Maximum and minimum temperature for 14 stations in Western Virginia were used to
develop a temperature estimation model. Locational data such as the UTM coordinates,
elevation, aspect, distance from West Virginia and the distance to the coast and certain
transformations of these variables were used as the independent variables. A variable
was developed, called the distance-weighting variable, using the inverse distance to each
of the 5 closest weather stations. The dependent variables selected were mean monthly
maximum, mean monthly minimum, and mean monthly average temperatures. The
statistical method used was stepwise regression analysis, with the diagnostic tools of the
partial R2, the Cp, and the PRESS statistic being used as deciding factors for choosing a
subset of 5 to 6 models to study closely. The Variance Inflation Factor and the Variance
Proportion values were used to check for multicollinearity and to choose the final model.
The models developed here were compared to those in a study that was done in 1981
(Anderson 1981). Anderson (1981) also developed temperature equations, using data
obtained from the same general area, but using a larger data set of approximately 120
weather stations, and using only locational data. An independent data set of years other
than those used to develop the model were used to validate the models by estimating the
temperatures and comparing these estimates with the actual temperature values, using a
paired t-test. A 2-sided t-test was used to compare the actual temperatures with the
estimates calculated with Anderson's (1981) models and to compare Anderson's estimates
with the estimates calculated from the models developed in this study. The t-test generally showed that this study developed models that fit the data well and seemed to
predict well. In two cases where the model developed did not estimate well, reasons for
this departure from the normal were discussed and possible solutions proposed. I also
explored a new way of describing a temperature zone. A negative exponential equation
was developed for the potential absolute maximum temperature estimate for the area of
Western Virginia. A BASIC program was developed for managers to derive the
temperature estimates, either on a point-by-point basis, or for a file to be entered into a
geographic information system.
next to an author's name indicates that all files or directories associated with their ETD are accessible from the Virginia Tech campus network only.
![[VT]](http://scholar.lib.vt.edu/images/ETD-db/restricted.gif)
