climate & hurricane report center...section i – comparison of hurricane strike probabilities...

CLIMATE & HURRICANE

REPORT

BROWNSVILLE, TEXAS

Comparative Assessment of Hurricane Threat, Historical Patterns, and Established Flood

Protection Mechanisms for the Brownsville, Texas Area

By: Jude A. Benavides

TABLE OF CONTENTS

Introduction.....................................................................................................................................4

Section I: Comparison of Hurricane Strike Probabilities and Return Periods along the U.S. Atlantic and Gulf of Mexico Coastlines......................................................................6

Section II: Important Factors Explaining the Reduced Threat of Impact from Tropical Cyclones for the South Texas Area.............................................................................12

Section III – Brief Discussion of Coastal versus Inland Risk for Historical Storms Impacting the Area...........................................................................................18

Section IV - The Rio Grande Valley Federal Flood Control Project and its Role in Providing Flood Protection for the LRGV...................................................................21

References.......................................................................................................................................23

CLIMATE AND HURRICANE REPORT 3

Introduction

This report provides an independent review of existing studies, data, and information that assess the relative threat of a hurricaneor tropical storm impact on Brownsville, Texas, as compared to other locations along the U.S. Atlantic and Gulf of Mexico coastlines.The report additionally provides information on factors influencing this relative threat through a review of peer-reviewed literature andgovernment agency reports (National Hurricane Center, National Climatic Data Center, etc.). A general review of inland hurricanepatterns, such as rates of inland decay for historical storms, is presented for comparison of typical storm impacts for the two Metropol-itan Statistical Areas (MSA’s) in the Lower Rio Grande Valley (LRGV), namely the McAllen – Mission – Edinburg MSA and theBrownsville – Harlingen MSA. Finally, due to its critical role in flood protection for both MSA’s, a brief discussion of the Lower RioGrande Flood Control Project (LRGFCP) is included.

The report was prepared in response to a request by the Brownsville Economic Development Council (BEDC) to investigate thethreat of hurricane impact in Brownsville relative to other coastal areas. Additional requested information included a discussion ofhow storm threat varies with inland distance – particularly relative to the observed and expected impacts between the two MSA’s inthe LRGV. The perceived threat of damaging hurricane impacts often either underestimates or overestimates the true threat in nearlyevery geographic area. The perception often is as cyclical as the pattern of hurricanes themselves, with overestimates of threats overhigher frequency periods and underestimates during hurricane “drought” periods for a region. This report attempts to present an unbi-ased review of established literature and technical findings to provide a statistically valid “best perception” for the threat of an impactrelative to other areas mostly in terms of return periods (period of time between events).

Report Structure (Table of Contents):u Section I – Comparison of Hurricane Strike Probabilities and Return Periods along the U.S. Atlantic and

Gulf of Mexico Coastlinesu Section II – Important Factors Explaining the Reduced Threat of Impact from Tropical Cyclones for the South Texas Areau Section III – Brief Discussion of Coastal versus Inland Risk for Historical Storms Impacting the Areau Section IV – The Rio Grande Valley Federal Flood Control Project and its Role in Providing Flood Protection for the LRGV

Summary of Findings:

Section I – Comparison of Hurricane Strike Probabilities and Return Periods along the U.S. Atlantic and Gulf of Mexico Coastlines

Findings show that the Brownsville, Texas area has a lower risk of being impacted by a hurricane than the majority of the U.S.Gulf of Mexico coastline and much of the Eastern seaboard up to Chesapeake Bay. Brownsville’s average return period for tropicalstorms and hurricanes of 5 years was longer than over 95% of the GOM coastline and 91 % of the coastline from Brownsville to Vir-ginia Beach, VA. In fact, in both cases, Brownsville’s return period was tied for largest along the entire coastline up to Virginia Beach,VA. Keim et al. (2007) found that the south Texas area had a return period of 5, 12, and 52 years for tropical storms, all hurricanes,and major hurricanes respectively. The same source also determined a THI (based on likelihood of impact and damage susceptibilityof each area) of 67 for Brownsville. This THI was lower than the average of 104 for the Gulf Coast and 98 for the coastline fromBrownsville to Virginia Beach, VA. The THI was lower than 82 % of the combined coastlines.

Section II – Important Factors Explaining the Reduced Threat of Impact from Tropical Cyclones for the South Texas Area

Several reasons for this reduced threat of impact to the South Texas area were investigated and are briefly explained. Among themost important is the advantageous geographic location and orientation of the South Texas coastline relative to the common track andapproach direction of tropical activity. Brownsville’s location appears to be a naturally reduced threat area from hurricanes based ontypical track patterns for tropical storms and hurricanes. Additionally, the east-facing South Texas coastline results in a narrow orshallow approach angle for storms approaching from the southeast, the most common approach direction of tropical cyclones for thisarea. This limits exposure to the more dangerous right front quadrant (RFQ) of storms that have not yet traversed at least some dis-tance over land. Lastly, Mexico’s mountainous Yucatan Peninsula lies directly in the path of a common “track alley” for storms mov-ing from the Caribbean to the Gulf of Mexico. The Yucatan serves as a critical buffer to many of the would be stronger stormsaffecting the western Gulf – limiting the track alley to the Yucatan Strait, a narrow and fairly shallow channel separating Mexico andCuba.

CLIMATE AND HURRICANE REPORT4

While storms do form in the Gulf of Mexico, a majority of these storms move in a northward direction and impact the east coast ofTexas or Louisiana. Storms that form in the Bay of Campeche are more likely to move in an easterly or south-southeasterly directionand often impact Mexico well south of Brownsville (Islam et al., 2009). Lastly, but just as importantly, the City of Brownsville is lo-cated 25 miles inland from the coast, providing a small, but seemingly important buffer from storm surge effects and an approachingstorm’s most intense winds. This buffer increases to 30-35 miles in the SSE to SE direction, the most common direction of approach-ing major storms.

