Trends in Extreme Temperature for July and January in North Carolina, 1949-1982

Southeastern Geographer Vol. 23, No. 2, November 1983. pp. 122-130 TRENDS IN EXTREME TEMPERATURE FOR JULY AND JANUARY IN NORTH CAROLINA, 1949-1982 Albert P. Barnett Attempts to identify world temperature trends have centered on analysis of mean monthly and seasonal temperatures (I), mean annual temperature (2), and to a lesser degree an indirect approach analyzing changes in growing seasons. (3) Another similar but undocumented approach is that of analyzing the July mean maximum temperatures and the January mean minimum temperatures. This study analyzes trends in the mean maximum temperature of the warmest month (July) and the mean minimum temperature of the coldest month (January) in North Carolina for the 34 year period 1949-1982. The selection of these years centers primarily on the quality of the data for the stations used in this analysis. The five stations chosen provide a representative cross-section of the varied physiographic divisions of the state. These stations include Wilmington, Raleigh-Durham, Greensboro, Charlotte, and Asheville. All are first-order National Weather Service stations (Table 1). Data utilized were made available from the HISARS data base. (4) The data records for the selected stations were considered complete. No station had more than one year in which data were missing. In such cases averages were computed from existing data. An attempt to test the homogeneity ofa station's data before and following a station's relocation was forestalled by insufficient comparative station data bases. By analyzing temperature extremes one not only interprets climatic alteration but concurrently assists in describing the environmental limits established for a given region. This in turn aids agronomists, entomologists, botanists, and geologists of the scientific community and many others in the society at large. METHODOLOGY'. Statistical analyses were performed for the five stations on the mean maximum temperature for the month of July and the mean minimum temperature for January. A time series was performed Dr. Barnett is Associate Professor of Geography at North Carolina Central University in Durham, NC 27707. Vol. XXIII, No. 2 123 TABLE 1 STATION LOCATION Station Latitude & Elevation Longitude(feet)(meters)County Asheville Charlotte Greensboro Raleigh/Durham Wilmington 35° 36' N 82° 32' W 35° 13' N 80° 56' W 36° 05' N '57' W 52' N 47' W 16' N 55' W 79° 35° 78° 34° 77° 2,242 735 897 434 28 673 220 269 130 Buncombe Mecklenburg Guilford Wake New Hanover to ascertain the general trend of temperature over time. A regression line (best fit line) was then computed for each station quantifying the temperature trend displayed in the time series. Each data set was analyzed using eight different regression equations (Table 2). The nonlinear equations were transformed to equations that are linear in parameters A and B (ex. Y = AeBX transforms to In Y = In A + BX). The trend line was determined by substituting the natural logarithms of the mean temperatures for the raw values. For example, ifa scatter diagram of In Y vs. X showed a linear relationship the equation had the form In Y = In A + BX, while if In Y vs. In X showed a linear relationship then In Y = In A + B In X was used. The A and B parameters were then transformed back to the original equation. Goodness of fit measures used by the procedure were R-square (an index of determination) and TABLE 2 SIMPLE REGRESSION EQUATIONS Original Equation Transformed Equation Restriction Y Y Y Y Y Y Y Y AX A + BX AeBX 1/(A + BX) A + B/X A + B log X AX» X/(A + BX) Unchanged Unchanged In Y = In A + BX 1/Y = A + BX Unchanged Unchanged In Y = In A + B In X 1/Y = A/X + B None None Y> 0 Y> 0 X> 0 X> 0 X, Y> 0 X, Y> 0 124Southeastern Geographer residual error (standard error of estimate). (5) The computations and graphics presented for the regression analysis were generated with the aid of a Tektronic 4051 microprocessor. The five-year running means of temperature were then computed and plotted on a time series. This yielded a line that was more indicative of the immediate variations surrounding a given instant in time and did not...