FLOOD08489 - Laserfiche WebLink

Home Browse Search

~ is the Espey Channelization factor cf is the watershed imperviousness factor cf = 1 - Ri Ri is the ratio of impervious watershed to pervious watershed. The regressiQn equation was tested using an indepen- dent set of data measured from Boneyard Creek at Urbana, Illinois. USE OF STATISTICAL METHODS The majority of the recent investigations on the effects of urbanization on flood hydrology have used some of the techniques of statistics dealing with cor- relation and regression. More effective techniques have been under development. Multivariate techniques are better suited to many problems in hydrology. Johnstone and Cross (1949) illustrated the application of correlation and regression in hydrology. Several examples of the test of significance are also given in this book. A more thorough treatment of the applica- tion of correlation to problems in hydrology was given by Beard (1962). Beard discussed the matters of mul- tiple correlation, nondetermination and criteria of statistical reliability. With the advent of the high speed digital com- puter, several improved procedures for complete multi- ple regression techniques evolved. These methods along with useful hints regarding their operation are given in a book by Draper and Smith (1966). Davis and Samp- son (1973) give and discuss a number of computer pro- grams written in Fortran for applying multiple regres.- sion and multivariate analysis. Reich (1962) utilized a stepwise multiple regression technique for selecting most effective regression equations between independent> and dependent hydrologic variables. Stepwi..6e. Mui..tipie. RegJLe.6.!o..i.on. -- In applying the stepwise multiple regression procedure, one independent variable is entered into the regression equation at a time and the coefficient of determination is found. The independent variable which yields the highest co- efficient of determination is selected. Following the selection of the initial independent variable, the re- maining independent variables are sequentially added to the regression equation and unexplained variance is computed. The independent variable which achieves the greatest reduction in the unexplained variance is the second independent variable added to the regression equation. The selection process is repeated -- each time a selection is made from the remaining independent variables until all of the independent variables have been added. In the practical case, all of the independent variables are seldom actually used in the operational regression equation. This is because of the problem of limited amoun~s of hydrologic data and the exces- sive costs involved in continuing the acquisition of large amounts of data. For this reason, early in the stepwise multiple regression analysis the matter of the "best regression equationll is considered. Draper and Smith (1966) suggest six general procedures in which this selection may be achieved: 1) All possible regressions, 2) Backward elimination, 3) Forward selection, 4) Stepwise regression, 5) Two variations on the four previous methods, 6) Stagewise regression. In general the fourth method, "Stepwise Regression" is the method used in the investigation reported herein. A computer program for completing these computations is available as a standard software package at the CSU Computer Center (STAT 38R-BDM02R revised). SIGNIFICANT PARAMETERS OF URBANIZATION There are many examples of a comparison of two or more photographs taken over intervals of time which graphically depict the evolution of an urban region. These comparisons witness to the fact that urbanization produces a profound change in the Watershed. The ex- tent of the changes caused by urbanization on the hy- drology of the watershed vary somewhat with local geo- logy, local customs, local laws, local climate and the intensity of the urban development. There is a need to be able to express ~he urbanization process quanti- tatively. Schulz (1971) listed eight measures which could be applied to quantify the urbanization. These were discussed in some detail. ) I P'-""en.t o. ImpeJw.coU1> W,ueMhed, 21 Le"gth o. S.tJt.ee.U Md Roado pel>. UrU.t a. Mea, 31 Length o. Paved S.tJt.ee.U pel>. urU.t 06 AlLea, 41 Length 06 CWLbed a"d Gu.t.tel>.ed S.tJLee.U pel>. UrU.t o. AlLea, 51 Le"gth 06 Stol11>1 Sewel>. Condu..i..t pel>. UrU.t o. Mea, 61 Ennec.Uve CMfme1. Ro~gh""M a. F.toodwa.y., 71 O.-Sill F.tood Ve.ten.tum StolW.ge, 81 Popu.f.aW" VeM.i.ty .c. W,ueMhed. Beginning ~ith this list of factors of urbaniza- tion, Lopez (1973) carried out a stepwise multiple re- gression analysis to select the most effective factors of urbanization. WMeJL6hed ImpeJtvlou.6nu.6. -- Referring to the re- sume of previous research work, the most obvious mea- sure of urbanization is the proportion of impervious watershed. This measure was listed by Schulz (1971) and used by Lopez (1973). The factor influences th~ hydrology in two ways: 1) Reduction of Inflitration, 2) Reduction of Response Time. Waananen (1969) shows two cases in different parts of the United States, where the water yield from a water- shed has been increased by urbanization. The explana- tion is that the paved and roof surfaced replace natural soil surfaces which in their former state have infiltered rainwater and in return lost water by evapotranspiration. 10