Abstract
The diurnal variation of surface temperature in mountainous terrain was modeled using the first two harmonics of the Fourier series. The temperature T at any time t is given by
Each of the Fourier coefficients in this equation was considered a linear function of eight variables (X1, X2, ..., X8). The eight variables were 1) daily mean temperature=(maximum+minimum temperature)/2; 2) daily range of temperature=maximum−minimum temperature; 3) length of day=time of sunset−time of sunrise; 4) elevation of the station; 5) slope of the station; 6) u component of aspect of the station=sin (aspect); 7) v component of aspect of the station=cos(aspect); and 8) time of sunrise. The expanded equation contained 44 variables which were treated as 44 potential predictors in stepwise regression. The predictand was the surface temperature at each hour for 41 days (1 day week−1) during May-October 1975–76 at 10 remote stations in the San Bernardino Mountains of southern California. Five predictors were selected for the model of the diurnal variation of surface temperature in mountainous terrain. The model is given by
where T̂t is the predicted surface temperature at any time t (local time), X1 is the daily mean temperature, and X2 is the daily range of temperature. Because only the mean and range of temperature are involved, the model requires only the maximum and minimum temperatures as input. Using surface temperature data taken during May-October 1975–76 the root-mean-square errors (rmse) of model-predicted values for eight times per day at nine remote stations in the San Bernardino Mountains of Southern California ranged from 0.79 to 3.11°C; at the Blue Canyon Weather Station in the Sierra Nevada they ranged from 0.67 to 2.26°C.