Section III – Discussion of Coastal versus Inland Risk for Historical Storms Impacting the Area

The current literature supports the claim that predicting inland decay rates and inland storm activity (direction, speed of advance,wind speed decay, rainfall intensity and accumulation, tornado formation, etc.) are still not possible with any high degree of certainty.Blanket forecasts for wind and wind gusts, precipitation amounts, etc. often offer a broad range of predicted values, reflecting this un-certainty. As such, any comparison of the relative inland threat of storms between the two MSA’s in the LRGV should remain generalin nature.

This section presents findings of published literature on the topic of inland decay and specifically reviews the decay rate of 21storms that made landfall within 50 nautical miles of Brownsville from 1850-2010. Findings show that despite the relatively fastdecay rates of tropical cyclones once inland (particularly for slower moving, higher intensity storms), the distance of 35 linear milesseparating the centers of the two MSA’s results in only minor differences in storm intensity for storms affecting both MSA’s. In thecase of fast moving minor hurricanes (Category 1 and 2) and tropical storms, both MSA’s historically experience very similar windspeeds, rainfall intensities and accumulations. The only substantive threat difference between the two is that stemming from stormsurge; however, only the far eastern portions of the Brownsville area are within storm surge risk areas and only in the case of powerfulCategory 4 and 5 storms. Again, the low number of events over the period of record, the wide variability in tropical cyclone behavior(particularly inland), and the wide variability in storm size and speed of advance should limit the application of these findings to ageneral discussion and comparison only.

Section IV – The Rio Grande Valley Federal Flood Control Project and its Role in Providing Flood Protection for the LRGV

The historical threat of flooding from the Rio Grande river was substantial enough for the creation of a system of levees, diversiondams, and diversion floodways specifically designed to prevent such an occurrence. A review of the system shows that substantialflood protection as been provided from river flooding and high flows in the Rio Grande resulting from tropically-induced heavy rain-falls in upstream portions of the watershed. In particular, the two diversion dams and floodways provide the ability to drastically re-duce and carefully control flood flows through the Rio Grande past the cities of Brownsville and Matamoros.

A review of the operation of the system through one of the wettest years on record (2010) revealed minor concerns with floodwayconveyance capacity and levee integrity along the floodways. The diversion dams and levees along the Lower Rio Grande channelfunctioned as designed, resulting in flood flows being effectively managed. However, the normal use of the floodways for drainageneeds stemming from local rainfall (as opposed to upstream rainfall) was restricted during the several weeks of floodway operation forsome areas. This presented some challenges for the McAllen MSA and upper valley, particularly during TS Hermine’s passage at theend of a very wet year. This resulted in long-term ponding, some property flooding, and storage issues in the upper Valley until waterlevels in the floodways were lowered by the International Boundary and Water Commission (IBWC). The Brownsville area does notrely on these floodways at all for local drainage, and only relies in a limited fashion on the Rio Grande. As a result, local drainage is-sues were addressed quickly after the wet period ended.


Section 1 - Comparison of Hurricane Strike Probabilities along the U.S. Atlantic and Gulf of Mexico Coastline

From a review of various sources containing strike probability data, hurricane hazard indices, and return period analyses, it is evi-dent that the Brownsville, Texas area has a lower threat of hurricane impact compared to most other portions of the U.S. Gulf of Mex-ico and Southeast Atlantic coast. Data presented in this section were collected from National Hurricane Center technical memos,peer-reviewed literature, and coastal hazard studies. Additional analysis was conducted in a geographic information system (GIS)framework by analyzing the tracks of historical hurricanes impacting the South Texas coast.

The National Weather Service and National Hurricane Center’s Technical Memo No. 46 (NWS NHC 46, 1992, updated 2010) (Jar-rell et al., ) provides a review of direct and indirect strikes by county along the U.S. Gulf of Mexico and Atlantic coastline. While theoriginal version was written in 1984, the study has been updated repeatedly since that time, with storm strike counts last updated in2010. Using GIS, Jarrell et al. (2010) compiled the text and tabular data of this original report into a much more user-friendly set ofmaps (Figures 1-4).

Figures 1-4 illustrate the total number of hurricane strikes per county for the period of record from 1900-2009 as per data collectedfor NWS NHC 46. Figure 1 provides a color-coded comparison of strike counts for the entire coast. The cumulative strike counts areshown from low (blue) to high (red). The highest strike count counties are limited to four main areas: the Houston / Galveston Texasarea; a stretch of coastline from eastern Louisiana (New Orleans) to western Alabama; the southern Florida coastline; and the barrierislands of the North Carolina coast. The South Texas coastline is not part of these higher strike count regions.

Figures 2-4 provide closer views of the same data presented in Figure 1 with the notable exception that exact counts for direct andindirect strikes for each county are provided via a county label. Nine storms are shown to have struck Cameron County from 1900-2009. (An indirect strike is counted when hurricane force winds are experienced in any part of the county as a result of a nearbystorm.)

Cameron County has a lower strike count total than 70 % of the coastal counties from the Rio Grande River to the ChesapeakeBay. Cameron County has a lower strike count total than 80% of all counties among coastal counties in the Gulf of Mexico (Cameronto Miami-Dade). Among the 17 Texas coastal or near coastal counties shown in Figure 2, Cameron County is tied for the loweststrikes with Jackson County. While this is strong support for the claim that Brownsville has a lower strike probability compared to themajority of other counties along the GOM and Atlantic coastline, the approach used for this study did have some limitations.

One of the shortcomings of NWS NHC 46 is that the size of the coastal county affects the strike count tally shown in Figures 1-4.In other words, a larger county will present a larger target for both direct and indirect strikes by approaching storms. This study alsoonly figured wind speed in its evaluation of a storm strike. As a result, additional sources of data and studies were considered neces-sary.

