Laserfiche WebLink
<br />Station-Year Method <br /> <br />Dalrymp Ie (1960) suggests that the station-yea r method of precipitation <br />analysis can be profitably applied to flood-frequency studies. By this method, <br />described in detail by Clarke-Hafstad (1938), records for a number of stations <br />are combined into a single record with a length equal to the total number of <br />station years involved. Use of this method, however, requires that the stations <br />used have homogeneous flood-frequency characteristics and that annual station <br />data are random or from different storms (Chow, 1964). As previously discuss- <br />ed, the regression analysis did not detect significant areal trends and it is, <br />therefore, assumed that the nonmountainous area of the Arkansas River basin in <br />Colorado has relatively homogeneous flood characteristics. The intense thunder- <br />storms that produce runoff on the small study basins are typically localized in <br />nature, and examination of the station data indicates the annual peak discharge <br />rarely occurred on the same day at more than two or three of the stations. The <br />average interstation correlation coefficient (Matalas and Gilroy, 1968), based on <br />the annual flood series, was determined to be 0.2, also indicating general ran- <br />domness or independence within the annual flood series. <br /> <br />~ <br /> <br />The objective of the station-year method of analysis was to develop an im- <br />proved regional flood-frequency relation using only recorded data. This relation <br />is not intended to be used for predictive purposes at ungaged sites, but rather <br />as a comparison with the regression results that involved extensive use of syn- <br />thetic data. <br /> <br />Because runoff characteri stics and, therefore, flood-frequency relationship s <br />vary with size of drainage area, the station-year analysis was done by grouping <br />the stations by effective drainage area, simi lar to the multiple-regression anal- <br />ysis. The groups selected were (1) 0.50 to 2.99 mi2, (2) 3.00 to 7.00 mi2, and <br />(3) 7.01 to 15.0 mi2, with average drainage areas of 1.74, 5.00, and 11.0 mi2, <br />respectively. (Record lengths for the three data sets were 55, 62, and 32 <br />years, respectively.) For the data to be compatible, all peak discharges were <br />expressed as runoff per square mile of effective drainage area. the most signifi- <br />cant factor in determining flood magnitude. Results of a frequency analysis of <br />these data are summarized in table 3. For the 100-year peak discharge, average <br />runoff was found to range from 1,240 (ft'/s)/mi2 for the first group (average <br />AE=1.74 mi2) to 724 (ft'/s)/mi2 for the third group (average AE=11.0 mi2), <br /> <br />The station-year results both for the drainage-area range of the group from <br />which they were derived. and for the average effective area within the range are <br />shown in figure 6. As was found with the regression results, these results also <br />indicate a non-linear relation with drainage area. For the average areas of each <br />group station-year results are 9 to 23 percent less than, but remarkably simi lar <br />to, the regression results. The relation by McCain and Jarrett (1976) is within <br />about 8 percent for the average drainage areas of 5.00 and 11.0 mi2, but indi- <br />cates about 57 percent greater discharge for a 1. 74-mi 2 basin. <br /> <br />23 <br />