Journal of Geophysical Research: Atmospheres

Variability of recent ground surface temperature changes in the Albuquerque basin, central New Mexico

Authors


Abstract

[1] Subsurface temperature measurements have been taken at four sites over a large area within the Albuquerque basin, central New Mexico. These data were taken in the unsaturated zone in order to avoid potential water movement in the saturated zone, so that a better estimate of surface temperature change might be realized. The data imply surface temperature change is not constant across the area. Unexpectedly, large upward water flow is suggested at one site having a very unique hydrogeologic environment. Different amounts of surface warming, likely caused by land cover change and encroaching urbanization, are estimated at two other sites. Data from a fourth site do not indicate a statistically significant surface temperature change, although the effects of urbanization and possibly increased evaporation in the past several years add ambiguity to the interpretation. These four sites indicate the potential variability of phenomena affecting subsurface temperatures in the Albuquerque basin.

1. Introduction

[2] Borehole temperature measurements have been used in many studies to indicate regional and global ground surface temperature warming over the past several centuries. Because of the coupling between air surface temperatures and ground surface temperatures these studies suggest climatic forcing warms ground surface temperatures. This analysis depends fundamentally on the condition that heat is being transferred only by conduction in the Earth's near subsurface. Whereas this is likely to be the case at many, if not most sites, there are a large number of temperature measured boreholes in regions where hydrogeologic and climatic conditions can cause considerable vertical as well as subhorizontal ground water flow, influencing the subsurface temperature profile.

[3] In this study I consider subsurface temperature measurements (T logs) across the vadose or unsaturated zone at four unique sites over an area of the order of ∼1000 km2 within the Albuquerque basin of central New Mexico. These data allow one to examine different T logs and consider possible perturbations to steady state heat conduction over a large local area. Past geothermal investigations in the Albuquerque basin at thirty sites over an area of the order of ∼3000 km2 have suggested considerable ground water flow in the saturated zone. I therefore began making T logs in the vadose zone several years ago because fluid flow in the unsaturated zone is generally thought to be very small (<1 mm a−1), therefore potentially permitting a better view of surface temperature changes. I briefly present previous results from two of the four sites in the study area and then consider new data from the other two sites. These data allow several interpretations of phenomena influencing temperature profiles in the area, present ambiguities in data interpretation, and suggest inverse forcing of local air surface temperatures from ground surface temperature increase.

2. Background

[4] Subsurface temperature measurements can provide valuable information regarding the history of the Earth's ground surface temperature. Because agreement between calculated ground surface temperature increase and meteorological data has been shown in a number of studies, ground surface temperature changes are likely to be important indicators of surface air temperature change [Chisholm and Chapman, 1992; Chapman and Harris, 1993; Harris and Chapman, 1995, 1997; Guillou-Frottier et al., 1998; Golovanova et al., 2001; Smerdon et al., 2004]. From subsurface temperature measurements, Lachenbruch and Marshall [1986] suggested 2 to 4°C warming of the permafrost surface over a period of a few decades to a century. Beltrami and Mareschal [1991] used borehole temperatures to suggest a 1 to 2°C warming in eastern Canada over the past century. Hundreds of subsurface temperature logs from four and six continents have been combined to suggest an average Earth surface warming of 1°C over the past five centuries, with a 0.5°C increase in the 20th century [Pollack et al., 1998; Huang et al., 2000]. A midlatitude warming of ∼1°C over the past 100 to 200 a is suggested by Harris and Chapman [2001]; and Majorowicz et al. [2006] indicated midwest North America ground surface temperatures have increased ∼2 ± 1°C over some two centuries. Interestingly, studies in the midcontinent of North America by Gosnold et al. [1997] suggest greater warming in the ground surface temperature histories in seasonally frozen ground is due to secular increase in soil moisture corresponding to greater precipitation over the last 50 a.