Commencing in 1978, the National Climatic Data Center (NCDC), in conjunction with the NHC, commenced publishing a seriesentitled, “Tropical Cyclones of the North Atlantic Ocean.” This series has been updated every few years since then with a most recentupdate of July 2009. The July 2009 version is hereafter referred to as McAdie et al. (2009). The report is arguably one of the mostcomprehensive and routinely updated series about Atlantic and GOM hurricanes, providing climatological, meteorological, and gen-eral data about storms in addition to a storm-season specific accounting of tracks, intensities, and landfalls. In contrast to the directand indirect counts used in NWS NHC 46, McAdie et al. 2009 studied both the variation of tropical cyclone frequency and intensityalong the same extent of coastline using a different counting procedure. Data were collected at fifty-seven, equidistant coastal loca-tions from Port Isabel, TX to Eastport, ME (approximately 50 nautical miles apart from each other). Counts were made of the num-ber of cyclones whose centers passed within 75 nautical miles of each sampling point from 1886-2008. The count was additionallydivided into standard tropical cyclone intensity thresholds: 34 knots (all tropical storms and hurricanes), 64 knots (hurricanes) and 96knots (major hurricanes). Figure 5 shows the results of their tally, with the vertical axis showing the number of storms counted by thisprocess normalized to the number of storms over a 100 year period. The horizontal axis shows well-known geographic locations forconvenient determination of the location of the sampling point.



Figure 1: Total number of hurricane strikes by counties/parishes/boroughs, 1900-2009 Source: NWS NHC 46

Figure 2: Total number of hurricane strikes by county/parish/boroughs, 1900-2009. Texas and Louisiana.Source: NWS NHC 46


Figure 4: Total number of hurricane strikes by counties/parishes/boroughs, 1900-2009, for South Carolina,North Carolina, and Virginia. Source: NWS NHC 46

Figure 3: Total number of hurricane strikes by counties/parishes/boroughs, 1900-2009, for Florida. Source: NWS NHC 46


Analysis of the figure reveals the low incidence count for Port Isabel (Cameron County area) in comparison to the vast majority ofthe coastline from Texas to Maine. In fact, the graph clearly illustrates, when considering all tropical cyclone activity (tropical stormsand hurricanes of any category), only the extreme portions of the Northeast United States, specifically the coast of Maine, had fewerstorms pass within the given radius.

Owing to the infrequent return periods for tropical cyclones in the northeastern United States, it is helpful and appropriate to re-duce the sample size to only those points between South Texas and the Chesapeake Bay before further analyzing the data shown inFigure 5. The exclusion of these 13 stations along the New England and Northeast coastline results in a total set of 45 observationpoints. Ranking these 45 points from highest incidence to lowest incidence, Port Isabel ranks as follows for the three different inten-sity categories analyzed: (Note: a higher ranking indicates a lower incidence rate for the corresponding intensity categories)

u Ranks 44th of 45 locations with respect to all tropical cyclones (tropical storms and all category hurricanes.)u Ranks 34th of 45 locations with respect to all category hurricanesu Ranks 27th of 45 locations with respect to all major hurricanes

Presenting this in terms of the percentage of coastline with an equal or higher chance for storms to pass within 75 nautical miles ofthe respective location results in:

u When considering a tropical cyclone event of any intensity (tropical storm, minor hurricane, or major hurricane) 98% of locations along the coast between South Texas and the Chesapeake Bay had a higher number of storms passing within a 75 nautical miles of the respective location.

u For the same area, but considering only hurricanes of any category (minor and major hurricanes), 76% of locations had a higher incidence of strikes.

u And finally, considering only major hurricanes, 60% of locations had a higher incidence of strikes.

A study by Keim et al. (2007) entitled,“Spatiotemporal patterns and return peri-ods of tropical storm and hurricanestrikes from Texas to Maine” presented areturn period based analysis of 105 yearsof tropical cyclone strikes at 45 locationsfrom Brownsville, Texas to Eastport. ME.A return period is the inverse of fre-quency or strike counts over a given pe-riod – in other words, a return period isthe average time it takes for a storm of aspecific intensity to return or be experi-enced again at any location of interest.For example, if an area experienced fourmajor hurricanes over the period of 100years, the return period for major hurri-canes in that location would be 25-years(or 100/4). The report also summarizesthe literature and research completed todate on return period analyses of hurri-cane strikes in the United States and pres-ents a detailed discussion of theirimprovements on these studies. Namely,the Keim study avoids the target size biasthat plagues the county-based analysis ofNWS NHC 46 by adopting a methodol-ogy that focuses on specific coastalplaces (towns or cities) rather than sectors of coasts.


Keim et al. selected 45 coastal cities along the Gulf and East coasts of the U.S. from South Padre Island, TX to Eastport, ME, andproduced return-periods for various intensity level storms. Strike intensities at these locations were estimated from reported maximumsustained winds when approaching centers were within 80 km of the specific location – regardless of whether or not the storm actuallymade landfall in that location.

Figure 6 shows the average return periods for tropical storms, hurricanes, and major hurricanes according to Keim et al. The fig-ure shows that South Padre Island, TX, has return periods of 5, 12 and 52 years respectively for tropical storms and hurricanes com-bined, all hurricanes, and major hurricanes. Converting these numbers to frequency of events over 100 years results in incident rateslower than those found by McAdie et al. Keim’s study shows that South Padre should expect approximately 20 tropical cycloneevents, 8.5 hurricanes of any magnitude, and just under 2 major hurricanes to pass within 80 nautical miles over a period of 100 years.Section 2 of this report shows the results of my independent review of storms passing within 50 nautical miles of Brownsville, Texas.My results are more closely aligned with the return period and incidence rates of those found by Keim et al. Additionally, other stud-ies analyzing the frequency and distribution of hurricane impacts on these portions of the U.S. coastline are more in-line with Keim’sresults. These studies include Elsner and Kara (1999) and Simpson and Lawrence (1971).

Figure 6: Average return periods for tropical storms, hurricanes, and major hurricanes.Source: Keim, Barry D. and Robert A. Muller. Spatiotemporal Patterns and Return Periods of Tropical Storm and Hurricane Strikes fromTexas to Maine.


Regardless of the differences between the Keim and McAdie studies with respect to incident rates (return periods), both studiesshow the same comparatively lower risk for impact by a tropical cyclone for the South Texas area when compared to many other por-tions of the GOM and Atlantic U.S. coastlines.

A Tropical Hazard Index (THI) was also created by Keim et al. in order to provide “additional geographic perspective on the vul-nerability of the Atlantic and Gulf Coast regions.” Strikes of tropical storm intensity were assigned two points, minor hurricanes givenfour points, and major hurricanes 8 points. The points were tallied over the 105-year period of data analysis – the same as their returnperiod analysis. Figure 7 shows Keim’s THI graphed for the coastal locations between Brownsville, TX and Nags Head, NC. TheBrownsville THI is the 7th lowest index along this portion of the coast. This additional, third manner of illustrating relative hurricanethreat along the U.S. Gulf and Atlantic coastlines confirms the lower, comparative threat of a storm strike to the Brownsville, Texas re-gion. The following section of this report investigates some of the reasons behind this reduced threat.

Figure 7: Tropical Hazard Index by location – Brownsville, TX to Nags Head, NC Source: Keim et al.(2007)


Section II - Important Factors Explaining the Reduced Threat of Impact from Tropical Cyclones for the South Texas Area

This section presents a brief discussion of some of the reasons for the reduced threat of impact from tropical storms to the SouthTexas area as illustrated and supported in Section I. While several reasons for this reduced threat were discovered through research,only those well-supported in the literature and government studies are discussed in any detail.

Some of the well-documented and supported reasons for the diminished threat are listed below:

a) Advantageous geographic location relative to common historical track patterns for tropical cyclones resulting in a lower than average strike probability relative to coasts both north and south of the Brownsville, Texas area as well as other areas as discussed in Section 1;

b) Orientation of the South Texas coastline relative to the most common southeasterly approach direction of storms resulting in less frequent exposure to the more dangerous right front quadrant (RFQ) of storms that have not yet traversed at least some overland distance;

c) The critical buffering effect of Mexico’s mountainous Yucatan Peninsula for storms approaching the Gulf of Mexico from the Caribbean.

Other reasons that could possibly explain the reduced threat to South Texas were discovered through discussions with expert per-sonnel, experienced weather watchers, and anecdotal evidence; however, little to no substantive, peer-reviewed research was found forthese additional explanations and as such, will not be detailed here. Examples of these as of yet unsupported theories include: the roleof the Rio Grande river deltaic protrusion in diverting storms around Brownsville (Hurricane Dolly of 2008); the shallower waters ofthe western Gulf of Mexico causing decay of approaching storms (data both supporting and refuting this were found); and the role ofmore persistent, larger scale weather patterns (steering currents) on the mesoscale and synoptic scales diverting storms either north-ward or southward depending on the storm’s area of genesis.

It is important to note that this section does not, nor does any part of this report, review the well-established literature relating thegeographic cycles of hurricanes in the GOM and Atlantic to much larger weather patterns and events such as El Nino or La Niña in thePacific. Relationships between hurricane heavy (poor) seasons and La Niña (El Nino) years have been observed and are becomingbetter understood; however, these multi-year oscillations affect much larger geographic regions and would not explain the longer re-turn periods seen at Brownsville over the period of 140 years of data. This being said, the cyclical nature of hurricane strikes from oneportion of the coastline to another plays an important role in the groups of years during which Brownsville experiences below orabove average hurricane activity. Readers wishing to look more carefully at this topic are referred to Elsner et al. (2000) and Kushnir(1994).

a) Advantageous geographic location relative to common historical track patterns for tropical cyclones resulting in a lower than average strike probability relative to coasts both north and south of the Brownsville, Texas area as well as other areas asdiscussed in Section 1;

There are a wide-array of on-line databases and tools available for efficient perusal of historical storm tracks. These databases canbe queried so that analysis can be performed and storms grouped by basin, area of impact, intensity, year, etc. An analysis of allstorms passing in close proximity (50 miles) to Brownsville, Texas, between the years of 1860-2010 was completed usinghttp://www.csc.noaa.gov/hurricanes/ and will be discussed and summarized later in this section.

The National Hurricane Center has analyzed the prevailing tracks and common areas of occurrence for tropical cyclones for eachof the six months in the Atlantic hurricane season. The results of this analysis have been summarized in a series of six images avail-able on their home website: www.nhc.noaa.gov. Figures 8 and 9 show the six images in succession, illustrating the areas within theNorth Atlantic, Caribbean, and Gulf of Mexico that are likely, more likely, and most likely to experience tropical cyclone activity ineach of the respective months. While these images were not created for explicit interpretation of the threat of a storm impact betweenany two specific point locations, the figures do show the prevalent tracks observed over a long period of record.


Analysis of Figures 8 and 9 shows that Brownsville is never shown in the “most likely” zone / category during any month in thehurricane season. Brownsville is just on the border between the “more likely” and “likely” zone once, during the traditionally activemonth of September. Brownsville is shown in the “likely” zone for the first month of hurricane season, June, and again in August.Brownsville is not in any of the likely occurrence categories during the months of July, October, and November.

Analysis of the prevalent track patterns shows that the more common directions of approach for storms affecting the South Texasarea move from the South-Southeast early in the hurricane season, to the Southeast in August, and finally the Southeast and possiblymore easterly direction in September. By October, the influence of larger scale weather patterns, namely frontal systems from thenorthwest and a weakening of the Bermuda high that peaks in the summer, permits the prevalent track pattern to shift more to the east-ern half of the Gulf of Mexico. A comparison of these generalized track patterns with the actual approach directions of storms impact-ing the Brownsville, Texas area follows later in this report.

b) Orientation of the South Texas coastline relative to the most common southeasterly approach direction of storms resulting in less frequent exposure to the more dangerous right front quadrant (RFQ) of storms that have not yet traversed at least some overland distance.

