Apr 10, 1998 - K) in central Canada than in eastern Canada. Meteo- rological records from Manitoba and Saskatchewan do not extend further back than 1870 ...
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. B4, PAGES 7385-7397, APRIL 10, 1998
Ground surface temperature history in central Canada inferred from 10 selected borehole temperature profiles LaurentGuillou-Frottier,• Jean-ClaudeMareschal,and JulienMusset GEOTOP, Centre de Rechercheen G•ochimie Isotopiqueet G•ochronologie Universit• du Quebec k Montreal, Canada
Abstract.
Among the 57 temperature-depth profilesrecently measuredin Mani-
toba and Saskatchewan (Canada), only 10 are suitablefor inferringrecentchanges in ground surfacetemperature. Many of the rejected temperature profilesshow an apparent climatic signal but are affectedby topography,changesin vegetation, or the proximity of lakes. At the northernmostsite, near Lynn Lake, Manitoba, shallow horizontal variations in temperature have been identified that are related to intermittent permafrost. Such variations can induce an apparent "climatic" perturbation. In areaswhere lateral thermal conductivitycontrastshave been measured, heat refraction effectscan also incorrectly be interpreted as ground surface temperature histories.The analysissuggeststhat forestfires that occurredat some of the sites have had little influence on the temperature profiles. The 10 selected
temperature profilescovera wide area in northern Manitoba and Saskatchewan. They have been independentlyand jointly inverted by a singularvalue decomposition method. The groundsurfacetemperaturehistory showstwo main episodes.A cold period, tentatively identified as Little Ice Age, with a minimum around 1820 A.D., was followedby marked warming after 1920. These trends are similar to those recognizedin easternCanada. 1. Introduction
dueto topography[Blackwellet al., 1980],waterflows [Lewisand Beck,1977],the presence of nearbylakes [Lewisand Wang,1992],or changes in the surfaceenvicentclimaticchanges in severalpartsof the world[e.g., ronment[$ebagenzi et al., 1992]. The thermalperturIn recentyears,boreholetemperature data havebeen extensively used to provide additional evidencefor reCermak, 1971; Vasseuret al., 1983; Lachenbruchand Marshall, 1986; Nielsen and Beck, 1989; Mareschal and Beltrami, 1992; Bodri and Cermak, 1997; Harris and
bations induced can incorrectly be interpreted in terms
of groundsurfacetemperaturehistory (GSTH). It is necessaryto investigate all possiblesourcesof thermal
Chapman,1997].Indeed,transientsurfacetemperature perturbationsat the surfacebeforeusinga temperature perturbations propagate downward and, although attenuated, are recordedin the Earth's subsurfaceas perturbations of a steady state temperature regime. Subsurfacetemperatureis routinely measuredin mining exploration boreholesto determine the heat flow. Some of these boreholetemperature data contain a record of recentclimate changesthat can be recoveredby careful interpretation.
profile as a record of past climatic changes.When surface and terrain conditions are adequate and all other sourcesof perturbation have been ruled out, the temperature profiles can be interpreted as the superposition of a transient component due to recent climatic
changesover the steady state profile. Many authors have inferred, from high-quality temperature profiles, GSTHs that are consistentwith meteorologicalrecords
Although the climatic perturbations are superposed and other proxy data [e.g., Beltfatal and Mareschal, on the equilibrium temperature profiles, many authors 1992; Gosnoldet al., 1997]. When sufficientlydeep have emphasizedthe danger of interpreting data with- boreholesare used (>500 m), two main episodeshave out careful selection. Several non climatic effects are been identifiedin the GSTH. A cold period lasting sevknown to affect temperature profiles, such as those eral hundred years before 1850 A.D., tentatively iden1Nowat D•partement de M•tallog•nie et G•odynamique, tified as the "Little Ice Age" [Grove,1988], was folBureau de RecherchesG6ologiqueet MiniSre, France. lowed by the global warming trend of the past century
[e.g.,Joneset al., 1986;HansenandLebedeff, 1987].In-
Copyright1998 by the AmericanGeophysicalUnion.
version of borehole temperature data suggeststhat the timing and amplitude of these climate episodesvary between regions. The cold period was marked by an
Paper number 98JB00021.