[5] Bullard [1939] noted that subsurface temperature profiles may be influenced by both ground surface temperature changes and subsurface fluid movement. Steady state conduction of heat within a depth interval of constant thermal conductivity will produce a linear temperature-depth profile. Curvature in borehole temperature profiles can be caused by both transient surface temperature changes and quasi-steady state ground water flow [Reiter, 2005]. In some areas it may be difficult to discriminate between the effects of transient diffusion due to ground surface temperature change and the quasi-steady state effects of vertical and/or horizontal ground water flow. Vertical only ground water flow less than ∼2 cm a−1 is difficult to recognize [Constantz et al., 2003; Reiter, 2003], and in such areas observed curvature in the T logs should represent ground surface temperature change [Harris and Chapman, 1995; Majorowicz et al., 2006]. Alternatively, borehole temperatures are frequently taken in areas of considerable topographic relief where an appreciable amount of both vertical and horizontal ground water flow is present and can therefore affect the profile of the temperature log. By curve matching appropriate expressions to temperature data, Reiter [2005] has statistically shown the relative importance of the different phenomena at two distinctly different geographical sites. Although more recent analysis of temperature profiles have employed sophisticated inversion techniques, first-order statistical considerations of possible influences to the temperature profile may be done by employing expressions for a surface temperature step and for vertical and/or horizontal ground water flow [Carslaw and Jaeger, 1959; Bredehoeft and Papadopulos, 1965; Stallman, 1963; Lu and Ge, 1996; Reiter, 2001, 2005].

[6] Interestingly a number of temperature logs used in the global analysis of ground surface temperature change show curvature compatible with surface cooling (e.g., ∼25% of the sites, [Pollack and Huang, 1998]). If such profiles are in fact caused by other than ground surface cooling, ground surface warming of large geographical areas might be somewhat underestimated. The variability of conditions between areas suggests that each T logged site be considered in terms of possible causes influencing the temperature profile. Ferguson et al. [2006] present hydraulic conductivity and gradient parameters that should be taken into account when considering the effects of vertical advection on the geothermal gradient. It appears likely that ground water flow is of second-order influence to T logs in the Great Plains and Prairies of North America [Majorowicz et al., 2006]; however, for much of mountainous western North America hydrogeological conditions have the potential to create vertical and horizontal hydraulic gradients that cause significant ground water flow in the saturated zone even if site recharge is quite small. For example, significant horizontal ground water flow resulting from mountain recharge is proposed across tens of km in the San Juan basin of New Mexico [McCord et al., 1992].

[7] Because of the potential for ground water flow to influence subsurface temperature profiles in the Albuquerque basin, a study of temperatures in the vadose zone of the basin was begun several years ago. In the desert southwestern U S vadose zones fluid fluxes are often small (typically less than 1 mm a−1; [Scanlon et al., 1997; Scanlon, 1999; Walvoord, 2002]) which should allow for any surface temperature change diffusing into to the Earth to be readily observed. The equipment necessary to make temperature measurements at closely spaced depth intervals within the air column of the vadose zone is described by Reiter [2004].

[8] A number of studies have also been concerned with ground surface temperature change resulting from man made changes of the land surface. At some data sites ground surface warming of up to 4°C is noticed and attributed to deforestation and/or urbanization [e.g., Lewis and Wang, 1992; Hayhoe and Tarnocai, 1993; Morin et al., 1999; Taniguchi et al., 1999; Reiter, 2006b]. Lewis and Wang [1998] point out that warming in Ontario over the past two centuries occurred when deforestation was taking place, and warming extended to regions surrounding the deforested areas. Skinner and Majorowicz [1999] discuss differences between ground surface temperature changes and air surface temperature changes, and suggest that land cover changes have contributed to a portion of the observed air surface temperature increase over the twentieth century from the boreal forests and Prairie Grasslands in Canada to the Great Plains in north Texas. Aquifer ground water temperatures beneath Winnipeg, Canada, have risen as much as 5°C in some areas because of warming buildings [Ferguson and Woodbury, 2004]. Hale et al. [2006] analyze hundreds of stations in the U.S. and conclude that significant warming has occurred at most stations associated with land conversion from forest to urban cover. In this note I will discuss the significant change of ground surface temperatures at some sites in the Albuquerque basin and the ambiguities associated with surface temperature change at other sites.