Figure 8: Typical areas of origin and tracks for tropical cyclonesover the first half of the hurricane season. Source: National Oceanic and Atmospheric Administrationhttp://hurricanes.noaa.gov/prepare/season_zones.htm.

Figure 9: Typical areas of origin and tracks for tropical cyclonesover the latter half of the hurricane season. Source: National Oceanic and Atmospheric Administrationhttp://hurricanes.noaa.gov/prepare/season_zones.htm.


A hurricane’s angle of approach to a coastline greatly affects the intensity of winds, storm surge, rainfall, and associated damagesthat is experienced by coastal and inland areas. The right side of a tropical storm relative to its direction is considered the “more dan-gerous” semi-circle. In particular, the right-front quadrant (RFQ) is typically the strongest portion of a tropical cyclone. This rightside typically has stronger winds, more intense rainfall both near the eye and in rain bands extending miles from the center, and thestrongest storm surge. A tropical cyclone striking a coast “head on”, or perpendicular, exposes the coastline to the direct force of theright front quadrant and the more dangerous semi-circle. Storms with a shallow angle of approach to the coastline that permits theweaker left side of the storm to impact the coastline first usually results in reduced storm surge and lower inland damages from highwinds. This is due to the fact that some decay occurs prior to the RFQ making landfall in these situations.

A review of the direction of approach for all tropical cyclones passing within 50 miles of Brownsville, TX was completed for theyears 1850-2010. Figure 10 illustrates that of the 25 storms that passed within that radius over the 160 year period, the majority ap-proached from the southeasterly direction – with the strongest storms approaching solely from that direction. The south-southeasterlydirection was the next most common approach direction. This means that approach angles of 45 degrees or less are more common forSouth Texas. The east facing coastline of South Texas, combined with the common approach direction from the Southeast and south-southeast results in less frequent exposure to the RFQ for storms that have not yet traversed at least some overland distance.

Figure 10 also shows that there may be a relationship between approach direction and the intensity of approaching storms. Thetwo most recent major hurricanes to impact Brownsville (Beulah-1967 and Allen-1980) approached from the SSE and SE respectivelyafter having passed through or near the Yucatan Strait. In fact, Brownsville only experienced the left-side of hurricane Allen and inconjunction with its rapid decay from a Category 5 to a Category 3 right before landfall, possibly explains the relatively lower winds,rain, and damages experienced in the region during that event.

Figure 11 shows the distribution of the 25 storms by category and month of landfall. As is well understood by people living inSouth Texas, September and August are the high frequency months for hurricanes. Additionally, it can be seen that the three majorhurricanes to strike the area occurred in those two months as well. An early season peak in June can be seen with the month of Junecontaining 2 tropical storms, 3 Category One storms, and 1 Category Two storm. While the hurricane season officially extendsthrough November, only one storm after September 30th has passed within 50 miles of Brownsville, an unusual late season Category 2that approached Brownsville from the northeast.


10

9

8

7

6

5

4

3

2

1

0June July August September October November

Tropical Storm

Category 1

Category 2

Category 3

Category 4

Category 5

Figure 11: Distribution of the 25 storms by category and month of landfall near Brownsville.

WIND SPEED(Knots)

>= 135114 - 13596 - 11484 - 9665 - 840 - -64

Calms: 0.00%

Figure 10: Radical schematic showing number of storms by approach angleand intensity passing within 50 nautical miles of Brownsville between theyears 1850 and 2010. (Note: TS Hermine, which approached from the southsoutheast is not shown)Source: Graphic designed by Alejandro Lara, Jr. under supervision of author. data ob-tained from HURDAT and NOAA Coastal Services Center online historical hurricanedata tracker (www.csc.noaa.gov/hurricanes)


Another mechanism that provides some “protection” to the South Texas coastline from more frequent tropical cyclone events (par-ticularly larger storms) is the buffering effect of the Yucatan Peninsula of Mexico. While there is ample evidence in the hurricanetrack record of the mountainous portions of the Yucatan disrupting and weakening stronger storms, it is not impossible for storms tosurvive this disruption and continue their path into the GOM, and possibly experiencing re-intensification while in the Gulf.

An analysis of the 25 storms influencing the Brownsville region between 1850-2010 with respect to their intensity level at impactand their area of initial formation yields a strong pattern. Figures 12 and 13 show that storms originating in the GOM and impactingBrownsville tend to be weaker compared to storms originating in the Caribbean or the Atlantic.

Of the 25 storms impacting the Brownsville area, 14 formed in the Gulf of Mexico. A simple intensity scoring system was estab-lished based off of standard hurricane categories. A Category One hurricane was given an intensity score of 1, a Category Two an in-tensity score of 2, etc. Tropical storms were given an intensity score of zero. After assigning these intensity scores to the 14 stormsthat formed in the GOM, their average intensity score resulted in 0.54. The 14 storms were comprised of 2 Category Two storms, 3Category One storms, and 9 Tropical Storms.

In contrast, the 11 storms that formed outside of the GOM and impacted the Brownsville area between 1850-2010 had an averageintensity score of 2.00, consisting of 2 Category Fours, 1 Category Three, 5 Category Two’s, two Category One’s and two TropicalStorms.

Figure 12 also shows that storms originating in the GOM tend to approach the Brownsville area from a more southeast to south-southeasterly direction. Islam et al. (2009) conducted focused research on tropical cyclone strikes for the coast of Texas and con-firmed that a majority of storms that form in the GOM move in a more northerly direction, more likely impacting the east coast ofTexas or Louisiana. Storms forming in the extreme southern portion of the GOM, the Bay of Campeche, often move more westerlyand frequently impact the Mexican coastline well south of Brownsville.

Figure 13 shows that the 11 storms that formed in the Atlantic and influenced the Brownsville area were in general much strongerand had a direction of approach slightly more toward the southeast and east-southeast as compared to storms that formed in the GOM.

