Trends in surface air temperature and temperature extremes in the Great Basin during the 20th century from ground-based observations

Authors


Abstract

[1] We analyzed trends in surface air temperature and temperature extremes in the Great Basin during 1901–2010. We found that annual average daily minimum temperature increased significantly (0.9 ± 0.2°C) during the study period, with daily maximum temperature increasing only slightly. The asymmetric increase in daily minimum and maximum temperature resulted in daily diurnal temperature range (DTR) decreasing significantly from 1901 to 2010. Seasonally, increases in daily minimum temperature and decreases in DTR occurred in winter, summer, and autumn, but the rate of increase was faster in winter. In contrast, daily maximum temperature showed no significant trend in any season. These trends in temperature measures, however, were not monotonic with decadal periods that included either reversal or acceleration of century-scale trends. The trend magnitudes in temperatures were not significantly associated with elevations. Increases in daily minimum temperature resulted in a decrease in the number of frost days (−0.14 ± 0.04 day yr−1) and cool nights (−0.09 ± 0.04 night yr−1) from 1901 to 2010, while the number of warm days (0.11 ± 0.04 day yr−1) and warm nights (0.19 ± 0.03 night yr−1) increased significantly. Surprisingly, the number of cool days and the length of the growing season showed no significant trend during the study period. Thus, the results of this study suggest that continuation of the overall warming trend would lead to markedly warmer conditions in upcoming decades.

1 Introduction

[2] Many studies during the past several decades have indicated that global land mean surface air temperature has trended upward in most parts of the world throughout the 20th century [e.g., Easterling et al., 1997; Alexander et al., 2006] or during the second half of the 20th century [e.g., Karl et al., 1993; Frich et al., 2002]. This increase in mean surface air temperature has been found largely to be due to more rapid increases in minimum temperature rather than to significant increases in maximum temperature [e.g., Easterling et al., 1997]. Consequently, these patterns have resulted in a narrowing of diurnal temperature range (DTR) in most parts of the world [Easterling et al., 1997]. Most studies also suggest that warming has been more rapid in winter than in summer [e.g., Stine et al., 2009]. In the many areas of the world experiencing these types of warming patterns, growing seasons generally have started earlier [e.g., Stine et al., 2009] and lasted longer [e.g., Myneni et al., 1997; Frich et al., 2002]. Despite the overall global warming trend, though, considerable spatial and temporal variability in warming has been observed during the last century [Zhang et al., 2000; Stine et al., 2009].

[3] A number of studies also have suggested that a large proportion of the world has been affected by significant changes in climatic extremes during the second half of the 20th century [e.g., Frich et al., 2002; Alexander et al., 2006; Vincent and Mekis, 2006] and that these climatic extremes can dramatically affect ecosystem processes for prolonged periods [e.g., Arnone et al., 2008]. Thus, compared to changes in mean condition, changes in frequency and severity of climate extremes caused by increased radiative forcing of the troposphere from continued anthropogenic greenhouse gas emissions [Easterling et al., 2000; Meehl et al., 2007] could have a more profound or serious effect on the natural environment and human society than increasing average temperature [Frich et al., 2002; Zhang et al., 2005; Vincent and Mekis, 2006]. For instance, temperature extremes can exacerbate drought conditions and increase the likelihood of fires as seen in the U.S. [Plummer et al., 1999; Meehl et al., 2007]. Therefore, timely analysis of climate trends and climate variation, especially extremes, becomes essential for providing guidance to policymakers to cope with climate-change-induced risk.

[4] Although studies on trends in temperature and temperature extremes at national or continental level included the Great Basin [e.g., Karl et al., 1996; DeGaetano and Allen, 2002; Peterson et al., 2008], detailed studies on trends in temperature and temperature extremes occurring in the Great Basin (Figure 1a) during the period of instrumental observation (e.g., 1901 to present) are necessary because this region experiences extremes of weather and climate that are not normally found elsewhere [Houghton et al., 1975]. Lying primarily in the rain shadow of the Sierra Nevada mountain range, the Great Basin is the driest region in the U.S. [Houghton et al., 1975]. One reason for the dearth of comprehensive, robust analysis of surface temperatures in the Great Basin over the last 100+ years is the sparseness of weather stations across this largely unpopulated land area (>500,000 km2) and the incompleteness of observational climate records at many locations. Nevertheless, available climate records in the Great Basin still can provide useful signals of historical climate variability and change that have occurred across this vast arid region.

Figure 1.

(a) The Great Basin defined in this study approximates the hydrographic Great Basin (http://water.usgs.gov/GIS/huc.html). (b) The 93 COOP stations that were selected for analysis in this study. The red points represent the 21 COOP stations in the early period of the 20th century. The square boxes of 1.34° × 1.34° in size in Figure 1b were used to provide a single value for a temperature index for the entire Great Basin. Panel (c) shows the number of stations at a given year with climate records available. Panel (d) shows the average elevation (the solid line) for all stations with climate records for a given year. The upper/bottom dashed line stands for mean elevation plus/minus one standard deviation of elevations of all available stations at a given year.

