Confronting the Problem of Lead in Drinking Water: What Can and Should Be Done - Global Drinking Water Management and Conservation - page 235

used, as the first option given in the Health Canada Guidelines (Health Canada

2009 ). We utilize data from the 2006 to 2007 Customer Lead Pilot Testing project
in Ottawa and assume that the statistical properties for lead from this dataset are
representative of the samples for the rest of Ontario. The level of pH affects the
dissolution of lead from pipes (i.e. lead being

) into the drinking water
and since the pH levels in Ottawa are much higher than the pH levels for the rest of
Ontario, we need to

“ dissolved ”

the pH levels in the Ottawa data so that it is com-
parable to the rest of Ontario data. We do this by utilizing data from U-MATE
International, which shows the relationship between pH and lead holding all else
constant,

“ reduce ”

fitting a functional form to the data and using the functional form to
convert Ottawa data to the average pH levels observed in the Ontario data. After the
completion of this process, we should have comparable data, in terms of pH,
between the rest of Ontario and Ottawa. The next step is to

find out what lead
values for the Ontario sample would be like if the Ontario samples were taken using
the 6-h stagnation protocol. Our first approach is a benchmark model and is a

simple percentage change between the 30-min Stagnation and 6-h stagnation pro-
tocols, holding pH constant. Since the lead values used in the analysis occur at only
two points in time (30 min and 6 h), we assume that regardless of the functional
form of the rate of dissolution of lead in water pipes (whether it is linear or not), the
percentage change between the two data points is fixed. After the completion of this

process, we should obtain estimated lead values from the 30-min Stagnation pro-
tocol data from the rest of Ontario

“ converted ” to what lead levels would have been

had the the 6 h stagnation protocol been used.

11.3.2 The Estimation of Lead

For this subsection, we estimate outcomes for the lead values in the Ottawa data
that would re

ect the pH values in Ontario. The range of pH for Ottawa data (under

the project) was between

8.57 and 9.46 while the range of pH for Ontario data (under the

“ City of Ottawa Customer Lead Pilot Testing 2006 2007 ”

-

“ Ontario tap Water

Order for 36 Municipalities in 2006) was between 6.37 and 8.4. Therefore, we

need the two datasets to be comparable in terms of pH. We utilize data from U-
MATE International that shows the relationship between pH levels and lead (see
Fig. 11.1 ) and

”

fit a nonlinear functional form.
Figure 11.1 above shows a similar functional form between lead and alkalinity in
Schock s( 1989 ) analysis of temporal variability in domestic plumbing systems

(replica graph from Schock ( 1989 ) is shown in Fig. 11.2 ).
Our estimated functional form is that of a quadratic with the estimated equation
being: y ¼ 0

'

003x 2 3622, with an R-squared value of 0.9247, where

y is the lead value and x the pH value.
A

: 0 : 0643x þ 0 :

cant
difference both in terms of slope and R-squared values. We also did not use higher-
order polynomials (above order 3) since the interpretation of coef

fitted cubic functional form for the U-MATE data did not show any signi

cients in higher