3. Data Presentation

3.1. Notes on the Study Area and Measurement Methods

[9] The four sites where vadose zone temperature measurements have been made in the Albuquerque basin are shown in Figure 1. These are piezometer nest sites where several different strings of 5.08 cm PVC casing have been put into a large drill hole and opened to the formation at different depths. The area is in the semiarid southwestern United States, mostly within the high desert grasslands, with the Chihuahua Desert at the southern edge. The Sandia, Manzanita, and Monzano Mountains boarder the basin to the east and the Colorado Plateau to the west. Precipitation is variable across the basin. For Albuquerque, the average annual rainfall from 1971 to 2000 was ∼24 cm; low and high average temperatures during January and July for this period where −4.6°C/8.7°C and 18.2°C/33.5°C respectively (http://www.rssweather.com).

Figure 1.

Location map for the study area and data sites (modified from Bartolino and Rankin [2001]).

[10] In order to make temperature logs with good depth resolution (1 m), the sensor must be lowered continuously into a well, otherwise the logging time requirement becomes great. This requires a sensor with a relatively fast time response in the media being logged. Reiter [2004] describes such a sensor for continuous temperature logging in air across deep vadose zones, as is presently the case. The sensor is a miniature thermistor having a time constant in air of 15 to 20 seconds. At 1 m depth intervals a digital multimeter sends the thermistor resistance measurement to be recorded on a laptop computer. Logging speeds begin at ∼0.75–1.0 m/min for the first 10–15 m where the yearly surface temperature cycle produces very large temperature gradients. Below ∼15 m logging speeds are ∼2.0–2.5 m/min. The system is calibrated with a platinum resistance in a stirred water bath. At the Tome site, six temperature logs measured over a two week period agreed to within 0.01°C at depths below 25 m [Reiter, 2004].

3.2. Review of Findings at Tome and Lincoln Middle School Sites

[11] One of the temperature logs from the Tome site is shown in Figure 2. Because of the very unique hydrogeology at the Tome site, an unexpectedly large vadose zone upward liquid water flow is predicted across a thick clay layer straddling the relatively shallow water table [Reiter, 2006a]. The curvature in the profile from ∼30 m to ∼58 m is best fitted by the expression for upward water movement, with horizontal water flow and ground surface cooling statistically second and third choices. The step function cooling prediction would be ∼1°C. These results, along with suggested warming of ∼0.4–0.8°C for the Colorado Plateau during the last century [Harris and Chapman, 1995] imply the curvature in the T log at the Tome site is likely caused by upward water flow even though the curvature is compatible with ground surface cooling.

Figure 2.

Example of one of the T logs at Tome [after Reiter, 2006a]. Fine-grained components across the vadose zone are also shown (modified from Jackson and Connell [1999]).

[12] An example temperature log for the Lincoln Middle School site is shown in Figure 3. The vadose zone temperature logs at Lincoln Middle School show profiles best fitted statistically by the complimentary error function describing the diffusion of a surface temperature step increase [Reiter, 2006b]. The estimated temperature step, ∼3.8°C, occurred about 20 a prior to measurements. Aerial photos of the site show natural vegetation removal and asphalt paving of two nearby roads occurred about 20 a before the temperature measurements. These observations suggest land cover change and urbanization as the cause of the surface temperature step.

Figure 3.

Example of one of the T logs at Lincoln Middle School (modified from Reiter [2006b]). Generalized lithologic description for vadose zone also given (from unpublished report, John Shoemaker and Associates, Incorporated, 4 August 1997).

