Observations of northern latitude ground-surface and surface-air

Observations of northern latitude ground-surface

and surface-air temperatures

Allan D. Woodbury†, AKM H. Bhuiyan‡ and John Hanesiak ‡

†

Department of Civil Engineering, University of Manitoba, Winnipeg, Manitoba, R3T

5V6, Canada.

‡ Center for Earth Observation and Science (CEOS), Faculty of Environment, Earth,

and Resources, University of Manitoba, Winnipeg, Manitoba, R3T 2N2, Canada

Abstract.

Since Lachenbruch and Marshall’s classic study of subsurface temperatures in

Alaska, there has been a great deal of interest in studying climate change through

reconstructing ground surface temperatures (GST) from borehole measurements

(BHT). Note that the magnitude of temperature increases reconstructed from BHT

records seems to contrast however, with some proxy based reconstructions of surface

air temperature (SAT) that indicate lower amounts of warming over the same period.

We present data suggesting that seasonal snowcover may bias climate reconstructions

based on BHT in portions of the Canadian northwest. Eight sites west of the Canadian

cordillera, were examined for long-term SAT and GST changes. At seven of these

sites precise borehole temperature profiles are used for the first time since the 1960s,

thereby exploring the linkage between GST and SAT. New readings were made at

four of these locations. All sites showed significant increasing SAT trends, in terms

of annual mean minimum and maximum temperatures. Over a 54 year period,

the minimum temperatures increased between 1.1o C and 1.5o while the maximum

increased between 0.8o C and 1.5o C, among those eight stations. Observations of

GST at those sites, however, showed no obvious climate induced perturbations. We

believe this disconnect between SAT and GST is attributable to an increase in snow

cover in early winter, followed by an increasing trend toward earlier snow melt in the

region. Such seasonal bias has important implications for GST reconstructions based

on borehole temperatures. These results support Mann and Schmidt’s conjecture about

a seasonal bias in the GST reconstructions from borehole surveys and counter assertions

of lower historic earth temperatures.

c©Allan D. Woodbury, Hassan Bhuiyan and John Hanesiak, 2007. Sumitted to Env.

Res. Let., December 2007.

† Corresponding address: E1-314 Engineering building, Department of Civil Engineering, University

of Manitoba, 15 Gillson Street, Winnipeg, Manitoba, R3T 5V6, Canada. ([email protected])

Northern latitude ground-surface-air temperatures 2

1. Introduction

The effects of climatic variations on subsurface temperatures have been known for some

time (Lane, 1923; Birch, 1948). However, it was not until Lachenbruch and Marshall’s

(1986) study of subsurface temperatures in Alaska was it realized that subsurface

temperatures could provide information on changes in climate over the last few centuries.

Since this time there has been a great deal of interest in using subsurface temperatures

to study climate change through reconstructing ground surface temperatures (GST;

for example, Harris and Chapman, 1997; Beltrami and Harris, 2001). Atmospheric

scientists are now beginning to explore the possibilities of using borehole temperatures

(BHT) as a source of data for the purposes of improving global circulation models, and

equally important for verifying the land surface schemes that are embedded in GCMs.

The GST reconstructions require the application of the theory of one-dimensional

heat conduction; the physics of which is well understood (Carslaw and Jaeger, 1959;

Jessop, 1990). However, the reconstructions are not the actual surface air temperatures

(SAT); the connection between them is a subject of active research (e.g. Schmidt et al.,

2001; Beltrami, 2001; Beltrami et al., 2005). Recent works (for example, see Beltrami

et al., 2005) have suggested that borehole GSTs are, in fact, robust reflections of past

climates.

Our understanding of the coupling between GSTs and surface air temperatures

(SAT) has improved. Subsurface temperature changes may result from factors other

than climate change and these are numerous: snow cover, changes in albedo, forest floor

organic layer thickness, and evapotranspiration, amongst other factors. For example,

studies have shown the effects on subsurface heat transfer from deforestation (Nitoiu

and Beltrami, 2005). Ferguson and Beltrami (2006) examined this issue in some detail

with numerical modeling. In their study, temperature anomalies associated with two-

dimensional (lateral) heat flow were evaluated for different scales of land-use. Changes

of land-use over large areas does indeed cause significant, but local changes in downhole

temperatures beyond the extent of the affected area. Their guidelines suggest that for

surface clearings of 500 m radius, borehole temperature measurements should be made

at least 60 m from the edge of an anomaly.