(Note: As of the time of this report, Hurricane Beulah’s status at landfall remains under review. Several literature sources list it asa high Category 3 at landfall; however, post-event analysis is trending toward a higher category rating – possibly even a Category 5.The author realizes that this is in contradiction to many older data reports as well as commonly agreed upon convention. However,HURDAT, the ‘best track’ dataset for hurricanes in the North Atlantic maintained by the Tropical Prediction Center, currently listsBeulah as a low range Category 5 just prior to landfall. According to this dataset, post-analysis review of the storm at landfall held itto have sustained 1-minute wind speeds of approximately 160 mph, 5 mph over the 155 mph threshold for a Category 5. The maxi-mum wind speeds recorded at South Padre of 138 mph do not directly support this; however, this could have been a result of suddenand rapid weakening just prior to landfall. For the purpose of this report, Beulah (1967) is listed as a Category 4 – a reasonable com-promise between the considered range from a high Category 3 to a low Category 5.)

Lastly, but just as importantly, the geometric center of the City of Brownsville is located 25 miles inland from the coast in astraight east – west direction , providing a small, but seemingly important buffer from storm surge effects and an approaching storm’smost intense winds. This inland buffer increases to 30-35 miles in the SSE to SE direction, the most common direction of approach-ing major storms. The next section of the report will discuss the decay rates of both historical storms that have struck in the vicinity ofthe Brownsville area and decay rates of storms in general.


Tropical Storm (<73mph) Category 1 (74 95 mph) Category 2 (96 110 mph) Tropical Storm (<73mph)Tropical Storm (<73mph) Category 1 (74 95 mph)Category 1 (74 95 mph) Category 2 (96 110 mph)Category 2 (96 110 mph)

Category 5 ( >155 mph)Category 3 (111 130 mph) Category 4 (131 155 mph) Category 5 ( >155 mph)Category 5 ( >155 mph)Category 3 (111 130 mph)Category 3 (111 130 mph) Category 4 (131 155 mph)Category 4 (131 155 mph)

Figure 12 (above): Angle of approach and intensity for storms passing within 50 nautical miles of Brownsville that origi-nated in the Gulf of Mexico. Figure 13 (below) shows the same but for storms originating in the Gulf of Mexico.


Section III – Discussion of Coastal versus Inland Risk for Historical Storms Impacting the Area

Tropical cyclones not only pose a threat to life and property along coastal counties, but also to inland counties and regions as theymove inland and dissipate. In many cases, aside from storm surge and wind damages resulting directly from the hurricane, damagesresulting from high winds due to tornados, heavy rains, and flooding for inland areas can equal or even exceed those experienced bythe coastal region where the hurricane made landfall (Kruk et al. (2010) and Rappaport (2000). In fact, Rappaport discovered that alarge portion of storm-related fatalities occurred inland, associated with decaying storm activity such as high winds, heavy rains, etc.Additionally, Rappaport found that in the 1970s, 1980s, and 1990s, freshwater floods accounted for 59% of the recorded deaths fromtropical cyclones.

This section presents a summary of previous work as well as independent analysis of post-landfall historical tracks for storms im-pacting the LRGV. The section commences with a brief summary of the state of the science of inland intensity decay and in particular,mentions one document that provides an excellent estimation of inland winds and associated damages from a comprehensive, GIS-based analysis of historical tracks and inland intensity data.

It is important to note that of the various uncertainties and difficulties associated with tropical event prediction, the prediction ofinland damages is perhaps the most difficult. The current literature supports this claim repeatedly (Kruk et al., 2010), by acknowledg-ing that hurricane intensity forecasting, rates of inland decay and geographically specific predictions of rainfall accumulations and re-sulting damages remain beyond the ability to predict with a high degree of certainty. This is particularly true as the geographicspecificity or precision of the forecast increases. Most of the work completed to date in this area has focused on numerical and com-puter simulations attempting to predict the inland decay of storms such as in Tuleya et al. (1984) and Kaplan and Demaria (1995,2001). Zandbergen (2009) examined the inland exposure of U.S. counties to tropical storm and hurricane force winds, concluding thatdespite relatively fast decay rates of hurricane winds, inland damages due to rainfall, flooding, tornados and general storm activitywere significant in many areas of the Deep South – sometimes extending as far inland as the Ohio Valley and New England.

Most of these studies, such as Emmanuel (2005), focus only on wind speed decay. For wind speed directly related to hurricanes(excluding tornado damage) Emmanuel showed that the decay rate of a land falling hurricane is rapid, losing half its wind speed valuein roughly 7 hours of inland travel, 75% in 15 h, and nearly 90% after just 1 day inland; however, this and other studies like the widevariability of inland decay rates for storms.

In general, the variety of studies on inland decay rates and associated damages can loosely be summarized as follows: decay ratesare a function of storm intensity and the rate of motion of the storm and damages are not limited to coastal communities or even near-coastal communities. Slower moving, more intense storms tend to decay faster just after landfall, while faster moving, less intensestorms tend to decay more slowly. (Kruk et al., 2010)

Historical track records of the 21 storms passing and making landfall within 50 miles of the Brownsville area between 1850 and2010 were analyzed for the purpose of obtaining a rough estimate of hurricane intensity decay rates after landfall. While a detailed in-land-risk assessment for the Lower Rio Grande Valley was beyond the scope of this report, this cursory analysis yielded useful and in-teresting results; however, the low number of events over the period of record, combined with the wide variability in storm behaviorshould limit the application of these findings to a general discussion and comparison only.

One of the primary reasons this analysis was conducted was to approximate the difference in hurricane intensity (category) for thesame storm affecting both the McAllen-Edinburg-Mission and Brownsville – Harlingen MSA’s. The Brownsville-Harlingen MSA isapproximately 25 miles inland, while the McAllen-Edinburg-Mission MSA is approximately 70 miles inland. Results of the threemajor hurricanes making landfall in the above area (Allen, Beulah, and the Category 4 Unnamed storm of 1880) maintained theirmajor hurricane strength (Category 3 or greater) for 85, 65, and 65 miles respectively – an average of 72 miles. Comparing this toother major storms along the Texas coastline with similar topographies and land roughness resulted in similar findings that large, fastmoving, higher intensity storms can maintain greater than hurricane force winds inland for over 125 miles.