3.3. 98th Street Site

[13] Profiles of the six temperature logs taken in the vadose zone over a 1-a (where a is years) period at the 98th Street site are presented in Figures 4a and 4b. Figure 4a shows the six overlaid temperature profiles from the surface. The variable temperatures near the surface indicate the yearly temperature cycle. At one wavelength depth of the yearly temperature cycle (∼19–20 m; [Carslaw and Jaeger, 1959, p. 66]) the surface temperature amplitude is reduced by a factor of exp(−2π) = 0.0019. Therefore in the Albuquerque area, the yearly temperature cycle at 19–20 m should be ∼0.03–0.04°C, (i.e., 0.0019 × 20°C, the approximate yearly air temperature cycle amplitude, http://www.rssweather.com). Figure 4b shows the six T logs from 20 m depth; they are offset so that one may better observe their characteristics by focusing attention along the plane of the figure. Irregularities in the profiles may result from the temperature sensor being caught on casing collars or the transducer wire while being lowered into the piezometer, as well as other causes not appreciated. Figure 4c indicates the temperature difference between the first log and subsequent logs. The first, second, and last T logs agree to within a few hundredths °C until 80 m depth, where the discontinuities in the T logs are noticed. The first and third T logs show a uniform offset of ∼0.08°C to 80 m. The other logs generally agree to within 0.05°C away from the erratic zones.

Figure 4a.

Composite of the six T logs at 98th Street. Generalized lithologic description also given for vadose zone (lithology given by Stone et al. [1998] for close-by core hole).

Figure 4b.

Offset T logs at 98th Street from 20 m to 117 m depth.

Figure 4c.

Temperature difference between T logs at 98th Street.

[14] In all the temperature logs a straight line fit (representing steady state heat conduction) to the data from 20 m depth to the bottom (117 m) is better statistically than the complimentary error function expression fit for diffusion of a surface temperature step (using the F statistic where F = mean square regression/mean square error, SSI Table Curve 2D). For example, considering the T log from 19 December 2003, over the interval 20 m to 117 m, the F statistic for the straight line fit is 158,614 where the F statistic for the complimentary error function fit is 62,011. These F statistics are used for a qualitative comparison of the fits because they are so large that both of the models are statistically significant at 0.99 confidence; that is, for the interval 20 m to 117 m the straight line fit to the data is probably better than the complimentary error function fit and therefore one cannot say that a surface temperature step has occurred. However, one may notice that all the T logs have a concave upward curvature approaching the surface (consistent with surface warming, Figure 4b). If I consider the depth interval 20 to 80 m for the first three T logs (because these intervals appear uninterrupted, Figure 4b), then the temperature step expression is about an order of magnitude better fit to the data than the straight line. Again, using the T log from 19 December 2003, over the interval 20 m to 80 m, the F statistic for the complimentary error function fit is now 451,466 where the F statistic for the straight line is 50,190. By using the shorter interval the F statistic results suggest a better data fit by the surface temperature step expression and therefore a surface temperature increase.

[15] Best-fit parameters for the first three T logs (20 m to 80 m) yield temperature steps of 1.72, 1.84, and 1.68°C occurring 3.34, 3.12, and 3.13 a before the logging took place (see Reiter [2006b] for a detailed explanation of how the expression best fit parameters are used to calculate the magnitude and timing of the temperature step). T logs four and five appear interrupted just below 50 m depth (Figure 4b), and therefore to achieve a much better F statistic fit of the temperature step expression, as opposed to a straight line fit, I consider the interval 20 to 53 m. The fitted step function parameters for T logs four and five yield a 1.16 and 1.26°C step occurring 6.1 and 8.5 a before logging. Lastly, the sixth T log is interrupted at 76 m; the best-fit temperature step expression to data from 20 to 75 m gives a value of 1.32°C occurring 4.5 a before logging.