[5] Study of the trends in the Great Basin climate is also significant for the sustainability of ecosystems and human society in the region because surface climatic conditions prevailing across the Great Basin strongly modulate the hydrology and ecology of the many biomes that exist along latitudinal and elevational gradients in this region [e.g., Scanlon et al., 2005; Grayson, 2011]. Temperature in particular regulates ecosystem functioning by modulating rates of biogeochemical processes and other hydrologic fluxes that determine water availability [Arnone et al., 2008; Cayan et al., 2010]. To discern how the diverse ecosystems of the Great Basin may be responding to anthropogenic climate change, temperature trends occurring throughout the region during the last century need to be rigorously analyzed using long-term ground-based instrumental data.

[6] The objectives of this study were to use available observational climate records of daily minimum and maximum surface air temperatures from the Great Basin to (i) quantitatively examine trends in daily minimum, maximum, and mean surface air temperatures, and DTR during the period of instrumental record from 1901 to 2010; (ii) explore spatial variation and elevational difference in trend magnitudes of these climate indices across the Great Basin; and (iii) investigate trends in temperature extremes—such as the number of annual frost days and cool nights (Table 1)—in the Great Basin during the study period.

Table 1. Definitions of Five Measures of Temperature Extremes and Growing Season Length (GSL)
Temperature IndicesDefinitionsa
  1. a

    All definitions are applicable on a station by station basis.

Frost dayNumber of days with daily mean temperature below 0°C
Warm dayNumber of days with daily maximum temperature exceeding the 90th percentile threshold of all maximums in the base period (1961–1990)
Cool dayNumber of days with daily maximum temperature falling below the 10th percentile threshold of all maximums in the base period (1961–1990)
Warm nightNumber of days with daily minimum temperature exceeding the 90th percentile threshold of all minimums in the base period (1961–1990)
Cool nightNumber of days with daily minimum temperature falling below the 10th percentile threshold of all minimums in the base period (1961–1990)
GSLNumber of days in a year starting when the daily mean temperature on six consecutive days is above 5°C and ending after the daily mean temperature on six consecutive days is below 5°C.

2 Data and Methods

2.1 Data Sets and Preparation

[7] Given the sparseness and incompleteness of climate records in the Great Basin, we collected all available climate records from a total of 993 stations (Figure S1 in supporting materials) that are or were located within the hydrographic Great Basin. Initial analysis of monthly temperature records from these 993 stations indicated that only one station from the Cooperative Observer Program stations (COOP) data set had full coverage (i.e., without data missing) for the period of 1901–2010, and records were mostly available for the period 1961–1990. Therefore, we selected the period 1961–1990 as the base (reference) period. Using the base period, we selected four criteria to further define data. First, we tolerated a maximum of six years of data missing for any station in the base period. Second, selected stations had to have at least 24 years of records (80% of coverage) for each of the 12 months in the base period. Third, stations to be selected needed to have more than 40 years of records (36% of coverage) during the 1901–2010 period [e.g., Frich et al., 2002]. Fourth, stations that were/are located near (within 3 km of the boundary of) big cities (e.g., cities with population of more than 200,000 in the 1980s)—including Reno and Las Vegas (Nevada), and Salt Lake City (Utah)—were excluded to avoid possible confounding effects of urbanization on weather conditions in nearby stations [e.g., Alexander et al., 2006]. By adopting these criteria, the derived trends of temperature at each station for the period with records represented at least 80% of climate variability at that station. As a result, 93 stations were selected for further analysis, and all of these were COOP stations (Figure 1b).

[8] Although the 93 stations were not evenly distributed across the Great Basin, they adequately represented latitudinal and elevational spatial variability of climate in the study region (Figure 1b). The number of stations that had records in a given year also varied temporally during the study period (Figure 1c) and increased from 21 in 1901 to a maximum of 91 in 1971. Despite the low number, the 21 stations present in the early 20th century were spread across the entire Great Basin and reasonably represented the region's spatial variability (including latitudinal, longitudinal, and elevational) during that period (Figure 1b). Moreover, the average elevation for all sites that had sufficient climate records in a given year was consistent throughout the study period (Figure 1d). This greatly minimized the effects of differences in averaged elevation across years on single-averaged values of climate indices for the whole Great Basin.

2.2 Quantifying Trends in Temperature, DTR, and Temperature Extremes

[9] We collected the time series data of daily minimum and maximum temperatures for each of the 93 stations from the Western Regional Climate Center data archive (http://wrcc.dri.edu) and assessed data homogeneity for each station. Thus, for each station in the study period, daily records that exceeded the long-term mean of all available records from that station by four standard deviations or greater were manually checked or removed on a case-by-case basis [e.g., Zhang et al., 2005]. For each station, the number of days with extremely abnormal values is few (<5), and no full months and years were excluded. We also graphed and visually compared derived monthly series minimum and maximum temperatures at each station with those from neighboring stations to further check for data inhomogeneity [e.g., Peterson et al., 1998]. In general, we did not find any sharp discontinuities or gradual biases within each of the 93 stations that could be considered to be inhomogeneous that may have affected the reliability of our trend analysis.