Tropical cyclones making landfall as a hurricane of any category (n=12) maintained hurricane status for an average of 129 milesafter landfall. Tropical cyclones making landfall as any category hurricane or tropical storm maintained at least tropical storm statusfor an average of 216 miles after landfall. Cyclones making landfall as a tropical storm (n=9) maintained that status for an average of152 miles inland. Cyclones making landfall as a Category 1 or 2 (n=9) maintained hurricane strength for an average of 102 miles in-land.


Kruk et al. (2010) conducted a very comprehensive analysis of the entire HURDAT track and intensity database with the primaryaim of producing a series of maps with tremendous value to forecasters, emergency managers, and the general public. Their mapsshow a series of frequency distributions and estimations of return intervals for inland tropical storm and hurricane force winds.

Figure 13 shows the return period for tropical storm force winds for inland areas. Figure 14 shows the same for hurricane forcewinds. Figure 14 shows that return periods for hurricane force winds do not drop off significantly with inland distances on the scaleof 40-50 miles from the coast. In other words, the two MSA’s in the LRGV do not differ substantially in return periods for hurricaneforce winds – with a return period of about 6-8 years for the Brownsville – Harlingen MSA and a return period of about 8-9 years forthe McAllen MSA.


There is sufficient evidence in the literature as well as our study of the decay rates of historical storms having impacted the SouthTexas area to make a few general, but well-supported conclusions. First and most importantly, hurricane damages are not limited tocoastal areas. In particular, a fast moving storm (greater than 15 mph speed over land), would likely expose both MSA’s to hurricaneforce winds. Secondly, aside from storm surge threat, both MSA’s experience about equal risk from heavy rains, localized floodingfrom heavy rains, tornado formation during hurricane passage, and high winds.

From a review of the historical track data for storms making landfall in the South Texas area, it is apparent that storms, on average,maintain their intensity for a long enough distance inland to affect the two MSA’s in the LRGV with only small changes in wind speedintensity and practically no apparent difference in rainfall intensities and total rainfall accumulation. Wind speed and rainfall recordsfor historical storms for cities in both MSA’s support this claim as do a number of independent, peer reviewed publications regardingthe inland decay rates of hurricanes in general.

The threat of tornado formation during hurricane passage is actually somewhat higher for areas just inland of coastal areas and formajor storms, often at inland distances of 100-200 miles. Hurricane Beulah held the record for the number of tornadoes spawned dur-ing hurricane inland transit, with the vast majority of these tornados forming in areas as far inland as 100-150 miles.

While the threat of storm surge is by its nature limited to coastal areas, the Brownsville – Harlingen MSA is largely outside thethreat zone for Category 1, 2, and 3 storms. For Category 4 and 5 storms, generalized maps show portions of the eastern half ofBrownsville within the storm surge area. However, the prediction skill in this area is admittedly low – particularly for higher intensitystorms. Additionally, the storm surge predictions often assume the worst-case strike angle of perpendicular to the coast. As discussedearlier, perpendicular storm strikes in the South Texas area are not common.

Given the above, it is again important to recall that the low sample size (n=25), the dependence of inland decay rates on stormspeed of advance, and the wide variability of storm decay rates as a whole mean that the above conclusions should be taken only as ageneral guide and should never be used for planning purposes, emergency management decisions, and/or by the general public.

One aspect not yet discussed, and an important difference between the two is the risk due to riverine flooding from a storm impact-ing the South Texas and Northeastern Mexico region – such as Hurricane Alex in 2010. In fact, the series of storms impacting thebroader region around this area in conjunction with the wet season of 2010 highlights the important role of flood protection in theLRGV as well as its limitations. The next and final section of this report will discuss a few important specifics about the Lower RioGrande Flood Control Project.


Section IV - The Rio Grande Valley Federal Flood Control Project and its Role in Providing Flood Protection for the LRGV

This section presents a brief summary of the Lower Rio Grande Flood Control Project (LRGFCP). The LRGVCP consists of a se-ries of diversion dams, dikes, levees, and interior floodways that serve to protect the Rio Grande Valley from river-related floodingstemming from tropically-induced rainfall upstream of the Valley. The project is detailed here to highlight the operations of the sys-tem during high-flow events and how this standard operation affects local drainage operations in the two MSA’s in the LRGV. Inter-estingly, while the project design is to protect both MSA’s from riverine flooding, the operation of the system during a high-flow eventrestricts some local drainage options for the McAllen-Edinburg-Mission MSA, but does not restrict those operations for the city ofBrownsville. Additionally, the systems two diversion dams – Anzalduas and Retamal – function to divert sufficient flood flows so thatthe design flow past Brownsville / Matamoros is 20,000 cfs, a mere 8% of the design flood flow into Anzalduas dam.

The LRGFCP covers 180 river miles from Penitas, TX to just downstream of Brownsville, TX. Approximately 270 miles of leveesare located along the Rio Grande, the Arroyo Colorado, and interior floodways. Major operational systems include the Banker Flood-way, Main Floodway, North Floodway, the Arroyo Colorado, Anzalduas Dam, and Retamal Dam.

While levees are an integral part of the system, perhaps the most vital components are the two diversion dams and the systems ofinterior floodways designed to handle diverted water – preventing this water from travelling downstream and impacting theBrownsville – Matamoros area. Anzalduas dam is located just south of Mission, Texas and has the primary function of diverting theU.S. share of floodwaters in the U.S. floodways. Retamal dam, located straight south of Donna, TX, diverts Mexico’s share of flood-waters into the Mexican floodways. Both diversion dams operate in concert to limit the flows in the Rio Grande downstream of thedams so that flood flows in the Rio Grande remain at safe levels in the Brownsville – Matamoros area.

