MEASURED AND PREDICTED AIR TEMPERATURES AT BASIN TO REGIONAL SCALES IN THE SOUTHERN APPALACHIAN MOUNTAINS

Landscape and temporal patterns of temperature were observed for local (13 station) and regional (35 station) networks in the southern Appal achian Mountains of North America. Temperatures decreased with altitud e at mean rates of 7 degrees C/km (maximum temperature) and 3 degrees C/km (minimum temperature). Daily lapse rates depended on the method a nd stations used in the calculations. Average daily temperature ranges decreased as elevation increased, from 14 degrees C at 700 m to 7 deg rees C at 1440 m, and daily temperature ranges were typically higher i n spring and fall at any given station. Daily maximum temperatures abo ve the forest canopy averaged 1.4 degrees C higher at a south-facing s tation relative to a comparable northwest-facing station, and above-ca nopy daily minimum temperatures were depressed at a valley-bottom stat ion. Regional regression models provided a more accurate estimates of station temperature than either kriging or local lapse models when tes ted using 35 National Climatic Data Center (NCDC) stations in the sout hern Appalachians. Data-splitting tests yielded mean absolute errors ( MAE) from 1.39 to 2.30 degrees C for predictions of daily temperatures . Ten-year biases for an independent data set collected at four statio ns in the Coweeta Basin ranged from -2.87 to 2.91 degrees C for daily temperatures, with regional regression performing best, on average. Ho wever tests against another independent data set indicate regional reg ression and local lapse models were not significantly different, with mean biases averaged from -2.78 to 2.91 degrees C for daily predicted temperatures. (C) 1998 Elsevier Science B.V. All rights reserved.