The Highland Lakes on the Colorado River are in an area periodically threatened by large storms and floods. Many storms exceeding 10 inches (in.) in depth have been documented in the area, including some with depths approaching 40 in. These storms typically produce large peak discharges that often threaten lives and property. The storms sometimes occur with little warning. Steep stream slopes and thin soils characteristic of the area often cause large peak discharges and rapid movement of floods through watersheds. A procedure to predict the discharge associated with large floods is needed for the area so that appropriate peak discharges can be used in the design of flood plains, bridges, and other structures.

The U.S. Geological Survey (USGS), in cooperation with the Lower Colorado River Authority (LCRA), studied flood peaks for streams in the vicinity of the Highland Lakes of central Texas. The Highland Lakes are a series of reservoirs constructed on the Colorado River. The chain of lakes (and year each was completed) comprises Lake Buchanan (1937), Inks Lake (1938), Lake Lyndon B. Johnson (1950), Lake Marble Falls (1951), Lake Travis (1942), and lake Austin (1890). The study area (fig. 1), which includes all or parts of 21 counties in the vicinity of the Highland Lakes, was selected because most streams in the area have flood characteristics similar to streams entering the Highland Lakes. The entire study area is in a region subject to large storms.

The purpose of this report is to present (1) peak-flow frequency data for stations and equations to estimate peak-flow frequency for large streams with natural drainage basins in the vicinity of the Highland Lakes, and (2) a technique to estimate the extreme flood peak discharges for the large streams in the vicinity of the Highland Lakes. Peak-flow frequency in this report refers to the peak discharges for recurrence intervals of 2,5, 10,25,50, and 100 years. A large stream is defined as having a contributing drainage area of at least0.5 square mile (mi’); and a natural drainage basin has less than 10 percent impervious cover and less than 10 percent of its drainage area controlled by reservoirs.

The mean annual precipitation in the study area for 1951–80 ranges from about 20 in, in western Kimble County to about 34 in. at the eastern edge of Williamson County (Riggio and others, 1987, p. 23). Many large storms and catastrophic floods have occurred along or in the adjacent area west of the Balcones escarpment (fig. 1) (Dalrymple and others, 1939, Breeding and Dalrymple, 1944; Breeding and Montgomery, 1954; Schroeder and others, 1979; Caran and Baker, 1986; Slade, 1986; and Hejl and others, 1996). About a dozen storms with precipitation depths exceeding 15 in. in a few days or less have been documented in this area during the past 60 years. Some of these storms have produced world-record precipitation depths for durations less than 48 hours. The documentation for these and for other large storms indicates that they are not uniformly distributed temporally or spatially; therefore, the recurrence intervals for such storms cannot be verified (Slade, 1986, p. 17). These large storms can cause flood peaks that would exceed those that can be predicted accurately by analyses of available precipitation or flood data.

The peak-flow frequency was estimated for each of 55 qualified stations in the study area (table 1) following guidelines established by the Interagency Advisory Committee on Water Data (1982). Qualified streamflow-gaging stations for the study area are those with at least 8 years of data from natural drainage basins (sites 1–55, fig. 1). Equations to estimate peak-flow frequency for large streams with natural drainage basins in the vicinity of the Highland Lakes were developed. These equations were developed from selected stations on the basis of the relation between peak-flow frequency and basin characteristics for each station. The entire period of systematic record (through 1993) was used in the frequency analyses for each qualified station except for stations at which streamflow was regulated during part of the record. These stations are Leon River near Belton (site 1): Lampasas River near Youngsport (site 5); North Fork San Gabriel River near Georgetown (site 6); San Gabriel River at Laneport (site 12); Brady Creek at Brady (site 16); San Saba River at San Saba (site 18); Rebecca Creek near Spring Branch (site 51); and Cibolo Creek near Boerne (site 54). One or more reservoirs were completed in the basin of each of these stations during the period of systematic record. These reservoirs caused the annual peak discharges to become regulated. The annual peak discharges for 1994 and 1995 at Sandy Creek near Kingsland (site 28) were used to include data associated with extreme flooding that occurred in 1995.

The extreme flood potential in the study area was investigated using an "envelope" or "extreme flood potential" curve. This curve is based on the relation between the contributing drainage area and (1) the maximum peak discharge of record for each qualified station (table 1); (2) substantial peak discharges documented for 84 sites without stations (sites 56–139, fig. 1, table 2); and (3) 100-year peak discharges from peak-flow frequency for stations (table 1). Peak discharges estimated from this curve represent the extreme flood potential for the study area.