Laserfiche WebLink
<br />I <br /> <br />I <br /> <br />Chapter 1 <br /> <br />INTRODUCTION <br /> <br />I <br /> <br />Many efforts (19)* have been made towards <br />rationalizing flood flow estimation for very small <br />watersheds within the past five years. A large num- <br />ber of peak-flow estimates are still being made by <br />inadequate means. Firstly. there is a paucity of hy- <br />drographs observed on watersheds within the range <br />of one-third to four square miles. Secondly, the sub- <br />ject is confused by the many design methods cited in <br />handbooks without reference to their regional restric- <br />tion and limited basis. <br /> <br />I <br /> <br />I <br /> <br />I <br /> <br />The time seems appropriate to associate the <br />flood hydrograph with the causative rainstorm. The <br />intensity-frequency-duration regime for rainfall in <br />the United States of America has been thoroughly in- <br />vestigated by the Weather Bureau (11). Designers <br />should be provided with methods for estimating flood- <br />hydrographs with commensurate accuracy. <br /> <br />I <br /> <br />I <br /> <br />Objectives <br /> <br />The purpose of this study was to describe <br />flood hydrographs observed on small agricultural and <br />range land watersheds by a mathematical model with <br />three parameters. These hydrograph parameters <br />were then correlated to other numerical variables <br />which describe: the rainstorm causing each particu- <br />lar flood, the topographic characteristics of the <br />watershedt and the soil and current land use. If this <br />could be successfully achieved it would provide a <br />means of synthesizing flood hydrographs for ungaged <br />watersheds. The present work should be viewed as a <br />pilot study to test the validity of the approach on <br />readily available data. Its success should signal the <br />analysis of a large quantity of unpublished hydrologic <br />data. This pilot study could greatly reduce the ex- <br />pense of such a large -scale follow up by establishing <br />techniques and by showing which variables do not <br />w arr ant further study. <br /> <br />I <br /> <br />I <br /> <br />I <br /> <br />I <br /> <br />I <br /> <br />The multiple regressions obtained at the con- <br />clusion of this study only represent tentative relations. <br />Their reliability can best be judged by seeing how <br />well they predict results in a new set of observed <br />data. Truer regressions with narrower confidence <br />intervals will doubtless arise as more and more data <br />be co me available. <br /> <br />I <br /> <br />I <br /> <br />Delineations <br /> <br />The forty-seven hydrographs and hyetographs <br />used in this study were observed by the U. S. Agri- <br />cultural Research Service (23) on fourteen watersheds. <br />Arranged according to the code numbers given them <br /> <br />I <br /> <br />I <br /> <br />*Numbers in parenthesis refer to items in the <br />"Bibliography. II <br /> <br />I <br /> <br />I <br /> <br />by the Agricultural Research Service, all watersheds <br />are listed in Table 1. They were located in the <br />following eleven widely spaced states: Virginia, <br />Illinois, Iowa, Ohio, Wisconsin, Texas, Nebraska, <br />Arizona, New Mexico, Washington and Mississippi. <br /> <br />The largest watershed studied had an area of <br />2086 acres (3. 26 square miles). The smallest was <br />290 acres (0.45 square miles). Beyond this arbi- <br />trary size range, the next-smaller and next-larger <br />watersheds for which the A. R. S. had published data <br />were 187 acres (.292 square miles) and 4430 acres <br />(6.93 square miles). Thus one may generalize by <br />saying that the present study aimed at presenting re- <br />sults applicable to watersheds within the general size <br />range from one-third to four square miles. <br /> <br />Only floods caused by rainstorms were con- <br />sidered. This is justified on very small watersheds, <br />since the flood series (21) from these stations shows <br />the annual maxima to occur almost exclusively in the <br />summer months, and not to be due to snow melt. <br /> <br />Certain watersheds seem to produce a charac- <br />teristic double-peaked hydrograph. Although the <br />mathematical model fitted was single peaked, not all <br />the observed double peaks were excluded from the <br />study. The approach was rather to approximate the.m <br />with a broad-crested theoretical hydrograph whose <br />parameters could hopefully be related to the particu- <br />lar basin characteristics responsible for this unusual <br />watershed response (28, 29, 30). Only one hydro- <br />graph with clearly separated peaks was omitted from <br />the study. It was the result of a storm moving trans- <br />versely across a small belt of long twin tributaries. <br />The peak rate was considerably less than for the <br />three hydrographs actually used for this watershed. <br />From the point of view of predicting design floods its <br />omission seems desirable. Furthermore, uneven <br />areal distribution is an undesirable feature for this <br />study of very .small watersheds. <br /> <br />The basic approach was to use the total ob- <br />served runoff. That is to say. groundwater and <br />interflow were not subtracted as in the unit hydro- <br />graph attempt to isolate 11surface runoff. II A simple <br />adjustment was made to the observed hydrographs to <br />account for antecedent flow when this was present. <br /> <br />It should be stressed that the forty-seven <br />hydrographs originally selected from the available <br />fifty were retained for the remainder of the study. <br />The degree to which the mathematical model could <br />be fitted to the corrected "observedll hydrographs, <br />in no way affected their retention in the study. Like- <br />wise no further adjustments were made to the <br />corrected ltobservedl1 hydrographs in an attempt to <br />influence the regressions. Whatever deviations <br />