[10] Statistics were first calculated over the base period (1961–1990) to provide a reference for assessing total time series anomalies. Daily mean temperature for each station and each day was calculated as the mean of recorded daily minimum and maximum temperatures, and daily DTR was calculated as the difference between recorded daily maximum and minimum temperatures. Due to the nonlinearity of daily temperatures, the calculated daily mean temperature should be considered as the best possible approximation of actual daily mean temperature in the absence of finer temporal measurements. Importantly, though, this measure of central tendency permits comparisons with data from other studies. Based on these daily values from the base period, we calculated monthly, seasonal, and annual means for each of the four temperature variables. When records were available, we also calculated anomalies from the mean for each of these temperature variables for each month, season, and year during the study period.

[11] To provide a single value for a quantity for the entire Basin, we divided the Basin into 1.34° × 1.34° boxes. The total number of boxes was 37, and most of them contained at least one weather station (Figure 1b). We then calculated box anomalies for each month, season, and year as the arithmetic mean of all available anomalies in the box. Afterward, we averaged, respectively, the box anomalies for each month, season, and year to compute their Basin-wide averages. The goal of this approach was to minimize effects of clustered stations on the Basin-wide averaged values for each month, season, and year [e.g., Zhai and Pan, 2003; Alexander et al., 2006].

[12] Following these procedures, we calculated time-series daily minimum, maximum, and mean temperatures during the entire record from 1901 to 2010. Similar to other studies [e.g., Plummer et al., 1999], we derived six additional temperature indices (Table 1) based on time-series daily minimum, maximum, and mean temperatures as well as base-period mean values. We used the nonparametric Kendall's tau (τ)-based slope estimator [e.g., Sen, 1968; Alexander et al., 2006] to compute trends for each climate index since this method does not assume a distribution for residuals and is insensitive to the effect of outliers in the series [e.g., Zhang et al., 2005; Alexander et al., 2006]. We also used a robust locally weighted regression and smoothing algorithm [Cleveland, 1979] to plot trends at different intervals in the study period.

[13] To analyze the spatial variation in trend magnitudes across the Great Basin, the period of 1961–2010 was selected because this period provided more complete records from 127 COOP stations (Figure S2 in supporting materials). The 127 COOP stations contained the 93 stations selected for the 110-year trend analysis. The addition of 34 stations helped minimize the limitation of sparseness of weather stations in the central and western Great Basin (Figure 1b vs. Figure S2) and thus enabled us to better compare the spatial variation in the trend magnitudes across the Great Basin. Besides, the addition of 34 stations did not produce any significant systematic bias of temperature records (see an example in Figure S3 in supporting materials). Following the same approach described before, we calculated the trend magnitudes of each of the four temperature variables for the 1961–2010 period and averaged the resultant trend magnitudes by 16 elevational zones of 200 m interval [e.g., Liu et al., 2009], ranging from 244 to 3260 m. The average trend magnitudes and corresponding average elevations were used to quantify the elevational dependences of trend magnitudes of the four temperature indices in the Great Basin.

3 Results

3.1 Trends in Annual Average Daily Temperatures and DTR in the Great Basin

[14] Annual average daily minimum temperature in the Great Basin averaged across the selected observation stations increased by a rate of 0.013 ± 0.002°C yr−1 (τ =0.42, p < 0.01), or 1.4 ± 0.2°C, during the period of 1901–2010 (Figure 2a). Annual average daily maximum temperature increased at only half of that (0.006 ± 0.002°C yr−1; τ =0.18, p < 0.01) during the same period (Figure 2b). Likewise, annual average daily mean surface air temperature increased by a rate of 0.009 ± 0.002°C yr−1 (τ =0.34, p < 0.01), or 1.0 ± 0.2°C during the 110 years (Figure 2c) largely as a result of the increase in daily minimum temperature. The more rapid increase in daily minimum temperature and lower increase in daily maximum temperature caused DTR to decrease by a rate of 0.007 ± 0.002°C yr−1 (τ =0.25, p < 0.01), or 0.8 ± 0.2°C during the 110-year observation period (Figure 2d).

Figure 2.

The general trends and variations in annual average daily (a) minimum (Min), (b) maximum (Max), (c) mean temperature, and (d) diurnal temperature range during 1901–2010 in the Great Basin. The plotted anomaly in a given year was relative to its long-term (1961–1990) annual mean. The solid red line is the fitted LOESS curve [Cleveland, 1979] with the smoother span 0.25.