It is apparent that the issues related to deforestation are part of a larger issue of

how changes in the land surface affects sub-surface temperature and rates of climate

change predicted from BHTs. According to Beltrami et al (2003) and Majorowicz and

Safanda (2005) there were large areas of eastern and central Canada that experienced

warming in the 19th century but there were also large areas of Western Canada that had

no change or even cooled (based on borehole reconstructions). These authors conclude

that climatic warming is the main forcing, but land use changes from deforestation or

conversion of grasslands over to agricultural lands can affect the surface and therefore the

subsurface temperatures. The instrumental (SAT) record is clear though, and identifies

the greatest global warming occurring in the 20th century and in Canada this was

experienced in western Canada, but typically east of the Cordillera (Zhang et al., 2000).


Some have suggested that warming of the ground surface has not occurred. Lewis et al.

(2003) for example, concluded that

“ In the Canadian Cordillera universal warming at the ground surface due

to climate change over the past two centuries and subsequent propagation of

temperature anomalies down to 100 - 200 m depth has not occurred ... as it

has in central and eastern Canada.”

However, Lewis et al. (2003) cite the work of Lewis and Wang (1998) who focused

on disturbances in Vancouver Island, much further to the south of the northern sites

quoted in Lewis et al. (2003). Therefore, other than the boreholes themselves, there

were no other independent data to support the conjectures noted above.

In this paper, eight sites in north-western Canada, west of the Cordillera, are

examined for long-term SAT changes (Figure 1). The selected stations are: (a) White

Horse (60o 45.0’N, 135o 11.0’ W), (b) Ruby Creek (59o 42.8’ N,133o 24.1’ W), (c) Dease

Lake (58o 38.9’N,130o 0.6’W), (d) Hotailuh (58o 9.6’N,129o 51.9’W), (e) Gnat Lake

(58o 15.3’ N,129o 49.0’W), (f) Buckley Lake (57o 53.6’ N,130o 51.3’W), (g) Stikine

Site Z (58o 6.0’ N ,130o 16’W), and (r) Ritchie (56o 25.1’ N ,129o 0.92’W). At seven

of these sites precise temperature profiles observed in the 1960’s - 1970’s are used to

explore the linkage between GSTs and SATs. The borehole logs and their descriptions

were obtained from the the Geological Survey of Canada’s repository of temperature

recordings. We attempt to link the results of new readings at four sites taken in 2006

with those previously, and also to the instrumental atmospheric records for observation

and interpretation. The exciting part of this analysis is that at all of the sites we looked

at very little land use changes have occurred for over 25 years.

Long term trends in monthly average maximum and minimum temperatures are

investigated using the adjusted Canadian gridded (CANGRID) dataset (Zhang et al.,

2000) for the eight stations. Two different time series are considered (1950 to 1998 and

1950 to 2003) for long term trend analysis. Seasonal trends (Winter -Dec to Feb, Spring

- Mar to May, Summer - Jun to Aug, Autumn - Sep to Nov) are also analyzed using

same time series.

2. Atmospheric Temperature Trend Analysis

The CANGRID dataset contains monthly temperature and precipitation that covers

Canada. The CANGRID data are adjusted using surrounding stations to ensure

homogeneous time series and corrected for bias due to observational system change,

station relocation and so on. The data grid is a polar stereographic projection with a 50

km spatial resolution. The grid is a 125 (columns) by 95 (rows) matrix. Further details

of CANGRID data can be found in Zhang et al. (2000).

The nearest CANGRID data point to each station was used for the atmospheric

temperature trend analysis (see Zhang et al., 2000). The statistical model initially

removes the autocorrelation from the time series, then the slope (the linear regression


�

�

�

�

��

�

�

��

�

��

�

��

�

��

��

��

�

��

��

��

��

��

��

� �

� �

�!�

�!�

�"�

�"�

#�"�#�!�#$ �

#$��

#$��

#$��

#$��

#$"�

#$"�

#$!�

#$!�

#" �

#" �

#"��#"��

#"��%%%%%%%%

&'()*+,-

.///.//0//12345)6)78

Figure 1. North West Canada Observation sites

between climate variable and time) and its significance level are obtained from the de-

autocorrelated time series. The estimated slope is then used to remove the trend from

the original series and resulting residuals are used to obtain a more accurate estimation

of the (lag - 1) autocorrelation. To determine the statistical significance of the trend,

the confidence interval for the slope is obtained in terms of lower and upper bounds

based on the order of all possible slopes (Zhang et al., 2000).

