Geoscience Reference
In-Depth Information
Fig. 9.8
Local
R
2
values for the GWR models of population density and water quality
have significant negative correlations with the urbanization level. In other words, the
relationships between population density and these water quality indicators decrease
as the urbanization of watershed increases.
A clear spatial pattern in the ability of PD to explain the variance in water quality
can also be identified from the maps of local
R
2
values from the GWR models
(Fig.
9.8
). Lower
R
2
values are mainly concentrated inside the 20-km buffer of
Boston, where PD explains low variance in the concentrations of most water qual-
ity indicators (Table
9.1
and Fig.
9.8
). In contrast, high
R
2
values are usually
observed outside the 20-km buffer. The
R
2
values generally increase as the sam-
pling sites move away from Boston metropolitan area (Fig.
9.8
). The highest
R
2
is close to 1 for some water quality parameters, such as SC, KN, Na, and SO
4
,
at some sites outside the 20-km buffer (Fig.
9.8
and Table
9.1
). Like the finding for
PDLU, this result indicates that population density is a better predictor to explain the
spatial variance in water quality indicators in less-urbanized areas than it in highly-
urbanized areas. Same degree of increase in population density may cause higher
increase in concentrations of water pollutants in less-urbanized suburban areas,
