Geographic Pattern Analysis of North Carolina Climate Division Data: 1895–2013

Introduction

The United States has observed an increase in mean annual temperature during the latter half of the 20th century, but multi-decadal periods of stasis and cooling have been embedded in the overall trend (Easterling and Wehner 2009; Meehl et al. 2009). Record-high temperatures have exceeded record-low temperatures in the United States by a two-to-one ratio (Meehl et al. 2009). Additionally, the number of cold days and nights has decreased as hot days and nights have increased (Field et al. 2012). Since 1950, both the number of extreme precipitation events and total precipitation amounts have increased (Peterson et al. 2008) and become more variable (Dulière et al. 2013), and regions of the Southeast have become cooler and wetter (Grundstein 2008; Zaitao el al. 2013). Drought conditions are expected to intensify for much of the American West (deBuys 2011), yet drought severity for the Great Plains (Hooerling 2012) and Southeast (Strezpek et al. 2010) are predicted to improve.

American mesoscale and macroscale climate analyses use a standardized set of climate divisions with monthly data beginning in 1895, although prior to 1931 missing data were estimated using least-squares regression (Guttman and Quayle 1996; McRoberts and Nielsen-Gammon 2011). Of the 334 United States climate divisions, eight divisions are located in North Carolina. Studies of climate in North Carolina have found that since the late 1970s, temperature has increased for western North Carolina leading to increased drought severity and frequency (Laseter et al. 2012). Up to 41 percent of southeastern U.S. droughts have been ended by tropical cyclone (TC) precipitation (Maxwell et al. 2012), yet future changes in TC frequency are inconclusive (Kuntson et al. 2010). Interannual and decadal El Niño and Pacific Decadal Oscillations have been found to affect hydro-logical processes in North Carolina (Anderson and Emanuel 2008). Boyles and Raman (2003) found increases in autumn and winter precipitation and summer and autumn minimum temperature as well as overall cooling of the Piedmont and northern mountain regions of North Carolina during 1949–1998, yet failed to report if such trends were significant.

This study observes significant seasonal trends in mean temperature, precipitation, and drought severity for the eight climate divisions in North Carolina from 1985–2013 to determine if changes for the three climate-parameters are congruent for all climate divisions. This research addresses three questions. First, do seasonal temperature, precipitation, and drought-severity means change temporally? Second, if so, are there geographic patterns to these changes? Third, if geographic patterns emerge do they covary between climate parameters?

Methods

All data were obtained from the State Climate Office of North Carolina (SCONC 2013). Monthly temperature, rainfall, and Palmer Drought Severity Index (PDSI) (Palmer 1965) means during 1895–2013 (full period) were acquired for each of the eight climate divisions and divided into four periods: winter = previous December–February, spring = March–May, summer = June–August, and autumn = September–November. Additionally, a state average was created by averaging all divisions. All data were converted to metric values. A reduced period from 1960–2013 was assigned to temperature and PDSI as initial analysis revealed a distinct rise in mean temperature and an increase in drought severity circa 1960 that persists through the remainder of the record. Similar time periods from the 1950s–1970s have been used as reference periods in climate studies (Giorgi et al. 2004; Mitchell et al. 2004; Hawkins and Sutton 2009; Coumou and Robinson 2013). Precipitation was excluded from the reduced period analysis as no distinct changes post 1960 were present.

Simple linear regression was performed for each climate division by season for the full and reduced periods. Linear regression provided least-squares estimates for annual climate parameter change as well as the state average, allowing for cumulative change for the full and reduced periods to be computed. Chloropleth maps with five equal interval bins to analyze geographic patterns in cumulative climate change for North Carolina climate divisions by season were produced. Additionally, Chow (Chow 1960) tests were performed for both temperature and PDSI to assess if structural breaks were present in the data. Data...