During the study we used Zhang et al. (2000)’s model to calculate lower and

upper bound of slope, and Z-statistics for maximum and minimum temperature for

our selected stations extracted from the CANGRID time series (1950 to 2003), and

results are presented in Tables 1 to 5 and temperature plots corresponding to mean

annual maximum and minimum temperatures are shown in Figures 8 and 9. A linear

regression line is also plotted on each figure along with the terms in the equation.

All eight sites show a statistically significant increasing trend, > 95% in most

cases, in annual mean minimum and maximum temperatures (Table 1). Over the 54

year period, the increase in minimum temperatures has been between 1.1oC and 1.5o

and the increase in maximum temperatures has ranged between 0.8C and 1.5C over the

eight stations. All stations, except for White Horse, have larger increases in minimum

temperatures compared to maximum temperatures, which is consistent with Zhang et al

(2000). Tables 2 - 5 suggest that the annual mean trends are significantly influenced by

winter (DJF) and spring (MAM) temperature trends as opposed to summer (JJA) and

autumn (SON). Mean annual winter minimum temperature trends range between 2o

and 4o over the 54 year period, while mean annual winter maximum temperature trends

range over 1.9o to 4.2o, all being statistically significant to > 90% confidence. Once


again, the minimum temperature trends tend to be slightly larger than the maximum

temperature trends at most stations. The spring temperature trends (Table 3) are much

smaller than the winter trends (Table 2), however, spring trends still remain all positive

with most of them statistically significant at the 90% level. Tables 4 and 5 indicate that

summer and autumn do not have any statistically significant temperature trends since

all confidence levels are less than 80%, with most of them much lower than this (not

indicated in the tables).

3. Borehole Measurement Techniques

Temperature recordings were made using standard analog techniques, but with devices

specifically adapted for northern Canadian use. Many of the sites are only accessible

by helicopter. Temperatures were measured with a portable winch and a resistance

temperature detector (RTD). The temperature sensor is a 100 ohm OMEGATM platinum

resistance element in a 3 mm probe tip. The present implementation includes a 300 m,

5.5 mm diameter kevlar strengthened cable. We used a four wire connection to the RTD,

with the data taken with a specially designed readout box (essentially a Wheatstone

bridge). The probe is 1.9 cm outside diameter and has been used in 2.54 cm boreholes.

The overall accuracy ( ±0.01o C) is controlled by precise calibration in a thermal bath

and was confirmed by comparison to a SEABIRDTM digital-platinum RTD. Field data

were recorded the summer of 2006.

The earlier GSC data (and the sites) were described by Jessop et al. (1984).

Temperature recordings were made at various times, but typically in the mid 1960s

to early 1970’s. Temperatures for this earlier GSC work were also measured with a

portable winch and a thermistor sensor. Jessop et al (1984) report accuracies of ±0.02o

C. Detailed comments on each site, as originally described by Jessop et al. (1984), are

given below.

In the majority of the observed sites presented below synthetic temperature-depth

profiles are also plotted. These theoretical responses are calculated by assuming a series

of linear changes of the surface temperature. The pattern consisted of an average GST

increase between 1800 and 1900 of 0.44 C and 0.71 C from 1900 to 1950, followed by

another linear increase based on the SAT records at each site noted in Table I (see

Beltrami et al., 2003). The temperature at surface is assumed constant before 1800. At

each site, this synthetic GST history is used as the driving signal for computation of

the theoretical temperature at depth (at an appropriate time) and is combined with a

linear slope and intercept determined from linear regression of portions of each borehole

temperature record. A homogeneous value of thermal diffusivity of 1.2 × 10−6 m2/sec

was used. If the conceptual model of the earth response is correct; namely a purely 1-D

conductive environment, then the observed borehole temperatures should be reflective

of a subsequent temperature propagation from surface. If not, then some mechanism

must be responsible for removal of the energy that is available from an overall warming

at each site.


