|
<br />REGIONAL ANALYSES OF STREAMFLOW CHARACTERISTICS
<br />
<br />7
<br />
<br />Use of short records on small streams
<br />
<br />f
<br />
<br />The usual regional analysis is based on some
<br />long records and some short ones, Records of
<br />floods from small drainage areas are usually
<br />short; consequently even the 10-year flood
<br />may be poorly defined, In this case a regional
<br />analysis by one of the methods previously de.
<br />scribed will tend to produce results of low
<br />reliability, On the other hand, there may be '
<br />more independence among the records for
<br />small streams than among those for large
<br />streams; if so, this should lead to increased
<br />reliability.
<br />
<br />The conterminous United States is covered
<br />by regional flood.frequency analyses, general-
<br />ly based on data for the larger streams. Since
<br />those analyses were made, 10 or more years of
<br />record have become available at many small- I
<br />basin crest.stage gage sites, and the demand
<br />for flood frequency characteristics of small
<br />streams has greatly increased, The short rec.
<br />ords on small streams could be used with the
<br />records from larger areas to produce another
<br />regional analysis, one that would encompass
<br />the whole range of drainage area sizes, Such
<br />a procedure probably would give the best an.
<br />swer, but one which would more or less dupli-
<br />cate the available results for the larger drain-
<br />age areas. Furthermore, 5 years from now one
<br />might be justified in reanalyzing the records
<br />from the small areas, and this would call for
<br />another general analysis, resulting in more
<br />duplication (or confusion),
<br />
<br />Therefore, it is sometimes desirable to pro.
<br />duce a regional analysis limited to small drain.
<br />age areas, one that will not duplicate or con-
<br />flict to any substantial extent with recently
<br />published analyses, This can be done in several
<br />ways.
<br />
<br />Given a regional analysis by the index-flood
<br />method, the defined relations can be extrapo.
<br />lated to small drainage areas, If the mean
<br />annual floods for the crest.stage stations
<br />check the extrapolated curves, the existing
<br />regional analysis may be considered applica-
<br />ble to small drainage areas, If not, the curves
<br />should be modified as indicated,
<br />
<br />Likewise, regression equations from an
<br />existing regional analysis may be checked
<br />
<br />,
<br />
<br />against data from small drainage areas, If
<br />this check indicates that those equations are
<br />not applicable, and if time and money are
<br />limited, a regression analysis applicable only
<br />to small drainage areas could be made, Be.
<br />cause of the short records available, such a
<br />regression usually will have a large standard
<br />error. Graphical regression may be adequate.
<br />An example, given in figure 4, is based on data
<br />used by Boner and Omang (1967), Note that
<br />some of the smalll0-year floods based on short
<br />records have been given little weight in de-
<br />fining the relation of figure 4, Some of the
<br />10.year floods at the larger drainage areas are
<br />defined by many years of record and are used
<br />to tie this relation into one based on records
<br />for large streams.
<br />At many sites on small streams, both flood
<br />hydrographs and the causative rainfalls at
<br />short time intervals (15 minutes or so) are
<br />being collected, After a few years, these data
<br />should be adequate to calibrate a hydrologic
<br />basin model such as the one described by
<br />Dawdy, Lichty, and Bergmann (1972), Then
<br />a long record of precipitation can be used to
<br />synthesize additional flood peaks, These syn.
<br />thetic peaks can be combined with those of
<br />record to define the frequency curve to recur-
<br />rence intervals of 50 years or more, Using
<br />frequency curves defined to that length, a
<br />standard regionalization process should give
<br />good results, The practicability of the method
<br />depends on obtaining a good relation between
<br />floods and precipitation and on the availabil.
<br />ity of an applicable long precipitation record,
<br />
<br />Defining the Row characteristic
<br />
<br />The better the frequency curves which
<br />form the basis for the regionalized relation,
<br />the better that relation will be. Therefore,
<br />some effort should be made to improve the
<br />frequency curves. Where data are available
<br />the rainfall. runoff approach described abov;
<br />can be used. Another method, utilizing his-
<br />torical data is described by Dalrymple (1960).
<br />Sometimes the definition of a frequency curve
<br />can be improved by correlation with a longer
<br />record, but this procedure generally results
<br />in improvement only if the correlation co-
<br />efficient between the two records is greater
<br />
|