0148-0227 / 98/ 98JB-00021 $09.00 7385
7386
GUILLOU-FROTTIER
ET AL'
GROUND TEMPERATURE
m 0.5ø- 1.0øC decreasein temperature, at m 1700 A.D.
HISTORY IN CANADA
Saskatchewan (Figure1). Theoretically,oneshoulduse
temperature profilesfree of any perturbation to the onedimensionalconductiontheory. Unfortunately, most of the data in central Canada do not satisfy this condition. We first discussthe data and how they are seThese differencesbetweenEuropean and American tem- lected. Examples of terrain effects are presented, the peraturesin the 18tn and 19tn centuriesare supported inversionprocedure is briefly described,and the results
in France[Marescha!and Vasseur,1992],1800A.D. in easternCanada[Beltfatalet al., 1992;$hen and Beck, 1992],1850A.D. in Alaska[Kakuta,1992],and m1880 A.D. in centralEurope[Clauserand Mareschal,1995].
from individual
by tree ring chronologies [Fritts and Lough,1985].
and simultaneous
inversions
of the 10
Temporal variations in ground surface temperature retained profiles are presentedand discussed. were first inferred
from borehole
data in northern
On-
tario by Beck and Judge[1969]and Cermak[1971]. 2. Temperature These authors recognizedthat a cold period, identi- Canada
Profiles in Central
fied as Little Ice Age (LIA), had precededthe current warming trend. These results were later confirmedby several
studies
in eastern
Canada
that
identified
the
LIA and the recent warming in boreholetemperature
2.1.
New
Data
and
Selection
For the past 4 years, several campaignsto measure terrestrial
heat flow have been conducted
in northern
data [Nielsenand Beck,1989;Beltramiand Mareschal, Manitoba and Saskatchewan (seeFigure I for location 1991; Mareschal and Beltrami, 1992; $hen and Beck, of the heat flow sites). Fifty-seventemperaturepro1992; Wang, 1992; Beltrami et al., 1992; Wang et al., fileswere obtained (26 heat flow sites), and coresam1992;$hen et al., 1995].In Ontarioand Quebec,inver- ples were collected for thermal conductivity and heat sion of borehole temperature data suggestsa warming productionmeasurements.One heat flow site consists, by I to 2 K over the past 150 years following a cold when possible,of severalboreholesseparatedby a few period (LIA) with minimumtemperaturem1800A.D. hundred meters. This allows identification and correc[Beltrami and Mareschal,1991; Beltrami et al., 1992; tion for local terrain effectswhen calculatingthe heat $hen and Beck,1992; Wanget al., 1992]. Althoughthe flow. detailed GSTH vary significantlyfrom hole to hole, this
Temperature measurements were made at 5- or 10-
generalpatternseemsubiquitous and is confirmed-bym intervalswith a calibrated thermistor with precision the simultaneousinversion of many temperature profiles from eastern
Canada.
of 0.002 K and accuracy better than 0.02 K. Thermal conductivity measurementshave been completed
Few heat flow siteswere availablein central Canada, for almost all the sites. We have identified refraction but inversion of individual profiles suggesteddiffer- effects due to horizontal variations in thermal conducences in climate variations between central and easttivity and eliminated from the study the correspondern Canada. A very deep hole at Flin-Flon (Mani- ing temperature-depth profiles. On the other hand, we toba) reportedby $asset al. [1971]was reanalyzed have not identified any systematicvariation with depth by Wang[1992],who concluded that the LIA might of the thermal conductivity.For all the sites,it appears have occured m100-200 years earlier in Manitoba than that the thermal conductivity variesrandomly around in easternCanada. Also, Beltrami et al. [1992]sug- its mean value. This observationis confirmedby the gestedthat the recentwarmingtrend is lessmarkedand quasi-linearity of the temperature profilesin the deeper
that the amplitudeof the previouscoolingis less(