3.1. Whitehorse

Measurements were originally obtained in a mineral exploration hole drilled by

Whitehorse Copper Mines Ltd. about 5 km west of the city of Whitehorse. The

geology at the site consists of grandioritic batholith, and the hole was drilled 388 m

at an inclination of 60o. The area is well vegetated and, unfortunately, evidence of the

original drill site has been obliterated. Even after an extensive search, the borehole

was not found. The earlier GSC dataset was obtained in 1976 and is shown on Figure

2. A change in climate in this area over the last two centuries should have caused a

temperature perturbation down to 100-200 m depth in this hole, but this is obviously

not recorded. The data are almost a perfect straight line.

Figure 2. Whitehorse hole. GSC refers to the Geological Survey of Canada.


3.2. Ruby Creek

The Ruby creek site is about 27 km northeast of Atlin, B.C. and, by the time of this

writing is an active Molybdenum mine. The area is in an area of significant topographic

relief at high alpine elevations. Vegetation is sparse and typical of alpine environments.

The two original Ruby creek boreholes were logged to depths of 305 and 350 m in a

granitic batholith. Unfortunately, the original hole collars were destroyed by exploration

activities. We managed to log another more recent hole (48), quite close (about 100 m)

to the original holes identified by the GSC (see Figure 3). The water table at depth at

this site is not known. Jessop et al. (1984) believed that the thermal gradients in the

upper portion of the holes were disturbed by groundwater flow. A comparison between

the synthetic T −z plot and the observed data indicates that the warming of the ground

surface and subsequent propagation downward according to a purely conductive heat

transfer model has not occurred. Yet, we know the SAT signal at the surface shows a

linear increase of about 1.45 C over 54 years.

3.3. Dease Lake

This hole was originally drilled with the aim of collecting heat flow data. The borehole

was found to be in excellent shape in a well forested creek area. The site is within a

few km of Dease Lake itself. Little vegetative changes have been noted in the area since

the 1960s. The water table is at the surface, and the hole was logged without incident.

Results are show on Figure 4. The older GSC data are shown as black dots. Only the

top 200 m of the hole is plotted, and the results show excellent correspondence between

the various data sets. The upward trending curve above 5 m is suggestive of the yearly

thermal cycles and not of climate change.

3.4. Hotailuh/ Gnat Lake

The original hole (Hotailuh) was found intact and in excellent condition. It was

completed to 427 m into a granitic batholith in an area that has seen its share of

exploration activities. It is just a few km from the Gnat Lake site. As mentioned by

Jessop et al. (1984) portions of this hole were considered to be disturbed by groundwater

flows in fractured horizons (150 m and 350 m) within the granite. The results of recent

surveys and that of the past are shown on Figure 5. The water table is at the surface,

and the hole was logged without incident. The older GSC data are shown as black

rectangles. Only the top 200 m portion of the hole is plotted, and the results show

excellent correspondence between the various data sets over the entire borehole length.

No change in surface temperature is evident at this site between 1972 and 2006, a

situation that is noticeably different with the SAT record at the site. The upward

trending curve above 5 m is suggestive of the yearly thermal cycles and not climate

change. We also show the data recorded by Jessop et al. (1984) in 1972 for Gnat Lake.

Unfortunately, only a few data points are available at depths less than 40 m. We were


Figure 3. Ruby Creek (Atlin) hole. Data recorded at site but not in the same

borehole. Dots recorded in the older hole.

not able to find the hole and of course, log more data.

3.5. Buckley Lake/ Stikine Site Z

This site was drilled near the shore of a small lake just north of the Mount Edziza

Volcanic complex, which is a continuous volcanic terrain the extends some 60 km south

of Buckley Lake. The site itself is a partially cleared area in a heavily forested zone.

The hole is deep, some 427 m into predominantly quartzose rock. Only the first 200 m

of the record is plotted (Figure 6) and very close correspondence is seen between the

earlier records and that of the most recent survey. The water table in the hole is at 4.7

m, and the two shallow measurements were made in the air filled portion of the hole.


Figure 4. Dease Lake

Again, none of this data suggests external climate forcing.

An additional site in the area was analyzed for long term temperature records, and

this is detailed in an earlier section, but no temperature logs are available. North of the

Edziza complex is the grand canyon of the Stikine River, which is a deep, narrow canyon

that cuts through bands of layered volcanic and sedimentary rocks. In the 1980’s B.C.

Hydro was actively engaged in groundwork for two very large dams in the Grand Canyon

of the Stikine. These projects were later abandoned due to environmental concerns.


