ABSTRACT
Many lives have been lost and much property damage incurred over many
years from floods in Central Texas, most of which have been caused by large
rainstorms. The meteorological characteristics of Central Texas, along with an
orographic influence caused by the Balcones Escarpment, produce conditions
that cause large rainstorms in the area. Many of the highest rainfall
intensities occurred in Central Texas, a 1921 storm in Thrall, Texas, for
example, produced 32 inches of rain in 12 hours, and a 1935 storm near D'Hanis,
Texas, produced 22 inches of rain in 2 hours 45 minutes.
The accurate determination of the recurrence probability of large-design
rainfalls in Central Texas is hindered by a lack of documentation of many
storms. Many large storms have been undocumented because of inadequate areal
and temporal coverage of rain gages. Lack of a complete data base contributes
to the lack of information concerning the recurrence of large storms. Accurate
prediction of design rainfalls is also hindered by the nonuniform areal and
temporal distribution of large storms. Large ranges occur areally in depths of
the largest storms and, at individual sites, large differences exist between
the depths of the greatest storms, along with temporal clustering of the large
storms.
Site specific rainfall data are commonly used to predict design rainfalls
for use in delineating flood plains and designing urban developments. Because
of the nonuniform occurrence of large storms in Central Texas, standard
statistical methods of predicting design rainfalls can produce inaccurate
results. Regional studies of the climatic and physiographic conditions in
Central Texas, along with analyses of the areal and temporal occurrences of
large storms, could be beneficial in providing methods to better predict large-design
rainfalls in the area.
Figure 1 : Locations, dates, and depths for selected large rainstorms in Central Texas.
Figure 2 : Magnitude-duration relationships for selected largest rainfalls of the world and Texas.
Figure 3: Maximum daily precipation values, 1900-1984.
Table 1: Number of months for which monthly precipitation exceeds 10 inches, 1900-1984
INTRODUCTION AND BACKGROUND
Despite immense public expenditures for flood protection, flood losses
remain substantial, costing many lives and averaging several billion dollars
per year nationally (U.S. Water Resources Council, 1968). A major part of the
national flood losses are from "catastrophic" floods--floods which have a
return period of 100 years or more, or cause failure of a flood-protection
project by exceeding the project design flood (Holmes, 1961). Many catastrophic
floods have occurred along the Balcones fault zone in Central Texas, most as a
result of extraordinary rainstorms. Many of these floods are catastrophic
because rainstorm depths exceed design amounts. The recurrence probability of
large design rainfalls in Central Texas cannot be accurately predicted. The
purpose of this report is to summarize the development and occurrence of large
rainstorms in Central Texas, to present rainfall data for some of the larger
storms, and to discuss problems in accurate determination of the recurrence probability of design rainfalls.
Many large storms have occurred in Central Texas. Most of these storms
have occurred during the months of May-July or September-October. A detailed
discussion of the causes for large storms during these months is presented by
Carr (1967). These two periods experience much precipitation because of
convective thunderstorm activity, and because migration of cooler air from the
north often encounters well-established moisture-laden winds from the Gulf of
Mexico. Also, upper-level areas of atmospheric convergence are then moving
over Texas from the west and east.
During the period May-July, the winds have intermittently prevailed from
the south long enough to have carried large quantities of water vapor from the
Gulf of Mexico far into the interior of Texas. The last of the cold air from
the winter season migrates from Canada and the Great Basin, and springtime low-pressure
troughs aloft in the westerly winds all contribute to precipitation
during this period.
By September, the first cold air of the autumn-winter season has begun to
clash with the long-established, moisture-laden, prevailing southerly winds.
Also, the severest hurricanes to affect Texas have occurred in September. The
remains of many of these hurricanes move inland to Central Texas, carrying much
moisture from the Gulf of Mexico.
Benson (1964) and Baker (1975) suggest that the physiography along the
Balcones fault zone also contributes to conditions that produce large
storms. The Balcones Escarpment, which occurs along the Balcones fault zone, separates
the gently sloping, lower altitudes of the Coastal Plains from the dissected
limestone terrains of higher altitudes prevalent in the Edwards Plateau (figure 1). The escarpment lies at right angles to the general direction of winds from
the Gulf. Moisture-laden air is cooled as it rises up the slopes, causing
condensation and subsequent precipitation along the escarpment. Close spacing
of mean-annual isohyets along the escarpment have been used to illustrate its
orographic influence (Carr, 1967).
The locations, dates, and amounts of many of the larger storms that have
been documented in Central Texas are shown in figure 1. Many long-duration storms with large rainfall depths have occurred along the escarpment; however, many shorter-duration storms of extremely high intensities have also occurred. Some of the highest reported rainfall intensities of less than 24-hour duration in the world have occurred in Texas (figure 2). The storm of September 9-10, 1921, in Thrall, Texas, for example, produced 32 inches of rain in 12 hours, and 38.2 inches in 24 hours, the greatest known
depths of these durations to occur in the continental United States. The storm
of May 31, 1935, produced 22 inches of rainfall in 2 hours 45 minutes near
D'Hanis, also a record rate for that duration.
