Viacom
December 19, 2002

On October 18, 2002, we sent you an analysis of storm volumes at Conard's Branch, Neal's Landfill. It has been determined that the statistical analysis in that report was in error. A revised report with a corrected statistical analysis is attached. Please disregard the previous report.

The corrected statistical analysis does moderate the conclusion that storm volume discharge has significantly decreased. It now appears that this is the case for smaller storms but perhaps not for larger storms.

If you have any questions on this analysis, please contact Mike McCann at (812) 355-0424.

Dorothy M. Alke
Director, Bloomington Project

Storm Volume Comparison at Conard's Branch, Neal's Landfill

Introduction

In the preliminary analysis distributed on September 11, 2002, a simple comparison was made between total rain and storm volume discharged for the 1993-94 storm set, the 2000 storm set, and the 2001 storm set. This analysis indicated a reduction in storm volume discharged for the post-remediation storm sets. However, the correlation coefficients were rather low and the linear regression equations were driven by a few large storms with much data scatter at the lower end of the spectrum. In order to strengthen confidence in the conclusion several additional steps were taken that form the basis of this analysis. They include:

• Incorporating the USGS data from 1983-84.
• Incorporating the spring treatment plant inflow records to account for all of the pre-storm flow.
• Updating the storm sets to account for the storms that have occurred since the weir was installed in Conard's branch. Using multiple regression analysis to account for antecedent moisture conditions as reflected by pre-storm flow and non-linear relations.

Storm Sets

The 1983-84 USGS data was entered into a spreadsheet so the hydrographs could be plotted and the storm volumes summed. The flows are summed from the beginning of the rising limb of the hydrograph to the point of intersection of linear and non-linear components on the receding limb of the hydrograph. This is the conventional point in hydrograph analysis which denotes the end of the storm (know as t_j in the discussion in Karst Geomorpholony and Hydrology. Ford and Williams, 1989, pages 193-203). These data would not have included North Spring flow, but would have included all flow from South Spring and the Overflow springs.

The records for the 1993-94 storm set included stage-discharge relationships for both South Spring and North Spring flow before those spring flows were collected and discharged to the treatment plant. These values were summed and included with flow in Conard's Branch to constitute the pre-storm flow.

A storm set was made from flow data collected after the September 26, 2001 installation of the weir on Conard's Branch until June 2002. These data are believed to be the most accurate. Inflow data from the spring treatment plant are included and added to the weir flow data prior to the storm onset to comprise the pre-storm flow. Table 1 lists the Conard's Branch storm sets used in this analysis. Back to back storms where the first storm does not recede to tj are combined as one event.

Regression Analysis

Table 2 shows the results of the multiple regression analysis using the STATISTICA computer program. The best correlation results were given with the natural log of the storm volume as the dependent variable (VAR1),the natural log of pre-stonn flow (VAR2) and total rain (VAR3) as the independent variables. The respective regression equations are written below the STATISTICA output tables for each storm set. Also included are the summary output tables of the regression analysis as calculated in Excel. The STATISTICA and Excel analysis are in agreement. The goal of eliminating data scatter was not particularly improved as the it-squared values ranged from 0.62 to 0.67. Table 3 shows for the largest storms of the data sets that the regression equation badly over predicts the amount. This is indicative of the lack of major storms that were measured and means comparison of the predictions of larger storms should be viewed skeptically.

Storm Volume Discharge Comparison

Inspection of Table 3 seems to indicate that volume reduction has occurred for post-remediation storms. A 2.45" storm in 1984 discharged 52.55 ac-ft while a 2.44" storm in 2001 discharged 50.10 ac-ft. A 2.85" storm in April 1994 discharged 65.09 ac-ft while a 3.06" storm in May 2002 discharged 44.49 ac-ft. The 3.63" storm of April 1984 discharged 98.38 ac-ft, the September 1993 storm of 4.00" discharged 94.49 ac-ft, while the largest storm of May 2002 of 3.51" only discharged 51.78 ac-ft.

Table 4 shows the predicted storm volumes based on the respective regression equations and an assumed pre-storm flow of 450 gpm. The table shows the 2001-2002 set consistently less than the 1983-84 set for storms less than 2.5 inches, and less than the 1993-94 set for storms less than 1.5 inches.

Conclusions