Figure 5. Hotailuh and Gnat Lake holes

3.6. Ritchie

Measurements were originally obtained in a large diameter (121 mm) petroleum

exploration hole to a depth of 1890 m by Dome Petroleum. The site is on the side

of a steep mountain (elevation 1100 m), along a tributary of the Bell Irving River, some

20 km east of Bowser Lake. Data from this hole originates from 1970 to 1972, and was

not surveyed in 2006 so we do not have more recent data to compare (see Figure 7).

Jessop et al. (1984) note that the temperature data had to be corrected for both climatic

changes and topographic relief in order to correctly determine heat flows. However, a

climate related GST signal is not plausible at this site and the observed profile is likely

affected by the steep topography at the site. The odd anomaly above 40 m is much to


Figure 6. Buckley Lake hole

deep to be related to an annual cycle, but the water table depth in the hole at the time

of the survey is not known.

4. Discussion

A review of Figures 1-6 shows that the known climate signal determined from the SAT

record has not propagated an anomalous temperature signal to depth as predicted by

a conceptual model of 1-D diffusive heat transport. Conversely any attempts at using

inverse methods that are based upon heat diffusion would, no doubt, predict a flat GST

history from 1800 to present, if based solely on the boreholes presented in this work.

Clearly, over a large area in northern British Columbia some other mechanism must


Figure 7. Ritchie hole

be responsible for removing the energy that should have been imported to the ground

surface in response to the SAT increases. As mentioned in the introduction, there is

a complex interplay between SAT and GST. It has been noted that the magnitude of

temperature increases reconstructed from BHT records seems to contrast with some

proxy based reconstructions of SAT that indicate less amounts of warming over the

same period (Zhang, 2005; and references therein). There can be many reasons for this,

but the influence of seasonal snow cover cannot be underestimated as a contributing

factor for modifying the climate signals reconstructed from BHT records.

According to Zhang (2005), snowmelt dates in Northern Alaska have advanced by

around 8 days since the 1960’s. Some modeling suggests that an advance of 10 days of

snow melt should lead to an increase in GST by 0.2 oC. However, Zhang notices that


a decrease in snow depth leads to a decrease in soil temperatures because of a lack of

snow insulating cover. Brown and Braaten (1998) report a snow depth decrease in much

of Canada over the period of 1946-1995. But, there has been about a 12 cm increase in

snow depth between 1946-1995 for each month December and January (total increase

of about 24 cm for the two months combined) for portions of northwestern British

Columbia (but nowhere else) and no change in snow depth for any other months for any

of our region. There has been no change in snow cover duration in our area of interest

except for spring (MAM) that shows about a 2 week decrease in snow cover duration

between 1946-1995 (49 years) - this essentially suggests that the snow disappears two

weeks earlier in some places compared to the mid/late-1940’s. Upon further reflection,

Zhang (2005) states that “The effect of early snow cover with late snow melt and and

late snow cover with early snow melt would change the mean annual ground surface

temperature by +2.0o and -1.5 oC, respectively”

So therefore, we conclude that a trend in our area towards late snow accumulation

and early thaw are masking an the increase in SAT temperatures. The snow cover

leads to a disconnect between SAT and GST over the autumn and winter months. The

important summer months, at least from the perspective of signal coupling, shows shows

a flat SAT response. Overall then, no (effective) increase in average climate is imported

into the ground.

Lastly Mann and Schmidt (2003), through coupling of GCM’s and land surface

schemes allowing for snow cover, show that SATs in North America closely track GST

only during the warm season. During the cold season an insulation effect is dominant

and this is supported by Zhang (2005). In our observational area, we have a trend in

warmer temperatures in the winter and spring, but flat or cool over the summer. This

may then explain why the boreholes do not reflect the GST, and the SAT is disconnected.

These results are important in that they support Mann and Schmidt’s conjecture about

a seasonal bias in the GST reconstructions from borehole surveys.

5. Conclusions

In this paper, eight sites in north-western Canada, west of the Cordillera, are examined

for long-term SAT changes. At seven of these sites precise temperature profiles are

used for the first time since (in some cases) the 1960’s in order to explore the linkage

between GSTs and SATs. The borehole logs and their descriptions were taken from

the the Geologic Survey of Canada’s repository of temperature recordings. We link the

results of new readings at four sites taken in 2006 with those previously, and also to the

instrumental atmospheric records for observation and interpretation. The exciting part

of this analysis is that in all cases we looked at little land use changes have occurred for

over 25 years.

