Information Technology Reference
In-Depth Information
represented as perennial streams on USGS topographic maps; therefore,
the Statewide database (sample) used in development of the equation may
not be random, because stream sites often selected for streamflow mea-
surements are represented as perennial streams on USGS topographic
maps. Also, the drainage area of stream sites selected for streamflow mea-
surements generally is greater than about 1.0 mi
2
, which may result in the
sample not being random.”
“The observed perennial or intermittent status of stream sites in the
South Coastal Basin database may also be biased, because the sites were
measured during the summer of 1999. The summer of 1999 did not meet
the definition of an extended drought; but monthly precipitation near the
South Coastal Basin was less than 50 percent of average in April, less than
25 percent of average in June, about 75 percent of average in July
(excluding one station), and about 50 percent of average in August
(excluding one station). Additionally, Socolow and others (2000) reported
streamflows and ground-water levels well below normal throughout most
of Massachusetts during the summer of 1999. Consequently, stream sites
classified as intermittent would have been omitted from the database had
this period been classified as an extended drought. This climatic condition
during the summer of 1999 could bias the logistic regression equation
toward a lower probability of a stream site being considered perennial in
the South Coastal Basin.”
“Basin characteristics of the stream sites used in the logistic equation
development are limited by the accuracy of the digital data layers used.
In the future, digital data layers (such as hydrography, surficial geology,
soils, DEMs, and land use) will be at lower scales, such as 1 : 5,000 or
1 : 25,000. This would improve the accuracy of the measured basin charac-
teristics used as explanatory variables to predict the probability of a stream
flowing perennially.”
“For this study, the area of stratified-drift deposits and consequently the
areal percentage of stratified-drift deposits included areas with sand and
gravel, large sand, fine-grained, and floodplain alluvium deposits. Future
studies would allow more specificity in testing the areal percentage of sur-
ficial deposits as explanatory variables. For example, the areal percentage
of sand and gravel deposits may be an important explanatory variable for
estimating the probability that a stream site is perennial. The accuracy of
the logistic regression equation also may be improved with the testing of
additional basin characteristics as explanatory variables. These explanatory
variables could include areal percentage of wetlands (forested and non-
forested), areal percentage of water bodies, areal percentage of forested
land, areal percentage of urban land, or mean, minimum, and maximum
basin elevation.”