[15] Although annual average daily minimum, maximum, and mean temperatures showed overall increasing trends from 1901 to 2010, increases were not monotonic during the 110 years. Decadal and multidecadal variability in the magnitude and sign of annual average daily temperature trends also were observed. For example, the first and last decade of the study period exhibited temperature decreases (Figures 2a–2c), and annual average daily DTR increased from 1940 to 1960 (Figure 2d). Further study suggested that the trend magnitudes of increases in annual average daily minimum and mean temperatures were greater in the last five decades (1961–2010) (Figure 3) compared to the overall trend magnitudes during 1901–2010 (Figure 2). For example, annual average daily minimum and mean temperature increased at a faster rate of 0.018 ± 0.005°C yr−1 (τ =0.35, p < 0.01) and 0.014 ± 0.004°C yr−1 (τ =0.31, p < 0.01), respectively, during the 1961–2010 period (Figures 3a and 3c). Annual average daily maximum temperature and DTR, however, had no statistically significant trends (p > 0.05) during the period of 1961–2010 (Figures 3b and 3d).

Figure 3.

The general trends and variations in annual average daily (a) minimum (Min), (b) maximum (Max), (c) mean temperature, and (d) diurnal temperature range during 1961–2010 in the Great Basin. The plotted anomaly in a given year was relative to its long-term (1961–1990) annual mean. The solid red line is the fitted LOESS curve.

3.2 Trends in Seasonal Average Daily Temperatures and DTR in the Great Basin

[16] When averaged for each season (each of four three-month sets), our results indicated that seasonal average daily minimum temperature increased in all seasons during the study period. Rates of increase ranged from 0.010 ± 0.003°C yr−1 (τ =0.19, p < 0.01) in spring to 0.014 ± 0.004°C yr−1 (τ =0.22, p < 0.01) in winter (Figures 4a–4d). In contrast, no significant trends (p ≥ 0.18) were observed in maximum temperature in any of the four seasons during the study period (Figures 4e–4h). Seasonal average daily mean temperature increased significantly by a rate of 0.011±0.004°C yr−1 (τ =0.17, p<0.01) in winter, 0.009±0.003°C yr−1 (τ =0.23, p<0.01) in summer, and 0.008±0.003°C yr−1 (τ =0.17, p<0.01) in autumn during 1901–2010. The increasing trend in seasonal average daily mean temperature, however, was not significant (at the 0.05 level) in spring (Figures 4i–4l). Increases in seasonal average daily minimum temperature and the absence of changes in seasonal average daily maximum temperature caused seasonal DTR to decrease in winter, summer, and autumn (Figures 4m, 4o, and 4p). This decrease ranged from −0.007±0.003°C yr−1 (τ =−0.21, p<0.00) in summer to −0.010±0.003°C yr−1 (τ =−0.20, p<0.01) in autumn. Overall, the upward trend in temperatures and the downward trend in DTR in the Great Basin occurred more rapidly in winter than in other seasons during the study period. In addition, the amplitudes of variations in seasonal average daily temperatures and DTR were much greater in winter and autumn than they were in summer (Figure 4).

Figure 4.

The general trends and variations in seasonal average daily minimum (Min), maximum (Max), mean surface air temperature, and diurnal temperature range during 1901–2010 in the Great Basin. The plotted anomaly in a given year was relative to its long-term (1961–1990) seasonal mean. The solid red line is the fitted LOESS curve. Winter—December, January, and February; Spring—March, April, and May; Summer—June, July, and August; Autumn—September, October, and November.

[17] Trends in seasonal average daily minimum, maximum, and mean temperatures, as well as in seasonal DTR, varied for different periods during the observation period (1901–2010). For example, seasonal average daily minimum temperature showed statistically significant increasing trend (p<0.01) in winter and autumn during 1901–2010, but such increasing trends were no longer significant (p ≥ 0.24) in the last five decades of the study period (Figure S4 in supporting materials). Besides, contrasting to the trend magnitudes observed for the whole study period, the trend magnitudes of increase in seasonal average daily minimum, maximum, and mean temperatures were much steeper in spring (> 0.031±0.013°C yr−1) and summer (> 0.021±0.008°C yr−1) during the 1961–2010 period (Figure 4 vs. Figure S4). Furthermore, although temperatures in spring increased significantly throughout the last five decades, they decreased markedly in the last two decades (1991–2010) (Figure S4). In terms of seasonal average daily DTR, the significant decreasing trends we observed for winter, summer, and autumn during the period of 1901–2010 were only applicable (p<0.02) to winter during the period of 1961–2010 (Figure S4).

3.3 Trends in Monthly Average Daily Temperatures and DTR in the Great Basin

[18] Monthly average daily minimum and maximum temperatures behave differently over the 110-year record. Monthly average daily minimum temperature showed statistically significant increasing trends in most months (except for April and December) as measured by a threshold of p ≤ 0.05 during the 110-year observation period (Table 2). Moreover, the increasing trends in monthly average daily minimum temperature were more pronounced and consistent in growing season months from May to September (p<0.01) relative to cold months (January, February, and November) (Table 2). Compared to daily minimum surface air temperature, monthly average daily maximum temperature in most months (except for September) showed no trend (p>0.10) during the 110-year observation period (Table 2).

