WCM) [12] that the scripted gates were based rise) in Harvey significantly deviated from the Water Control Manual (WCM) [12] that the scripted on, the recorded opening data from the USACE was also modeled and used for the calibration. The two gates were based on, the recorded opening data from the USACE was also modeled and used for the runs presented only differ by the method of operation for the reservoir gates and shows the difference calibration. The two runs presented only differ by the method of operation for the reservoir gates and between the WCM and the human operation of the structures. More information detailing the gate shows the difference between the WCM and the human operation of the structures. More information operation scripts (Section S.1), a GUI image of the code (Figure S1), and a version to copy (Section S.2) detailing the gate operation scripts (Section S.1.), a GUI image of the code (Figure S1), and a version are presented in the Supplemental Materials. Finally, the downstream boundary conditions for the to copy (Section S.2.) are presented in the Supplemental Materials. Finally, the downstream boundary conditions for the model were rating curves at the corresponding United States Geological Survey

Water 2020, 12, 2259 4 of 8 Water 2020, 12, x FOR PEER REVIEW 4 of 8 model(USGS) were stream rating gauge curves locations at the corresponding [13] along the United edge Statesof the Geological model domain Survey and (USGS) normal stream depth gauge for locationsoverland [areas.13] along the edge of the model domain and normal depth for overland areas.

Figure 3. Plan and cross-sectionalcross-sectional view of modeled outlet structure for AddicksAddicks Reservoir.Reservoir.

3. Results and Discussion 3. Results and Discussion The RAS 2D model of the observed operations (labeled “Modeled” below) accurately computed The RAS 2D model of the observed operations (labeled “Modeled” below) accurately computed the spillway overflows, the outlets’ tailwater conditions, and the local fluvial and pluvial flooding the spillway overflows, the outlets’ tailwater conditions, and the local fluvial and pluvial flooding which is evident from the comparisons at the reservoir pools (Figure4), at the channel gauges (Figure5), which is evident from the comparisons at the reservoir pools (Figure 4), at the channel gauges (Figure and across the modeled floodplain (Figure6a,b, and Table1). These figures also show the major 5), and across the modeled floodplain (Figure 6a,b, and Table 1). These figures also show the major distinctiondistinction betweenbetween observed operations and the coded WCM operations (labeled “Coded” below). TheThe averageaverage Nash-SutcliNash-Sutcliffeffe eeffifficiencyciency (NSE) (NSE)[ [1414]] ofof thethe ModeledModeled runrun forfor the gauges excluding Cypress CreekCreek was was 0.97 0.97 with with an an average average root-mean-squareroot-mean-square errorerror (RSME)(RSME) of of 0.18. 0.18. The Coded runrun clearly clearly performedperformed muchmuch worseworse havinghaving aa didifferentfferent operationoperation scheduleschedule thanthan thethe observed,observed, butbut itit showsshows thethe similaritysimilarity of of the the two two runs runs at the at Cypress the Cypress Creek Creek gauge whichgauge was which una wasffected unaffected by the reservoir by the operations. reservoir Theoperations. key result The for key the result Coded for run the is Coded the Piney run Point is the gauge Piney shown Point ingauge Figure shown5. This in gaugeFigure is 5. used This for gauge the determinationis used for the determination of releases in the of releases WCM and in the a distinct WCM and drop a odistinctff around drop 9 Septemberoff around 20179 September 12:00 can 20 be17 seen12:00 in can the be observed seen in andthe observed Coded runs. and This Coded drop runs. off is This due drop to the off reservoir is due poolto the levels reservoir dropping pool belowlevels thedropping range forbelow emergency the range releases. for emergency Because releases. of the much Because milder of the release much schedule milder ofrelease the actual schedule event, of thethe jumpactual on event, 9 September the jump 2017 on 9 in September the USGS 2017 record in the is much USGS less record pronounced is much thanless pronounced the more aggressive than the releasesmore aggressive from the releases Coded run.from Asthe a Coded result ofrun. this As di aff resulterence, of thethis observed difference, jump the observed is lost to attenuationjump is lost byto attenuation the time it reaches by the time Shepherd it reaches Drive Shepherd whereit Drive can still where be seenit can in still the be Coded seen in run. the TheseCoded similarities run. These insimilarities timing and in overalltiming and shape overall show shape that theshow coded that operationsthe coded operations are performing are performing the functions the functions to mimic theto mimic WCM the as WCM desired. as desired. While the While Modeled the Modeled run results run results generally generally matched matched the observed the observed records records very wellvery forwell regions for regions at and at downstreamand downstream of the of reservoirs, the reservoirs, the lack the lack of spatially-distributed of spatially-distributed rainfall rainfall in the in domainthe domain led toled an to overestimation an overestimation of the of volumesthe volumes in Cypress in Cypress Creek Creek (recorded (recorded total total precipitation precipitation was 22was cm 22 less cm in less the northwestin the northwest portion portion of the domain of the thandomain the southwestthan the southwest during the during storm). the To addressstorm). theTo rainfalladdress distribution the rainfall distribution limitation seen limitation in the Cypress seen in the Creek Cypress results Creek in the results future, in new the workfuture, is new currently work beingis currently conducted being with conducted the setup with and the linking setup of and multiple linking smaller of multiple 2D domains, smaller each2D domains, with its individual each with hyetographsits individual to hyetographs account for theto account spatial andfor the temporal spatial variabilityand temporal in the variability rainfall. in the rainfall.