[16] Although these six T logs present variable estimates of the surface warming parameters, the estimates are not comparable with the 0.4–0.8°C warming of the Colorado Plateau over the past century [Harris and Chapman, 1995]. An asphalt road was constructed about 15 m from the site in 1982 and the piezometer nest was completed in early 1997. From examination of available aerial photos there does not appear to be any major land surface change near the site coincident with the estimated step occurring about 2000 to 2001. I suggest that the close-by asphalt road with increased activity over the years, some destruction of natural vegetation at the site, the considerable recent development of the Albuquerque area in general, and large surface casing for the well, may all be contributing to the estimated surface temperature increase.

3.4. Mesa del Sol Site

[17] The last site to be considered is Mesa del Sol, the most remote and the least surface disturbed of the four sites in the study area. A composite of seven logs taken over a year and a half is shown in Figure 5a. Below the influence of the annual surface temperature cycle (∼20 m) it is noted that the logs appear to be quite consistent. The data are offset and shown in more detail from 20 m depth in Figure 5b; the consistent character of the logs may be observed by focusing attention along the logs in the plane of the illustration. The temperature difference between the first and subsequent T logs is shown in Figure 5c. Most of data agree to within a few hundredths of a °C.

Figure 5a.

Composite of seven T logs at Mesa del Sol. Generalized lithologic description for vadose zone also given (from unpublished report, John Shoemaker and Associates, Incorporated, 14 May 1997).

Figure 5b.

Offset T logs at Mesa del Sol from 20 m to 121 m depth.

Figure 5c.

Temperature difference between T logs at Mesa del Sol.

[18] Considering the entire depth interval for all the T logs in Figure 5b, the F statistic is about two times greater for the straight line characterizing steady state conduction than for the complimentary error function expression representing a surface temperature step. For example, given the T log from 8 August 2003, the F statistic is 298,700 for the straight line fit and 123,079 for the complimentary error function fit. Therefore this interval does not provide a basis for a surface temperature step. The same is true for the depth interval 30 m to the bottom of the piezometer.

[19] However, if I consider the interval 30 to 80 m, the F statistic becomes five to ten times greater for the temperature step expression fit than for the straight line fit (e.g., T log, 8 August 2003, F = 109,057 for the complimentary error function fit and 27,359 for the straight line fit). One chooses this depth interval because examination of the T logs in Figure 5b shows a subtle continuously concave upward curvature between ∼30 to 80 m, consistent with surface temperature warming. However it is difficult to justify excluding depths from 20 to 30 m (indeed if I consider the interval 20 to 80 m the F statistic is 1.2–1.8 greater for the straight line than for the temperature step expression). Perhaps some as yet unrecognized process is operating in the vadose zone. From the seven T logs very consistent temperature step parameters can be estimated for the 30 to 80 m interval: 0.98–1.08°C occurring 15.9–18.8 a before logging. These results are not consistent with regional trends (0.4–0.8°C increase in the Colorado Plateau over the past century [Harris and Chapman, 1995]) and little surface disturbance is noticed at the site. From area photos one may notice that development at a distance of a few km had taken place by 1967. This land disturbance along with the activity near the Albuquerque airport ∼7–8 km north of the site, and the considerable development of the city a few km to the west, may contribute to the possible estimated temperature step.

4. Discussion

[20] The four data sites in the Albuquerque basin do not appear to show a consistent aerial surface temperature change. Data from one site, Tome, suggest the influence of upward water movement in the vadose zone. Data from the Lincoln Middle School site demonstrate the significant effect that devegetation and urbanization can have on increasing the Earth's surface temperature. A first-order estimate can be calculated for the change in the rate of energy emission at the surface for the temperature increase of 3.8°C at the Lincoln Middle School site. From the law of black body radiation the radiancy at the surface is often written R = e*σ*T4 (where R is the radiancy of the outer surface, e is the emissivity, σ is the Stefan-Boltzmann constant, 5.67*10−8 W m−2 K−4, and T is temperature, K). Taking the emissivity to be constant, dR = e*σ*4*T3(dT), where e = 0.71 [Wielicki et al., 2005] and the black body temperature for the Earth = T = 254.3 K (http://mssdc.gsfs.nasa.gov/planetary/factsheet/earthfact.html). The calculation shows that the change in radiated energy corresponding to a 3.8°C rise in surface temperature is ∼10 W m−2. This energy flux is about 100 times the heat flux coming from the Earth's interior in the region of the Albuquerque basin.