Table 2. Trends in Monthly Average Daily Minimum, Maximum, and Mean Temperatures, as well as Trends in DTR, in the Study Region During the Period of 1901–2010 Using COOP Data
 MinimumMaximumMeanDTR
 Slope ≈p≤Slope ≈p≤Slope ≈p≤Slope ≈p≤
  1. Note: the unit for slope is in °C yr−1. The p-value is dimensionless.

January0.0190.020.0080.210.0130.030.0090.04
February0.0120.050.0100.130.0110.08−0.0020.68
March0.0120.010.0120.100.0120.040.0000.92
April0.0030.41−0.0040.490.0000.85−0.0070.12
May0.0130.010.0110.170.0120.04−0.0040.37
June0.0150.010.0080.210.0110.03−0.0070.09
July0.0130.010.0060.110.0090.010.0060.01
August0.0100.010.0040.290.0070.04−0.0060.06
September0.0170.010.0140.010.0150.01−0.0040.29
October0.0120.020.0050.510.0080.08−0.0070.18
November0.0100.05−0.0110.120.0000.940.0190.01
December0.0110.150.0010.830.0040.460.0140.01

[19] Increases in monthly average daily minimum temperature in the growing season and the lack of change in monthly average daily maximum temperature resulted in a significant increase (p<0.05) in monthly average daily mean temperature for growing season months such as months from May to September (Table 2). For January and March where monthly average daily minimum temperature increased significantly (≥ 0.012°C yr−1, p ≤ 0.02) during the period of 1901–2010, so did monthly average daily mean temperature (Table 2). Monthly average daily DTR showed no significant (p ≤ 0.05) trends in most months while it decreased significantly (≤ − 0.006°C yr−1, p<0.05) in January, July, November, and December (Table 2).

[20] Although monthly average minimum and mean temperatures increased for growing season months, these overall upward trends again were not monotonic at shorter temporal scales. For example, monthly average daily minimum, maximum, and average temperatures all increased significantly (p<0.03) in March and July during the period of 1961–2010 (Table S1 in supporting materials). Such increasing trends, however, were not applicable to other months when an overall significant increasing trend was observed during 1901–2010 with the exception of April and May for minimum temperatures. Also, compared to the overall trend magnitudes, monthly average daily temperatures in March and July increased at accelerated rates (>0.031°C yr−1, p<0.03) during 1961–2010 (Table S1 in supporting materials). Monthly average daily DTR no longer showed a significant decreasing trend (p ≥ 0.31) in July, November, and December during 1961–2010, compared to the overall decreasing trend observed for the period of 1901–2010 (Table 2 vs. Table S1 in supporting materials).

3.4 Spatial Variation in Trend Magnitudes of Temperatures and DTR in the Great Basin

[21] Our results based on a greater network of 127 stations (Figure S2) suggested that not all stations in the Great Basin showed uniformly increasing trends in annual average daily minimum, maximum, and mean temperatures, or decreasing trends in daily DTR, for the period of 1961–2010. Among 127 stations, 85 stations showed significant trends in minimum temperature, 58 stations showed significant trends in maximum temperature, 79 stations exhibited significant trends in mean temperature, and 84 stations exhibited significant trends in DTR. Stations that had significant trends were distributed all across the Great Basin, as were stations that showed no significant trends (Figure 5). There also were stations (e.g., green points for temperature and red points for DTR in Figure 5) that showed trends that were opposite to the trends calculated using all 127 stations during 1901–2010 [Figure 2 vs. Figure 5]. Analysis of data from 127 stations also suggested that trend magnitudes in annual average daily minimum, maximum, and mean temperatures, as well as DTR, were not significantly (p ≥ 0.17) correlated with elevation (Figure 6). Even for those stations showing significant trends during 1961–2010 (Figure 5), their trend magnitudes were not correlated with elevations either (data not shown).

Figure 5.

Red/green points stand for stations where annual average daily minimum (Min), maximum (Max), mean surface air temperature, and diurnal temperature range showed a statistically significant increasing/decreasing trend (p<0.05) during 1961–2010 in the study region. Black points stand for stations where no significant trends were observed during 1961–2010. The results shown here are based on 127 stations (Figure S2 in supporting materials).

Figure 6.

The relationship of trend magnitudes of annual average daily minimum (Min), maximum (Max), mean temperature, and diurnal temperature range with elevation. The plotted data were based on 127 stations (Figure S2 in supporting materials) for the period of 1961–2010.

3.5 Trends in Temperature Extremes and Growing Season Length in the Great Basin

[22] The number of frost days per year decreased during the period 1901–2010 in the Great Basin at about −0.14±0.04 day yr−1 (τ =−0.23, p<0.01) (Figure 7a). In other words, we estimated that the number of frost days decreased by 15 days (10%) with regard to its 110-years mean of 147 days during the study period. The number of warm days (Table 1) occurring per year increased significantly (0.11±0.04 day yr−1, p<0.01) during the period of 1901–2010. The number of cool days, however, showed no significant (p=0.82) trend in the study period (Figure 7c). Interestingly, the number of warm nights per year increased (0.19±0.03 night yr−1) while the number of cool nights per year decreased (−0.09±0.04 night yr−1) (Figures 7d and 7e). These patterns yielded no long-term significant changes (p=0.95) in growing season length (GSL) (Figure 7f).

Figure 7.

Trends in the number of frost days, warm days, cool days, warm nights, cool nights, and the length of growing season during 1901–2010 in the Great Basin. The solid red line is the fitted LOESS curve.

[23] Again, like trends in temperature indices, the overall trends in each of temperature extreme indices are not monotonic during the 110 years. For example, the increasing trend in the number of warm days occurring per year was accelerated (≈ 0.6±0.3 day yr−1, p ≤ 0.05) in last three decades starting around 1980s (Figures 7a and 7b). Similarly, the length of growing season in each year was extended significantly by about 1.9±0.03 day yr−1 (p<0.02) during the period 1970–1995, presumably due to the rapid anthropogenic warming of climate occurring in the last three decades.

4 Discussion

4.1 Trends in Annual Average Daily Temperatures and DTR

[24] The upward trends in temperature statistics, and the downward trend in daily DTR, we observed during the period of 1901–2010 in the Great Basin paralleled trends reported for many regions of the world [e.g., Zhang et al., 2000; Karoly and Braganza, 2005; Zhang et al., 2005]. The more rapid increase in annual average daily minimum temperature, relative to the rate of increase we calculated for annual average daily maximum temperature, led to a reduction of annual average daily DTR (Figures 2a, 2b, and 2d). This pattern has been reported for more than half of the land area in the Northern Hemisphere [Karl et al., 1993], globally [Easterling et al., 1997], and for the western United States [Pepin et al., 2011]. In addition, the absolute rate of decrease in daily DTR (−0.007°C yr−1) was equal to the rate of increase in daily mean temperature (0.007°C yr−1), a relationship that also has been reported for half of the land area in the Northern Hemisphere [Karl et al., 1993]. The small upward trend in annual average daily maximum temperature we calculated for the Great Basin concurred with results showing nonchanging or slight increases in the U.S. during the last century [Easterling et al., 1997]. At shorter temporal scale, our findings for the period of 1961–2010 [Figure 3] also agreed well with those reported for Australia [Karoly and Braganza, 2005] and the world [Alexander et al., 2006], in which significant increases in annual average daily maximum temperature were observed during 1951–2003. Such spatial similarity in the signs of temperature trends between spatially disjointed regions may be due to the uniform warming global climate resulting from anthropogenic greenhouse gas emissions since the late 1800s.

[25] The overall increasing trend we observed in average daily minimum temperature of 1.3±0.2°C per 100 years in the Great Basin, however, was 0.6°C less than the 1.86°C per 100 years reported for the world by Easterling et al. [1997]. The trend we observed (0.18±0.05°C increase per decade) for the period of 1961–2010 also was less than the increase of 0.21°C per decade calculated for the western United States by Pepin et al. [2011] for the period of 1948–2006. These contrasting temperature trends reported for different world regions suggest either true regional differences or artifacts generated by use of different analytical methods for estimating trend magnitude (e.g., linear regression vs. Kendall's tau), or differences in periods of observation [e.g., DeGaetano and Allen, 2002]. The uniqueness of Great Basin climate caused by its location in the rain shadow of the Sierra Nevada mountain range that makes it the driest region in the United States may account for some of the differences observed in the rates of temperature change relative to those seen in other regions of the world.

4.2 Trends in Seasonal and Monthly Average Daily Temperatures and DTR

[26] Increases in minimum temperature occurring in all seasons in the Great Basin during 1901–2010 (Figures 4a–4d) were similar to findings from earlier studies that suggested a globally warming trend in all seasons during the 20th century [e.g., Alexander et al., 2006]. In addition, the more dramatic increases in seasonal minimum temperatures we observed in the Great Basin, relative to the absence of statistically detectable (p>0.18; due to very high interannual variability in maximum daily temperatures, Figures 4e–4h) changes in maximum temperature, mirror global trends. The more significant and rapid increases in minimum temperature, and decreases in DTR, we observed during winter for the study period in the Great Basin, were consistent with results of Easterling et al. [1997], suggesting that the strongest changes in seasonal DTR in the Northern Hemisphere occurred in the winter season. Karl et al. [1993] suggested that decrease in DTR in the U.S. was found to be weak in spring but strong in autumn. Nevertheless, our observations of significant increases in temperatures and decreases in DTR in summer contrasted to results of Easterling et al. [1997], indicating that the smallest changes occurred in the summer season. Analyses of Alexander et al. [2006] suggest that, globally, spring showed the most significant change in minimum temperature while autumn showed the least change during the 20th century, which also contrasted to our observations that increase in minimum temperature in spring is smallest compared to other seasons.

[27] Regionally, Hamlet et al. [2007] found that temperature trends were more pronounced in the cool season (October–March) than in the warm season (April–September) during the period of 1916–2003 in the western U.S., which are somewhat contrary to our findings. Similarly, Gaffen and Ross [1999] reported that regional warming in the U.S. mainly occurred in winter, spring, and summer, while fall showed a less significant warming trend. These differences point to large spatial heterogeneity of seasonal trends in temperatures and DTR and highlight the need for careful regional studies to understand local changes in temperatures. Differences in regional land surface characteristics such as shifts in snow cover may produce regional signals that differ from continental or hemispheric trends in temperatures. For example, the accumulation of snowpack in February in the western U.S. can decrease regional warming due to an increase in surface albedo and consequent enhancement of the reflection of solar radiation. This would help explain the lack of a detectable warming trend in our study in February in the Great Basin. In contrast, a warming-induced melting of snowpack in March can reduce snow cover and surface albedo, which could enhance regional warming [Hall, 2004; Kapnick and Hall, 2012]. This would help explain the pronounced warming trend we observed for March in the Great Basin (Table 2), especially in the last five decades (Table S1 in supporting materials).

4.3 Spatial Variation in Trend Magnitudes of Temperatures and DTR

[28] The spatial variation of trends in temperatures and temperature indices we observed across the Great Basin suggested that factors other than the increased tropospheric radiative forcing from rising atmospheric CO2 concentration also affected local temperature regimes during the study period. Daly et al. [2010] and Pepin et al. [2011] suggest that local-scale processes, such as cold air drainage flow or the trapping of cold dense air masses by relief, can cause surface air temperatures to vary over space. In the western U.S., topography strongly modulates temperature [Hamlet et al., 2007]. Dettinger and Cayan [1995] have demonstrated that these meteorological and topographic factors can result in highly variable spatial patterns of temperature trends in the western U.S. They found that the upward trend in monthly temperature was less pronounced on the eastern slope of the Sierra Nevada during the period of 1948–1990 than on the western slope during the same period. Thus, it is possible that trends in temperatures and climate indices can differ from station to station across the Great Basin, as indicated in other studies [DeGaetano and Allen, 2002].

[29] Our finding indicating very weak relationships between the magnitude of temperature trends and elevation may partially be due to the relatively large spatial extent of the study region and also are subject to the temporal resolution of temperature indices (e.g., annual instead of seasonal). These findings, however, appear to agree with results from some studies. For example, Pepin et al. [2011] stated that, on a mean annual basis, they observed no significant relationships between the magnitude of temperature trends and elevation. Hamlet et al. [2007] also reported only limited observational data about differences in trends in temperature with elevation in the western U.S. An earlier study focusing on the Northern Hemisphere extratropical belt (30°N to 70°N) [Diaz and Bradley, 1997] indicated that changes in annual mean minimum temperature with height are equal to or greater than those near the surface.

[30] Nevertheless, our findings of weak relationships between trend magnitudes and elevations contrasted to other studies. Liu et al. [2009] indicated that monthly mean minimum temperatures increased more dramatically at higher elevations than at lower elevations in and around the Tibetan Plateau. Such an elevation dependency of climate warming might be a result of amplification of CO2-induced climate warming through cloud-radiation and snow-albedo feedbacks. Diaz and Bradley [1997] suggested that there was a clear tendency for maximum temperature trends to be smaller at stations above 2000 m compared to stations at lower elevations during 1951–1989 on the Northern Hemisphere extratropical belt. Weber et al. [1997] reported that mountain stations in the Italian Alps showed only small changes in DTR during 1901–1990, whereas low-lying stations in the western part of the Alps showed a significant decrease in DTR, caused by a strong increase in the minimum temperature. These differences in findings suggest high spatial variation in the relationships between temperatures/temperature indices and elevation [e.g., Diaz and Bradley, 1997] due to differences in regional climate (e.g., rainfall and snow cover) and local barometric pressure [e.g., Horton, 1995].

4.4 Trends in Temperature Extremes and GSL

[31] The downward trend in frost days and cool nights, and increasing trends in warm nights, in the Great Basin also occurred in other parts of the U.S. and world. Easterling et al. [2000] indicated that the number of frost days decreased slightly in the U.S. from 1910 to 1998. Alexander et al. [2006] reported that more than 70% of global land area showed a significant decrease in the annual occurrence of cold nights and a significant increase in the annual occurrence of warm nights throughout the 20th century. Vincent and Mekis [2006] found a significant decrease in the number of days with extreme cold temperature in southern Canada. In fact, nearly all earlier studies have reported significant decreases in the number of frost days and increase in the number of warm nights during the last century due to increases in daily minimum temperature [e.g., Frich et al., 2002]. Recent study [Peterson et al., 2008] indicated that the number of days exceeding the 90th percentile of daily minimum temperature (similar to warm nights defined in this study) in North America has been going up while the number of days below the 10th percentiles (similar to cool nights defined in this study) has been going down since the late 1960s. Such trends were most significant in the western North America (include the Great Basin) and consistent with our findings.

[32] Our findings of significant increasing trends in annual warm days in the Great Basin also seemed to agree well with findings from other studies. For example, Vincent and Mekis [2006] found that the number of days with extreme warm temperature showed a significant increase during the last century in southern Canada. Similarly, Peterson et al. [2008] found that the number of days exceeding the 90th percentiles of daily maximum temperature (similar to warm days defined in this study) has been going up in North America. Our findings of no significant trend in annual cool days in the Great Basin, however, seem to be unique. Alexander et al. [2006] found that, globally, the annual number of cold days decreased during 1951–2003. The little change in the number of cool days occurring each year during the 110-year observation period may have been due to the occurrence of only a slight overall upward trend in annual average daily maximum temperature (Figure 2b) while such an increasing trend was no longer significant in the last five decades (Figure 3b). In addition, differences in the study periods also can cause compared trends to differ among studies. For example, DeGaetano and Allen [2002] found that extreme warm minimum temperature (< the 5th percentile of daily maximum temperature) in the U.S. showed an overall decreasing trend during the period of 1930–1996 but an overall increasing trend during the period of 1960–1996. Again, the spatial heterogeneity of environmental conditions (e.g., snow cover, vegetation, atmospheric circulation, and rainfall pattern) can cause significant regional variations in the observed trends in temperature extremes [e.g., Easterling et al., 2000].

[33] The absence of any detectable changes in GSL in the Great Basin in our analysis contrasts ubiquitous observations of increasing GSL in other parts of the world, especially in the Northern Hemisphere midlatitudes [Frich et al., 2002]. At the global scale, Stine et al. [2009] found that the phase of the annual cycle of surface temperature over extra-tropical land areas shifted toward earlier in season by 1.7 days between 1954 and 2007. For the U.S., Myneni et al. [1997] reported that GSL was extended in the Great Lakes region during 1981–1991. A possible reason for the discrepancies in results may be that the increases in daily mean temperatures we observed mainly occurred in months from May to September in the Great Basin (Table 2). As a result, such increases had the least effect on GSL because daily mean temperatures in these months were already above the 5°C threshold. Myneni et al. [1997] stated that the extension of GSL mainly occurred in areas north of 45° N in the Northern Hemisphere, while the Great Basin is located south of 45° N. Further analysis based on 17 stations (with a maximum of 4 years of missing data allowed for each station) in the Great Basin still suggested no significant trend in GSL during the 1901–2010 period.

5 Summary

[34] Based on temperature records from 93 COOP stations in the Great Basin, we found the following to apply to the U.S. Great Basin from 1901 to 2010:

  1. [35] Annual average daily mean temperature in the Great Basin increased by 1.0°C during the 110-year period. The overall warming trend mainly resulted from a more rapid increase in daily minimum temperature and a lower increase in daily maximum temperature in the study period. Asymmetric changes in daily minimum and maximum temperatures caused daily DTR to decrease significantly by 0.8°C during the period of 1901–2010. These trends, though, were not monotonic with decadal periods that included either reversal or acceleration of century-scale trends.

  2. [36] The upward trend in daily minimum and mean temperatures, and downward trend in DTR, occurred in all seasons but more dramatically in winter during the period of 1901–2010. In contrast, monthly average daily maximum temperatures exhibited no significant trend in any season during the study period but increased significantly in spring and summer in the last five decades of 1901–2010. When averaged by month, increases in monthly average daily minimum and mean temperatures mainly occurred in the growing season (May to September) and between January and March. Monthly average daily DTR decreased mainly in cold months (November to January). Monthly average daily maximum temperatures showed no significant trends in most months during 1901–2010. Spatial and temporal variations in the strength (and sometimes direction) of trends occurred.

  3. [37] Increases in daily minimum temperature caused the number of frost days and cool nights per year to decrease, and the number of warm days and nights to increase during the period of 1901–2010 across the Great Basin. In contrast, the number of extreme cool days showed no significant trends in the study period partially because daily maximum temperature had only a slightly (0.006°C yr−1) upward trend. Although daily mean temperature showed an increase in the Great Basin, the extension of GSL in the region was not significant largely because the increases in daily mean temperature occurred predominantly in months from May to September, when temperatures were above the growing season temperature threshold (0.5°C).

[38] Thus, the results of our analysis in the Great Basin not only provided evidences of the 20th century global scale climate warming but also suggested that continuation of these overall trends would lead to markedly warmer conditions in coming decades that may be drier than in past decades if annual mean precipitation decreases in the southwestern U.S. as predicted by most current regional general circulation models (Christensen et al., 2007), albeit with a considerable uncertainty.

Acknowledgments

[39] This project was supported by NSF EPSCoR grant (NSF Cooperative Agreement EPS-0814372) for Nevada. We greatly appreciate the valuable comments from three anonymous reviewers. We thank Gregory McCurdy and Grant Kelly for their help in retrieving climate records from the U.S. West Regional Climate Center Data Archive.

Ancillary