Water 2020, 12, 2259 5 of 8 Water 2020, 12, x FOR PEER REVIEW 5 of 8

Water 2020, 12, x FOR PEER REVIEW 5 of 8

FigureFigure 4. Addicks 4. Addicks and and Barker Barker reservoirs reservoirs pool elevation elevation during during Hurricane Hurricane Harvey Harvey.Harvey. .

FigureFigure 5. 5.Channel Channel flow flow comparisonscomparisons during during Hurricane Hurricane Harvey Harvey..

ApartApart from from the the lackFigure lack of spatiallyof5. spatiallyChannel distributed distributedflow comparisons rainfall, during a anothernother Hurricane major major limitation Harvey limitation. of HEC of HEC-RAS-RAS 2D is 2D is the limitedthe limited options options for for modeled modeled structures. structures. Currently,Currently, the the types types of of hydraulic hydraulic structure structure that thatcould could be be Apartmodeled from within the HEClack -ofRAS spatially 2D are distributedfew and do rainfall, not have a thenother same major computational limitation detail of HEC of the-RAS 1D 2D is modeled within HEC-RAS 2D are few and do not have the same computational detail of the 1D version the limitedversion options of the model.for modeled For example, structures. bridges Currently, could be the conceptually types of hydraulic represented structure in RAS that 2D ascould a be of the model. For example, bridges could be conceptually represented in RAS 2D as a culverted weir, modeledculverted within weir, HEC but-RAS the computations2D are few and would do lack not thehave eddies the same from roundedcomputational deck piers detail or ofbank the 1D but the computations would lack the eddies from rounded deck piers or bank obstructions along the versionobstructions of the model. along the For channel. example, Although bridges inconsequential could be conceptually in regional watershed represented studies in such RAS as o 2Dne as a channel. Although inconsequential in regional watershed studies such as one presented in this paper, culvertedpresented weir, in but this the paper, computations this limitation would could lack pose the a significant eddies from problem rounded as the deck model piers domain or bank this limitationdecreases in could size because pose a significantof the compounding problem effects as the of model each individual domain decreases bridge or in the size end because results. of the obstructions along the channel. Although inconsequential in regional watershed studies such as one compoundingWhen these eff issuesects of are each addressed individual, by bridge either ormodel in the updates, end results. combined When 1D these-2D domains, issues are or addressed, other presented in this paper, this limitation could pose a significant problem as the model domain by eitherdiscovered model methods, updates, this combined methodology 1D-2D will domains, supply or more other accurate discovered and detailedmethods, information this methodology for decreases in size because of the compounding effects of each individual bridge or in the end results. will supplyreservoir more operations accurate modeling and detailed to minimize information the flooding for effects reservoir in major operations storms. Overall, modeling the excellent to minimize When these issues are addressed, by either model updates, combined 1D-2D domains, or other the floodingmatches e showffects that in major this work storms. has laid Overall, the foundation the excellent for matches this modeling show framework that this workin future has laid discoveredapplications. methods, Despite this the methodology current limitations will stated, supply HEC more-RAS accurate 2D can accurately and detailed simulate information reservoir for the foundation for this modeling framework in future applications. Despite the current limitations reservoiroperations operations in addition modeling to the tohydraulics minimize it wasthe fdesignedlooding for.effects in major storms. Overall, the excellent stated, HEC-RAS 2D can accurately simulate reservoir operations in addition to the hydraulics it was matches show that this work has laid the foundation for this modeling framework in future designed for. applications. Despite the current limitations stated, HEC-RAS 2D can accurately simulate reservoir operations in addition to the hydraulics it was designed for.