[21] Although T logs at the 98th Street site show a less pronounced effect of a temperature step than those logs at Lincoln Middle School, a first-order estimate of likely surface temperature change is possible. If I consider the first three T logs from 20 to 80 m depth, above the depth of disturbances, I estimate a temperature increase of 1.7 to 1.8°C at a time 3.1 to 3.3 a before logging. I suggest for the 98th Street site this increase of temperature is influenced by both nearby and aerial land use change.

[22] The Mesa del Sol site is minimally disturbed for a drilling location, although the immediate area around the piezometer (∼5 m radius) has been devegetated during the drilling operation. If the entire depth of the T log is considered, steady state conduction in the vadose zone appears to be the statistically preferred environment (no surface temperature change); this is also the case for the depth interval 20 to 80 m. Figure 6 shows an FFT filter (Fast Fourier Transform filter) of annual average temperature and precipitation data taken at the Albuquerque airport (∼7–8 km north of the Mesa del Sol site) from about 1920 to 2005 (http://www.wrcc.dri.edu). The data are transformed to the frequency domain, where about 4/5ths of high frequencies are removed, and inverse transformed to the time domain. These data show a general increase in temperature of ∼1.4°C from 1986 to 2005, with a dip in the profile from 1995 to 2001. This information would be generally compatible with the increase in surface temperature predicted from the T log over the interval 30 to 80 m (∼1°C, ∼17.4 a before logging or ∼1987). However there is no robust reason to exclude the data from the 20 to 30 m interval, which in fact are curved in such a way to suggest surface cooling (Figure 5b). Could this cooling result from increased evaporation suggested by the smoothed precipitation increase over the past 3 or 4 a (Figure 6)? The data from the Mesa del Sol site are ambiguous and more study must be done to appreciate the effects of evaporation. There is presently conflicting evidence concerning ground surface warming.

Figure 6.

Fast Fourier Transform (FFT) filter of average annual precipitation and temperature at the Albuquerque Airport from about 1920 to 2005.

5. Conclusions

[23] The data shown in this note indicate the variability that is possible when using subsurface vadose zone temperature logs to estimate surface temperature change within the Albuquerque basin. The four sites where near surface vadose zone temperature measurement were made in the Albuquerque area do not correlate well with global or regional studies suggesting 0.4 to 0.8°C surface warming during the past century [Harris and Chapman, 1995; Pollack et al., 1998; Huang et al., 2000]. The sites appear to be differently influenced, in some cases by different phenomena. This study suggests several potential influences to near ground surface subsurface temperature profiles in the arid southwest U.S. vadose zones: water flow in unique hydrogeologic environments, land use change and urbanization, evaporation and evapotranspiration, and perhaps a disturbance of surface conditions at many drill sites.

Acknowledgments

[24] I thank Bonnie Reiter for helping gather the temperature data used in this study. The help and cooperation of the U.S. Geological Survey and the Office of the New Mexico State Engineer in providing access to the data sites is appreciated. Leo Gabaldon drafted Figure 1, Figures 2–6 drafted using SSI Sigma Plot, and the F statistics for expression fits to the data were done using SSI Table Curve 2D. The FFT filter of the average annual precipitation and temperature were done using SSI Table Curve 2D. The Earth Data Analysis Center at U.N.M. provided aerial photos of the sites used in this study. W. Haneberg made many helpful suggestions to improve the manuscript.

Ancillary