7073 - Laserfiche WebLink

Home Browse Search

,. .~, Instreau Te~perature Model """DRAFT" " Page 6 March 27, 1989 was used to fill in missing bi-weekly averages for water temperatures at headwater and validation nodes. The simple multiple linear regression option was used here because the correlation coefficients were an improvement over the physical regression option. The temperature model filled in missing bi-weekly discharges with the mean for the respective time period. Regression statistics for water temperature are in Table 4. Relative humidity on a daily basis was unavailable from the National Oceanic and Atmospheric Administration. However, daily dew point temperature was obtained so that Relative humidity could be calculated using (Linsley, et al. 1975): HR - 112 - .1T + Tp 112 + .9r ,g f where HR Relative Humidity T ambient Air Temperature T„ Dew Point Temperature Several ancillary Fortran-77 programs were developed to format the data into required input files processed by the temperature model. All validation tables were requested when the model was first executed to compare predicted water temperatures with observed or regressed water temperatures at all validation nodes. RESULTS Validation Nodes Comparisons were made between predicted and observed (or regressed) mean daily water temperatures at all validation nodes for every two week period. Validation statistics produced by the model include the mean difference, the probable difference, and maximum error. The mean difference (~) is the average of the differences between observed (or regressed) water temperatures and those from the heat transport model. The probable difference (S) sets 50i; confidence limits for the predicted water temperatures. Thus, 50e of the actual water temperatures fall within ~ ± S of the model predic- tions. The explicit equations for ~ and S are: