|
<br />Detennination of design
<br />stonn parameters
<br />
<br />Some investigators such as Bleich (1935) and
<br />Preul and Papadakis (1973) determined the a, b, and
<br />c values by plotting rainfall data points with various
<br />assumed values of b on 10g.log paper until a straight
<br />line was established while others such as Wagnitz and
<br />Wilcoxen (1931) evaluated a, b, and c values using the
<br />method of least squares, The duration, td, of a design
<br />hyetograph must be the one that produces the
<br />maximum runoff from a given drainage area. The
<br />maximum runoff happens at the time of coneen.
<br />tration, te ' at which all parts of the drainage area may
<br />contribute to the flow concurrently. Although in the
<br />actual hyetograph, td, may be greater than, equal to,
<br />or less than te, all previous investigators (Kiefer and
<br />Chu, 1957; Bandyopadhyay, 1972; Preul and Papa-
<br />dakis, 1973) set the duration of the design hyeto-
<br />graph equal to te.
<br />
<br />Time concentration
<br />
<br />The strict determination of tc is very difficult
<br />because te depends not only on the physiographical
<br />factors such as the slope and character of runoff sur-
<br />faces, but also on the meteorological factors involved
<br />in the rainfall-runoff process. According to Jens and
<br />McPherson's (1964) study, where the drainage area
<br />served by an inlet is entirely paved, tc is assumed to
<br />vary from about 5 to 10 minutes as the length of run-
<br />off to the inlet varies from 100 ft to about 500 ft. For
<br />turfed areas, tc is usually considered to vary from
<br />about 10 minutes for lengths of runoff less than 100
<br />ft to about 30 minutes for 400 to 500 ft. For bare
<br />ground, tc may be taken somewhere between the
<br />values of paved and turfed areas, decreasing with the
<br />expected smoothness of the surface. However, de-
<br />tailed consideration of the several components con-
<br />stituting inlet concentration times is often circum-
<br />vented through establishment of a flxed tc for particu-
<br />lar types of highly developed urban areas, with 5 to
<br />15 minutes in common use (ASCE Manual No, 37,
<br />1960),
<br />
<br />Jens and McPherson (1964) have found that in
<br />small watersheds such as most urban drainaga areas,
<br />the brief interval between the occurrence of short,
<br />intense rainfall and succeeding peak runoff has a
<br />more significant effect on the magnitude of the peak
<br />rate than the time of concentration, The influence of
<br />this effect becomes less, however, as the size of the
<br />watershed increases, In general urban drainage areas
<br />possess neither the overall detention storage nor long
<br />times of concentration and other peak-flow-reducing
<br />characteristics of large watersheds, Note that use of a
<br />uniform rainfall intensity for a duration equal to Ie is
<br />only a simplifying assumption since rainfall does not
<br />truly persist at a uniform intensity for even as short a
<br />time as 5 minutes,
<br />
<br />Stonn pattern skewness
<br />
<br />The critical arrangement (i.e" time distribution)
<br />of rainfall intensities is essential to the sound design
<br />of urban drainage systems, The most recent study by
<br />Pilgrim and Cordery (1975) has clearly indicated that
<br />heavy storms, with the exception of isolated thunder.
<br />storms, vary almost randomly in the time patterns
<br />because very heavy rainfalls are generally associated
<br />with highly turbulent unstable air-streams, A temporal
<br />pattern with average variation of intensities within
<br />the design burst can be formulated by using their
<br />method which, however, requires the recorded in-
<br />tense burst of a given duration.
<br />
<br />Huff (1967) has found, after analyzing data
<br />from two concentrated rain gage networks in central
<br />llIinois, a trend for the longer, heavier storms to
<br />dominate the fourth quarter of the storm period,
<br />whereas short-duration storms account for a major
<br />portion of the flrst and second quarters of the storm
<br />period. His classification of the storms according to
<br />whether the heaviest rainfall occurred in the first,
<br />second, third, or fourth quarter of the storm period
<br />as well as dimensionless representation of the time
<br />distribution minimizes the effects of mean rainfall,
<br />storm duration, and other storm factors on the
<br />variability in the time distribution. Huff (1970) has
<br />also found that time variability increases with de-
<br />creasing sampling area, the relative variablity (per-
<br />centage distribution) with respect to average rainfall
<br />intensity decreases with increasing intensity, and the
<br />absolute variability increases as the mean intensity
<br />increases.
<br />
<br />The skewness (y value) of a storm pattern varies
<br />greatly with numerous factors so that its accurate
<br />determination seems impractical, if not impossible.
<br />Various storm factors such as mean rainfall intensity
<br />and storm duration cause relatively large variations in
<br />the quartile distributions between storms, but no
<br />single parameter dictates the characteristics of the
<br />distribution (Huff, 1967). To select approximately
<br />the y value for a specifled frequency by using Huffs
<br />(1967) rainfall mass curves is possible, but his method
<br />is subjected to the quartile groupings of the storms
<br />recorded only in centrallllinois.
<br />
<br />Miller and Frederick's (1972) analysis of the
<br />sample of 1,484 stonns over the Ohio River Basin
<br />resulted in a typical time distribution of storm which
<br />contained two bursts with the smaller one near the
<br />beginning and the larger near the end of the long
<br />duration (4 to 10 days) in their study. They found
<br />that the number of bursts and time of occurrence
<br />within the storm were independent of geography,
<br />magnitude, and season. A similar study was also
<br />conducted by Frederick (1973) for storms over the
<br />Arkansas.Canadian River Basins. Miller and Frederick
<br />
<br />5
<br />
|