All eight sites show a statistically significant increasing trend, > 95% in most

cases, in annual mean minimum and maximum temperatures (Table 1). Over the 54

year period, the increase in minimum temperatures has been between 1.1o and 1.5o and


the increase in maximum temperatures has ranged between 0.8C and 1.5C over the

eight stations. All stations, except for White Horse, have larger increases in minimum

temperatures compared to maximum temperatures, which is consistent with Zhang et

al. (2000). Tables 2 - 5 suggest that the annual mean trends are significantly influenced

by winter (DJF) and spring (MAM) temperature trends as opposed to summer (JJA)

and autumn (SON). Mean annual winter minimum temperature trends range between 2o

and 4o over the 54 year period, while mean annual winter maximum temperature trends

range over 1.9o to 4.2o, all being statistically significant to > 90% confidence. Once

again, the minimum temperature trends tend to be slightly larger than the maximum

temperature trends at most stations. The spring temperature trends (Table 3) are much

smaller than the winter trends (Table 2), however, spring trends still remain all positive

with most of them statistically significant at the 90% level. Tables 4 and 5 indicate that

summer and autumn do not have any statistically significant temperature trends since

all confidence levels are less than 80%, with most of them much lower than this (not

indicated in the tables).

This research shows that the known climate signal determined from the SAT record

over a large portion of north-west British Columbia and the southern Yukon has not

propagated an anomalous temperature signal to depth as predicted by a conceptual

model of 1-D diffusive heat transport. Conversely any attempts at using inverse methods

that are based upon heat diffusion would, no doubt, predict a flat GST history from

1800 to present, if based solely on the boreholes presented in this work. Clearly, some

other mechanism must be responsible for removing the energy that should have been

imported to the ground surface in response to the SAT increases. Our hypothesis is

that the observed disconnect between the SAT and GST signals at eight of our sites is

caused by an increase in snow cover in early winter, followed by an increasing trend to

an earlier snow melt in the region. A trend in our area towards late snow accumulation

and early thaw are masking the increase in SAT temperatures. The snow cover leads

to a disconnect between SAT and GST over the autumn and winter months. The

important summer months, at least from the perspective of signal coupling, shows a flat

or statistically insignificant SAT response.

There are limitations in our study. We were not able to obtain recent BHTs at

four of the eight sites. We do not have precise snow depths, and other detailed weather

records from any of the sites. The temperature trends at each site are determined from

statistical analyses of the CANGRID dataset and finally, no detailed modeling efforts

were carried out at any of the sites. Obviously, more work needs to be done but we

do believe that they support Mann and Schmidt’s (2003) conjecture about a seasonal

bias in the GST reconstructions from borehole surveys and counters assertions of lower

historic earth temperatures west of the Cordillera in northern British Columbia.


Table 1. Annual trends analysis (from 1950 - 2003) results of minimum and

maximum temperature (t-min/t-max) at different stations.

Station name oC per 54

years period

Z-Stats. Confidence

level (%)

(min/max) (min/max) (min/max)

White Horse 1.49/1.49 2.42/2.60 99.1/98.4

Ruby Creek 1.45/1.42 2.32/2.50 98.7/97.9

Dease Lake 1.26/1.07 2.16/2.42 98.4/96.9

Grant Lake 1.22/1.02 2.25/2.36 98.1/97.5

Buckley Lake 1.14/0.92 2.35/2.31 97.9/98.1

Stikine Site Z 1.13/0.88 2.23/2.26 97.6/97.4

Ritchie 1.06/0.79 2.09/1.68 90.7/96.3

Table 2. Winter (Dec to Feb) trends analysis (from 1950 - 2003) results of

minimum and maximum temperature (t-min/t-max) at different stations.


years period

Z-Stats. Confidence

level (%)


White Horse 4.03/4.24 1.78/2.02 92.5/95.7

Ruby Creek 3.86/3.73 1.84/2.02 93.4/95.7

Dease Lake 3.24/3.25 1.84/1.74 93.4/91.8

Grant Lake 3.06/2.92 1.87/1.81 93.9/92.9

Buckley Lake 2.84/2.59 1.79/1.98 92.7/95.2

Stikine Site Z 2.82/2.64 1.87/1.96 93.9/95.0

Ritchie 2.05/1.91 1.81/2.05 92.9/95.9

Table 3. Spring (Mar to May) trends analysis (from 1950 - 2003) results of



years period

Z-Stats. Confidence

level (%)