CHARACTERISTICS OF STORMS
The characteristics of many large storms in Central Texas, however, are
unknown due to lack of documentation. Almost all storm documentation is from
rain gages, most of which are operated by the National Weather Service. The
areal and temporal coverage of these gages, as well as the type of data being
collected, are inadequate to properly document many of the large storms. In many
areas, distances between rain gages are greater than 40 miles; gaging density
is as low as one gage per 1,000 square miles. Also, many of the gages have only
short periods of record, many less than 10 years. Another gaging problem occurs
because most of the gages in Central Texas are non-recording collectors of
rainfall. At these gages, rainfall depths are measured once per day by
observers, thus only daily rainfall values are available. Storm intensities
are available only for those few gages that record incremental rainfall.
Because of these facts, the greatest depths and intensities for many storms are
not recorded, and many storms are totally undocumented. Lack of a complete data
base contributes to the lack of information concerning the recurrence of large
rainstorms.
Another problem in predicting large storms is caused by large areal ranges
in the depths of the greatest storms and large differences between the largest
storm depths at individual sites. The rainfall records for many rain gages in
Central Texas are analyzed to demonstrate these characteristics. Ten gages
operated by the National Weather Service with long-term data are chosen for the
analyses. The mean-annual precipitation for these sites, which are shown in Figure 1, range from about 26 to about 36 inches. With the exception of the
Smithville gage, all the gages were installed before 1900. A common period of
1900-1984 is chosen for the analysis.
The values of the greatest daily rainfalls for each of the gages are shown by bar graphs in figure 3. The horizontal 1ines in
each bar represent the depths of the highest 3 to 6 daily values for each of the
gages during the period 1900-84. These values are listed at the top of each
bar. These data illustrate the range in the highest daily rainfall values
between gages in Central Texas. The maximum-daily rainfall for 2 of the gages
is less than 7.5 inches, while 3 of the gages have had daily rainfalls greater
than 15 inches. Figure 3 also shows the large differences between the largest storms at individual sites. For example, the highest daily value at the Smithville gage is 16.05 inches, while the second through fifth highest values are between 6.60 and 6.01 inches. Incremental
rainfalls can vary as significantly a s t he daily rainfalls.These variations and
inconsistencies in rainfall illustrate the difficulties in predicting
rainfall magnitude and intensity at specific sites.
Large storms are also unevenly distributed in time throughout sites in Central Texas. Table 1 shows, for five-year periods, the number of months for which the monthly rainfall for each gage exceeded 10 inches. The irregular frequency at which large storms occur at each gage is indicated in the table. For example, at the Austin gage, 12 of the 19 months that exceeded 10 inches of rainfall occurred during the first 30 years of the 85-year period. The Austin gage, installed in 1856, represents the first rainfall gage in Central Texas. The data for that gage demonstrate that the large storms can be irregular or "clustered" in time. For example, 11 of the 12 "wettest" months on record occurred before 1930. A rain gage that records incremental rainfall was installed in Austin in about 1928. Rainfall frequency-duration statistics, based on values from the gage, are used throughout the area as the basis for flood-plain delineations and designs for urbanization. It is likely, however, that these data are not representative of the "wet" period occurring before 1930. In Austin's case, the 130 years of rainfall data indicate that the first half of the period had many more large storms and greater storm depths than the second half of the record. The largest storms for the other gages also are temporal1y clustered, a problem that can bias statistical studies of the depths and frequencies of large rainstorms.
The most common method used to predict design rainfalls can be inadequate
because of the areal and temporal characteristics of these storms. Rainfall
frequency-duration statistics are commonly used by governing officials as the
basis for delineating floodplains and for designing urban developments.
Generally, rainfall statistics for a community are based on one rain gage in the
area. Standard statistical methods for rainfall prediction assumes the
recorded
depths or intensities to be linearly related to frequency of occurrence. This
method of prediction cannot account for the large ranges in depths of the
largest storms at the site or for temporal clustering, both of which may bias the
statistics.
SUMMARY
In summary, areal and temporal documentation of large storms is hindered by lack of appropriate gaging. Also,
large ranges occur areally in depths of the largest storms. At individual sites, large differences
between the depths of the largest storms occur, along with temporal clustering of the large
storms. These characteristics present problems in planning and managing land and water
resources. Regional studies of the magnitude, frequency, and location of large storms
would probably be very beneficial in developing methods for better predicting these
occurrences. If all relevant climatic, physiographic, and rainfall data and information
were gathered, analyzed, and interpreted, better planning and managing might reduce the threat to life and property caused by rainstorms.
REFERENCES
Baker, V.R., 1975, Flood hazards along the Balcones Escarpment in central Texas; alternative approaches to their recognition, mapping, and management: Univ. of Texas at Austin, Bureau of Economic Geology Circular 75-5, 22 p.
Baker, V. R., 1977, Stream-channel response to floods: Geological Society of America, v. 88, no. 8, p.1057-1071.
Benson, M. A., 1964, Factors affecting the occurrence of floods in the southwest: U.S. Geological Survey Water-Supply Paper 1580-D, 72 p.
Carr, J. T., Jr., 1967, The Climate and physiography of Texas: Texas Water Development Board Report 53, 27 p.
Holmes, R. C., 1961, Composition and size of flood losses, in White, G. F., ed., Papers on Flood Problems: Univ. of Chicago, Dept. of Geography Research Paper 70, p.7-20
U.S. Water Resources Council, 1968, The Nation's water resources, the first national assessment: Washington, U.S. Government Printing Office, 32 p.
go to: Contents : Next Article
in Abbott, Patrick L. and Woodruff, C.M., Jr. eds., 1986,
The Balcones Escarpment, Central Texas: Geological
Society of America, p. 15-20.
About the Library