Water 20202020,, 1212,, 2259x FOR PEER REVIEW 66 of of 8

(a)

(b)

Figure 6. (a(a) )Modeled Modeled hindcast hindcast of ofobserved observed USGS USGS operations operations data data for forHarvey Harvey floodplain floodplain extent extent;; (b) (Codedb) Coded operations operations hindcast hindcast of Harvey of Harvey floodplain floodplain extent extent..

Table 1. Flood depth comparison of USGS observed high water marks from Figure6a,b. Table 1. Flood depth comparison of USGS observed high water marks from Figure 6a,b.

PointPoint ID IDUSGS USGS High High Water Water Mark Mark (m) (m) Modeled Modeled Operations Operations Figure6 a6a (m) (m) CodedCoded Operations Operations Figure Figure6b (m) 6b (m) 0 01.51.5 1.91 1.91 1.201.20 1 1.3 0.84 0.36 1 1.3 0.84 0.36 2 1.2 0.82 0.77 2 31.21.2 0.82 0.71 0.730.77 3 41.21.1 0.71 1.13 0.190.73 4 51.11.0 1.13 1.04 1.000.19 5 61.00.8 1.04 0.98 0.911.00 7 0.8 0.98 0.13 6 80.80.6 0.98 0.38 0.180.91 7 90.80.6 0.98 0.68 1.080.13 8 100.60.2 0.38 0.19 0.620.18 9 0.6 0.68 1.08 10 0.2 0.19 0.62

Water 2020, 12, 2259 7 of 8

4. Conclusions This study successfully demonstrates the capability of using a single model, HEC-RAS 2D, for combined hydrologic and hydraulic simulations, reservoir operations, and floodplain analysis in a hydrodynamically complex region (West Harris County, TX, USA). Using HEC-RAS 2D, the point source limitation of the conventional CWMS or WAT modeling framework is circumvented while still using free-to-use HEC software. Moreover, by using a single model to simulate all the relevant processes instead of linking separate models, the entire workflow is streamlined opening up the possibility for a reservoir WCM to account for more than just pool elevation and rate of rise in the future. Although this methodology has not yet been directly compared to CWMS or WAT, both methodologies using RAS for the hydraulics allows these conclusions to be drawn aside from computational run times. Preemptively knowing how the emergency operations of an urban reservoir will impact flooding can mitigate the damages like those seen during Hurricane Harvey in the future. This framework can also be used for other distributed modeling functions that integrate structure operations such as dam breach modeling or regional flood protection modeling as intended with this work. This approach reduces the computational errors associated with point source connections and unaccounted gate flows seen in the traditional methodology and is more accurate in spatially distributed flow regions. These improvements over the current methods could also improve regional modeling in other related fields such as stormwater reuse, flood warning, and emergency planning. Despite the current limitations of the model (e.g., lack of infiltration loss and spatially distributed rainfall), HEC-RAS 2D is a capable and robust tool that effectively handles combined reservoir operations and flood inundation modeling.

Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4441/12/8/2259/s1, Section S.1: HEC-RAS 2D Sluice Gate Operation Scripts, Figure S1: Rule operation code for Addicks reservoir sluice gate, Section S.2: Full Coded Script of Addicks Reservoir Gate Copied from the HEC-RAS Interface. Author Contributions: M.G. student, led the modeling, hypothesis testing, and writing. A.J., research scientist, supervised the modeling and aided in writing. P.B., faculty advisor, supervised the project. All authors have read and agreed to the published version of the manuscript. Funding: This research received external funding from the “ Flood Mitigation Consortium” https://www.houstonconsortium.com/ and the “Katy Prairie Conservancy” https://www.katyprairie.org/. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Klipsch, J.D.; Evans, T.A. Reservoir operations modeling with hec-ressim. In Proceedings of the 3rd Federal Interagency Hydrologic Modeling Conference, Reno, NV, USA, 3 April 2006; Available online: https://acwi.gov/hydrology/mtsconfwkshops/conf_proceedings/3rdFIHMC/7E_Klipsch.pdf (accessed on 11 August 2020). 2. Linder, J.; Fitzgerald, S. Hurricane Harvey—Storm and Flood Information. Harris County Flood Control. District Memorandum to HCFCD Flood Watch/Partners. Available online: https://www.hcfcd.org/Portals/62/ Harvey/immediate-flood-report-final-hurricane-harvey-2017.pdf (accessed on 4 June 2018). 3. USACE. Galveston District. Corps Releases at Addicks and Barker Dams to Begin. Press Release. 2017. Available online: https://www.swg.usace.army.mil/Media/News-Releases/Article/1291369/corps-releases-at- addicks-and-barker-dams-to-begin/ (accessed on 11 August 2020). 4. National Weather Service. Houston International Temperature and Precipitation Climate Graphs. Available online: https://www.weather.gov/hgx/climate_graphs_iah (accessed on 3 July 2020). 5. Gori, A.; Blessing, R.; Juan, A.; Brody, S.; Bedient, P. Characterizing urbanization impacts on floodplain through integrated land use, hydrologic, and hydraulic modeling. J. Hydrol. 2019, 568, 82–95. [CrossRef] 6. Juan, A.; Gori, A.; Sebastian, A. Comparing floodplain evolution in channelized and unchannelized urban watersheds in Houston, Texas. J. Flood Risk Manag. 2020, 13, e12604. [CrossRef] 7. Panakkal, P.; Juan, A.; Garcia, M.; Padgett, J.E.; Bedient, P. Towards Enhanced Response: Integration of a Flood Alert System with Road Infrastructure Performance Models. In Proceedings of the American Society of Civil Engineers SEI Structures Congress, Orlando, FL, USA, 25 April 2019. [CrossRef] Water 2020, 12, 2259 8 of 8

8. Gary, B. Hydraulic Reference Manual, version 5.0. In HEC-RAS River Analysis System; USACE Hydrologic Engineering Center: Davis, CA, USA, 2016; Available online: https://www.hec.usace.army.mil/software/hec- ras/documentation.aspx (accessed on 11 August 2020). 9. Multi-Resolution Land Characteristics Consortium. NLCD 2016 Land Cover (CONUS). Available online: https://www.mrlc.gov/data (accessed on 7 August 2020). 10. Kalyanapu, A.J.; Burian, S.J.; McPherson, T.N. Effect of land use-based surface roughness on hydrologic model output. J. Spat. Hydrol. 2009, 9, 51–71. Available online: http://spatialhydrology.net/index.php/JOSH/ article/view/84/83 (accessed on 11 August 2020). 11. NOAA, National Severe Storms Laboratory. Multi-Radar/Multi-Sensor System (MRMS). Available online: https://www.nssl.noaa.gov/projects/mrms/ (accessed on 5 August 2020). 12. USACE. Addicks and Barker Reservoirs, Buffalo Bayou and Tributaries, San Jacinto River Basin, TX: Water Control Manual; USACE Digital Library/USACE Publications: Washington, DC, USA, 2012; Available online: https://usace.contentdm.oclc.org/digital/collection/p16021coll11/id/1868/ (accessed on 11 August 2020). 13. USGS. Texas Water Dashboard, Gauges 08073600, 08074000, and 08068800. Available online: https: //txpub.usgs.gov/txwaterdashboard/index.html (accessed on 7 August 2020). 14. Nash, J.E.; Sutcliffe, J.V. River flow forecasting through conceptual models part I—A discussion of principles. J. Hydrol. 1970, 10, 282–290. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).