Figure 15 shows the location of the diversion dams, interior floodways on both the U.S. and Mexican sides, the component por-tions of the U.S. interior floodways (Main, Arroyo Colorado, and North floodways) and the location of major cities, notablyBrownsville / Matamoros, McAllen and Harlingen.

The system operates under specific design flood flows that were updated after Hurricane Beulah in 1967, owing to the devastatingfloods experienced in the McAllen and Harlingen areas after that storm. The current design flows are as follows:

a) 250,000 cubic feet per second (cfs) at Rio Grande City (inflow to Anzalduas Dam);b) 105,000 cfs for diversion by Anzalduas to the U.S. floodway (Main floodway)c) 105,000 cfs for diversion by Retamal to the Mexican floodwayd) 84,000 cfs in the North Floodway (diverted from the Arroyo Colorado)e) 21,000 cfs in the Arroyo Colorado at Harlingenf) 20,000 cfs in the Rio Grande at Brownsville / Matamoros.

As can be seen from Figure 15 and from the list of design flood flows, the LRGFCP can control flows in the Rio Grande pastBrownsville and Matamoros at numerous upstream points. The system successfully maintained flood flows below design levels forthe Brownsville / Matamoros area during the 2010 hurricane season – one of the wettest water years on record.

Unfortunately, one of the significant drawbacks of this system is the that the interior floodways are contained by levees out of ne-cessity – in order to not flood the areas the floodways pass through. The levee contained floodways, while preventing the Rio Grandefrom flooding downstream, severely hinder the local drainage of areas near the floodways. The reason for this is that the normaldrainage conduit for the upper valley MSA is the same floodways used by the LRGFCP. As a result, if the LRGFCP is in operationand diverting floodwaters, the upper valley MSA cannot utilize the floodways because they are already full with Rio Grande diversionflood water. Continuing, and as occurred in 2010, several upper valley municipalities were forced to retain and/or store significantfloodwaters in streets, property, and under-designed detention ponds for several weeks. In contrast, the city of Brownsville utilizes aseries of drainage canals that drain to the Brownsville Ship Channel. As such, the flood condition of the Rio Grande River does notaffect the drainage operations of Brownsville as significantly.


MatamorosMexican Floodway

References

u Elsner, J. B., A. B. Kara, M. A. Owens, 1999: Fluctuations in north atlantic hurricane frequency. J. Climate, 12, 427–437

u Elsner, James B., Kam-biu Liu, Bethany Kocher, 2000: Spatial variations in major u.s. hurricane activity: statistics and a physical mechanism. J. Climate, 13, 2293–2305.

u Emanuel, K., 2005: Divine Wind – The History and Science of Hurricanes. Oxford University Press, 285 pp.

u Fernandez-Partagas, J., and H.F. Diaz, 1996: Atlantic hurricanes in the second half of the nineteenth century. Bull. Amer. Meteor. Soc., 77, 2899-2906.

u International Boundary and Water Commission, IBWC. Home website: www.ibwc.gov

u Islam, T., W. Merrel, W. Seitz, R. Harriss, 2009: Origin, Distribution, and Timing of Texas Hurricanes: 1851-2006. Natural Hazards Review, 10, 136-145.

u Jarrell, J.D., Ph.H. Hebert, and M. Mayfield, 1992 (revised 2009): Hurricane experience levels of coastal county populations from Texas to Maine. NOAA Tech. Rep. NWS NHC-46, U.S. Department of Commerce, Washington, DC, 154 pp.

u Kaplan, J., and M. DeMaria, 1995: A simple empirical model for predicting the decay of tropical cyclone winds after landfall. J. of Applied Meteorology, 34, 2499-2512.

u Keim, Barry D., Robert A. Muller, Gregory W. Stone, 2007: Spatiotemporal patterns and return periods of tropical storm and hurricane strikes from Texas to Maine. J. Climate, 20, 3498–3509.

u Kruk, Michael C., E.J. Gibney, D. Levinson, M. Squires, 2010: A Climatology of Inland Winds from Tropical Cyclones for the Eastern Unites States. J. of Applied Meteorology, 49, 1538-1547.

u Kushnir, Y., 1994: Interdecadal variations in the North Atlantic sea surface temperature and associated atmospheric conditions. J. Climate, 7, 141-157.

u McAdie, Colin J., C.W. Landsea, C. Neumann, J. David, and E. Blake, 2009: Tropical Cyclones of the North Atlantic Ocean, 1851-2006 (with 2007 and 2007 track maps included). Historical Climatology Series 6-2. National Climatic Data Center, Asheville, NC. 243 pp.

u NOAA Coastal Services Center website: www.csc.noaa.gov/hurricanes/

u NWS NHC 46, 1992. See McAdie et al.

u Rappaport, E. N., 2000: Loss of Life in the U.S. associated with recent Atlantic tropical cyclones. Bull. Amer. Meteor. Soc., 81, 2065-2073.

u Simpson, R. H., and M. Lawrence, 1971: Atlantic hurricane frequencies along the United States Coastline. NOAA Tech. Memo. NWS-SR-58, 14 pp.

u Tuleya, R. E., M.A. Bender, and Y. Kurihara, 1984: A simulation study of the landfall of tropical cyclones. Mon. Wea. Rev., 112, 124-136.

u Zandbergen, P. A., 2009: Exposure of US counties to Atlantic tropical storms and hurricanes, 1851-2003. Nat. Hazards, 48, 83-99.


.145.659:ehonP

tSocixMe301

xaF2553.552.800dna3181.

, ellivsnworB1- FetiuSt,eert

8393.645.659:x

12578saxeT

wwww omcCDEBwww

omc.CDEB.wwwwwww

:SH UTIWTCENNOC

climate & hurricane report center...section i – comparison of hurricane strike probabilities...

Documents