White Horse 1.73/2.16 2.14/2.05 96.8/95.9

Ruby Creek 1.68/1.88 2.01/1.92 95.4/94.5

Dease Lake 1.66/1.69 2.08/1.83 96.2/93.3

Grant Lake 1.65/1.67 2.08/1.72 96.2/91.5

Buckley Lake 1.55/1.29 2.23/1.86 97.4/93.7

Stikine Site Z 1.52/1.49 2.36/1.86 98.2/93.7

Ritchie 1.19/1.29 1.92/1.59 94.4/88.8


Table 4. Summer (Jun to Aug) trends analysis (from 1950 - 2003) results of



years period

Z-Stats. Confidence

level (%)


White Horse 0.68/0.51 1.23/0.77 -/78.1

Ruby Creek 0.56/0.42 1.01/0.63 -/68.3

Dease Lake 0.33/-0.17 0.63/-0.28 -/-

Grant Lake 0.28/-0.22 0.44/-0.45 -/-

Buckley Lake 0.35/-0.28 0.57/-0.51 -/-

Stikine Site Z 0.29/-0.30 0.47/-0.59 -/-

Ritchie 0.31/-0.17 0.81/-0.36 -/-

Table 5. Autumn (Sep to Nov) trends analysis (from 1950 - 2003) results of



years period

Z-Stats. Confidence

level (%)


White Horse 0.01/-0.22 0.00/-0.24 -/-

Ruby Creek 0.05/-0.14 0.01/-0.21 -/-

Dease Lake 0.00/-0.16 0.00/-0.25 -/-

Grant Lake 0.11/-0.04 0.10/-0.01 -/-

Buckley Lake 0.23/0.09 0.44/0.01 -/-

Stikine Site Z 0.23/0.01 0.33/0.00 -/-

Ritchie 0.46/0.38 0.93/0.82 -/-

[] Beltrami, H. (2001), On the relationship between ground temperature histories and meteorological

records: A report on the Pomquet station, Global and Planetary Change, 29(3-4), 327-348.

[] Beltrami, H. and R.N. Harris (2001), Foreword: Inference of climate change from geothermal data,

Global and Planetary Change, 29(3-4), 149-152.

[] Beltrami, H., Ferguson, G. and R. Harris (2005), Long-term tracking of climate change by

underground temperatures, Geop. Res. Lett., 32 (L19707), doi:10.1029/2005GL023714.

[] Birch, A.F. (1948), The effects of Pleistocene climatic variations upon geothermal gradients,

American Journal of Science, 246(12), 729-760.

[] Brown, R.D. and R.O. Braaten (1998), Spatial and temporal variability of Canadian monthly snow

depths, 1946-1995, Atmosphere-Ocean, 36, 37-54.

[] Carslaw and Jaeger (1959), Conduction of Heat in Solids, Oxford Univ. Press, 386 pp.

[] Ferguson, G. and Beltrami, H. (2006), Transient Lateral Heat Flow due to Land-Use Changes, Earth

and Planetary Science Letters, 242, 217-222.

[] Harris, R.N., and D.S. Chapman, (1997), Borehole temperatures and a baseline for 20th century

global warming estimates, Science, 275, 1618-1621.

[] Jessop, A.M. (1990), Thermal Geophysics, Elsevier Science, New York, 316p.


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−3.5

−3

−2.5

−2

−1.5

−1

−0.5

0

0.5

1

1.5

White Horse

Time (Year)

Mea

n a

nnual

max

imum

tem

per

ature

(oC

) Equation of the line: Y = 0.030022X −1.6917

1950 1956 1962 1968 1974 1980 1986 1992 1998 20032.5

3

3.5

4

4.5

5

5.5

6

6.5

7

Ruby Creek

Time (Year)

Mea

n a

nn

ual

max

imu

m t

emp

erat

ure

(oC

) Equation of the line: Y = 0.027482X + 3.9192

1950 1956 1962 1968 1974 1980 1986 1992 1998 20033.5

4

4.5

5

5.5

6

6.5

7

7.5

8

Dease Lake

Time (Year)

Mea

n a

nn

ual

max

imu

m t

emp

erat

ure

(oC

)

