FLOOD08489 - Laserfiche WebLink

Home Browse Search

Chapler 4 PROCESSING BASIC DATA CSU SMALL WATERSHED FLOOD DATA FILE Beginning in 1962 Colorado State University set out to assemble high quality rainfall-runoff data for use in research on floods from small watersheds. Since 1962, the flood data file has evolved into a system for storing and retrieving the pertinent facts from magnetic tape. The data storage system is now entirely computer based. The basic data for the flood data file are assem- bled as a series of IBM cards as shown schematically on Fig. 2. Originally the flood data file was intended entirely for flood events from pristine watersheds. The data are prepared for magnetic tape storage as six sets of IBM cards. The arrangement is "open-endedll so that additional floods can be added at any time. Like- wise new watersheds with their flood events can also be added at any time. To adapt this storage system for urban flood events, two additional sets of data cards defining the extent of urbanization have been added. Each flood event from an urban watershed will also be accompanied by a set of cards to define the state of urbanization existing for that flood event. A detailed description of these sets of data cards are given in Appendix B. DERIVING A UNIT HYDROGRAPH Since the rainfall and runoff event is stored on magnetic tape and since the procedure of deriving a unit hydro graph from a set of rainfall-runoff observa- tions can be tedious hand or desk calculator operation, computer-based methods for deriving have been deve- loped. Six different computer-based methods were used with data stored in the CSU Flood Data File by Jawed (1973) . Lopez (1973) used three of these methods for deriving the unit hydrographs. The method which ob- tained the largest number of realizable unit hydro- graphs was the FINVER program which was developed by Kavvas (1972). The unit hydrographs and the recorded floods are given in Appendix C. Vdta Evatuation.--A preliminary evaluation of the available data was performed before any unit hydro- graphs were derived. The events chosen were primarily single peaked events. Complex events having a well defined peak several times larger than a secondary peak were chosen only when the volume under the second- ary peak was insignificant compared to the primary peak. Care was also exercised in choosing events with relatively dry antecendent moisture characteristics whenever possible. However, Since rainfall below 0.015 inch per five minute interval was not recorded it is possible that some events may have had wetter antecedent conditions than others. The volumes of rainfall and runoff were calculated and all events having recorded runoff in excess of the rainfall were discarded because this was an indication that the rain gage data did not correctly represent the causal rainfall. ~n6 06 Vata.-- The initial evaluation revealed several shortcomings of the data: a) The intervals in recording time are relatively large when the volumes of rainfall and runoff typi- cally found in the data are considered. Smaller intervals would be desirable in the determination of initial abstractions for instance, where the volume of rainfall prior to the beginning of runoff occurs some- time within the five minute interval. The error in this case is magnified by the fact that often a rain- fall duration of fifteen minutes is recorded which peaks during the first five minutes. As an example, an event registered at Hillcrest Drain, Northglenn, watershed area--0.28 square miles, on August 20, 1970, had a total abstraction of 0.073 inches and an initial abstraction of 0.09 inch. In this case the high inten- sity rainfall occurring in the first five minutes seem to not only satisfy the initial abstractions but also produces runoff. A malfunction in the instrumentation and nonuniform areal distribution of the storm could also account for this effect. The average rate of rainfall for the five minute interval also hides the actual time distribution of the event which is often necessary in deriving unit hydrographs of events of relatively small duration. b) In some events, the volume of runoff exceeded the volume of rainfall as in the event at Westerly Creek Tributary, Aurora, watershed area--0.20 square miles, August 19, 1971. A malfunction of the instru- mentation is possible, but a nonuniform areal distri- bution of the rainfall seems to be the more likely cause of this type of error. Even though the effect of nonuniform areal distribution is minimized in small basins, it should be kept in mind that it is possible. c) In order to investigate the effect of antece- dent conditions on the volume of rainfall excluded from the runoff, information concerning rainfall occur- rences prior to the reported event is necessary. d) In analyzing the effect of the volume of rain- fall on the response time of the basin, multiple events occurring continuously could be very useful. When rain falls interruptedly, the initial abstractions are minimized and the infiltration approaches a constant value. UNIT HYDROGRAPH PARAMETERS The rainfall runoff data were used to derive unit hydrographs. A five-minute unit hydrograph was ob- tained from each event. (The unit hydrographs are given in Appendix C.) The choice of the time interval was dictated by the available computer capabilities to invert large matrices. Of the chosen events, only one was too long to obtain a unit hydrograph. Another consideration in choosing the five-minute interval was the fact that the rainfall data were measured at this interval and could be used as given. The inter- polation of data often results in the unnecessary introduction of errors since one can only guess the possible time distribution of the rainfall within the recorded interval. 14