Equation of the line: Y = 0.022003X + 4.8755

1950 1956 1962 1968 1974 1980 1986 1992 1998 20033

3.5

4

4.5

5

5.5

6

6.5

7

7.5

8

Grant Lake

Time (Year)

Mea

n a

nn

ual

max

imu

m t

emp

erat

ure

(oC

)


1950 1956 1962 1968 1974 1980 1986 1992 1998 20031.5

2

2.5

3

3.5

4

4.5

5

5.5

6

Buckley Lake

Time (Year)

Mea

n a

nn

ual

max

imu

m t

emp

erat

ure

(oC

)


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−0.5

0

0.5

1

1.5

2

2.5

3

3.5

Stikine Site Z

Time (Year)

Mea

n a

nnual

max

imum

tem

per

ature

(oC

)


1950 1956 1962 1968 1974 1980 1986 1992 1998 20031

1.5

2

2.5

3

3.5

4

4.5

5

Ritchie

Time (Year)

Mea

n a

nn

ual

max

imu

m t

emp

erat

ure

(oC

)


Figure 8. Mean annual maximum temperature (oC) trend for seven stations [White

Horse, Ruby Creek, Dease Lake, Grant Lake, Buckley Lake, Stikine Site Z, and Ritchie].


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−13

−12

−11

−10

−9

−8

−7

White Horse

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC

)Equation of the line: Y = 0.02933X −10.3092

1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−8

−7.5

−7

−6.5

−6

−5.5

−5

−4.5

−4

−3.5

−3

Ruby Creek

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−8

−7.5

−7

−6.5

−6

−5.5

−5

−4.5

−4

−3.5

Dease Lake

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC

)

Equation of the line: Y = 0.025775X −6.4836

1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−8

−7.5

−7

−6.5

−6

−5.5

−5

−4.5

−4

Grant Lake

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−7

−6.5

−6

−5.5

−5

−4.5

−4

−3.5

−3

Buckley Lake

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC

)


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−9.5

−9

−8.5

−8

−7.5

−7

−6.5

−6

−5.5

Stikine Site Z

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC

)


1950 1956 1962 1968 1974 1980 1986 1992 1998 2003−6.5

−6

−5.5

−5

−4.5

−4

−3.5

−3

−2.5

Ritchie

Time (Year)

Mea

n a

nnual

min

imum

tem

per

ature

(oC

)


Figure 9. Mean annual minimum temperature (oC) trend for seven stations [White

Horse, Ruby Creek, Dease Lake, Grant Lake, Buckley Lake, Stikine Site Z, and Ritchie].


[] Lachenbruch, A.H. and B.V. Marshall (1986), Changing Climate: Geothermal evidence from

permafrost in the Alaskan Arctic, Science, 234, 689-696.

[] Lewis, T.J., Hyndman, R.D., and P. Fluck (2003), Heat flow, heat generation and crustal

temperatures in the northern Canadian Cordillera: Thermal control of tectonics, J. Geop. Res.

108(B6), doi:10.1029/2002JB002090.

[] Lewis, T.J. and K. Wang (1998), Geothermal evidence for deforestation induced warming:

Implications for the climatic impact of land development, Geop. Res. Lett. 25(4), 535-538.

[] Lane, A.C (1923), Keweenaw geothermal gradients and the ice age, Geological Society of America

Bulletin, 34(1), 86.

[] Majorowicz, J. and J. Safanda (2004), Measured versus simulated transients of temperature logsa

test of borehole climatology, J. Geophys. Eng, 2, 1-8.

[] Mann, M.E. and G. Schmidt (2003), Ground vs. surface air temperature trends: Implications for bore-

hole surface temperature reconstructions, Geop. Res. Lett., 30(12), doi:10.1029/2003GL017170.

[] Schmidt, W.L., W.D. Gosnold, and J.Enz (2001), A decade of air-ground temperature exchange from

Fargo, North Dakota, Global and Planetary Change, 29(3-4), 311-325.

[] Zhang, T. (2005), Influence of the seasonal snow cover on the ground thermal regime: An overview,

Reviews of Geophysics, 43, 1-23.

[ ] Zhang, X., L. Vincent, W. D. Hogg, and A. Niitsoo (2000), Temperature and precipitation trends

in Canada during the 20th century, Atmosphere-Ocean, 38 (3), 395-429.

Documents

Observations of northern latitude ground-surface and surface-air