On the relationship between ground temperature histories and meteorological records: a report on the Pomquet station

Abstract

An experimental air–ground climate station is operating in Pomquet, Nova Scotia, monitoring meteorological (surface air temperatures at three heights, wind velocity and direction, incoming solar radiation, precipitation, snow depth and relative humidity) and ground thermal variables (soil temperatures at depths of 0, 5, 10, 20, 50 and 100 cm). Readings are taken every 30 s and 5 min averages are stored, in order to characterize the energy exchanges at the air ground interface. Here, I report on the first year of operation. For spring, summer and fall, we find that soil temperatures track surface air temperatures with amplitude attenuation and phase lag with depth confirming that heat conduction adequately describe the soil thermal field at the Pomquet site. For winter conditions, we find that heat transfer is dominated by latent heat released during soil freezing and to a lesser extent by the insulating affect of snow cover. A numerical model of heat conduction was used in order to estimate the magnitude of the heat released by freezing during the winter months. I also show that there is an inverse correlation for the difference between soil (100 cm) and air temperatures and the incoming solar radiation at the site.

Introduction

Analyses of meteorological records Jones et al., 1999, Hansen and Lebedeff, 1987 and proxy climatic indicators (e.g., Groverman and Landsberg, 1979, Jacoby and D'Arrigo, 1989, Overpeck et al., 1997) suggest that surface air temperatures have been steadily rising since industrialization, particularly in the last century. Such evidence, and the coincidence with the beginning of the emissions of large quantities of greenhouse gases into the atmosphere, raises questions about the effects of anthropogenic activities on the Earth's climatic system and about the potential consequences of associated changes in current world climatic patterns.

Debate on the effects of anthropogenic activities on the climate of this century and on future climate continues to occupy a central part of policy developers and the scientific community. The controversy centers mainly on the capacity of general circulation models (GCMs) to simulate a complex system such as the Earth's climate. Hence, it is extremely important to ascertain the past climatic variations to allow robust validation of GCMs. This has been accomplished in part by the collection and analysis of meteorological and proxy climatic indicators in the last decades. It remains, however, crucial to compile a consistent picture of climatic changes of the past reconciling as many source of paleoclimatic information as possible (Overpeck et al., 1997). Because of shortcomings in the meteorological record (Karl et al., 1989) and the fact that proxy data climatic reconstructions involve multiple assumptions (see Bradley, 1985), a complementary record has started to be explored in the last decade. It is the determination of ground temperature histories from geothermal data. The interest in this method lies in the fact that it examines a direct measure of temperature, free of problems such as variable standards and noise found in meteorological data, and unlike proxy records, it has a very clear physical interpretation (i.e., temperature). Reconstruction of past climatic changes from geothermal data has proven, in the last few years (e.g., Huang et al., 1997, Pollack et al., 1998) to be an additional source of information to complement meteorological and proxy records of climatic change.

It has long been known that past ground surface temperatures can be estimated by analyzing the perturbations to the steady state geothermal gradient (e.g., Lane, 1923, Hotchkiss and Ingersoll, 1934, Birch, 1948). Variations of the ground surface temperature are recorded in the subsurface as deviations from steady state. Indeed, it has been customary to eliminate the effects of known climatic oscillations (mainly the last ice age) from the temperature profiles used for HFD determination (Powell et al., 1988). Furthermore, data obtained at many locations around the globe often show that temperature gradients are in fact disturbed for the first hundred meters (e.g., Cermak et al., 1984, Pinet et al., 1991). Although, non-climatic causes of the nonlinearities in temperature–depth profiles are possible (e.g., thermal conductivity variations, subsurface heat production, ground water flow, topography, urban heat, changes in surface conditions, etc.), most of these factors can be examined and eliminated before ground surface temperature (GST) analyses are performed. The abundance of borehole temperature data around the world has the potential to yield ground surface temperature histories in vast areas of the continents (e.g., Lachenbruch and Marshall, 1986) which otherwise would remain undocumented.

In the Earth, these temperature perturbations appear superimposed on the equilibrium temperatures. In a source free half space, the steady state heat flow is constant. The steady state heat flow is usually evaluated in the deeper part of the temperature profile, which is not affected by the recent temperature changes. The steady state temperature is then continued to the surface and the perturbation is determined as the difference between the measured temperature and the upward continuation of the temperature profile. The magnitude of the anomaly is proportional to the total amount of heat absorbed by the ground. The shape of the perturbation is determined by the thermal history of the surface. This history can be inferred by comparing the calculated temperature perturbation for a model of surface temperature with the data and adjusting the model parameters until a fit is obtained or the surface temperature history can be inferred directly by inversion (see Lewis, 1992; see Pollack and Chapman, 1993, Beltrami and Chapman, 1994 for introductions to this subject; see Beltrami and Chapman on-line teaching aid at http://www.geo.lsa.umich.edu/IHFC/climate1.html).

The attractiveness of this approach to climatic reconstruction rests on the characteristics of heat conduction into the ground. Unlike meteorological records subjected to high frequency variability and noise, the Earth behaves as a low pass filter recording long-term trends of ground surface temperature (GST) changes (Joss, 1934). For example, daily and annual temperature variations penetrate into the ground to depths of approximately 1 and 20 m, respectively, whereas a 100-year long trend will be recorded in the subsurface and will be detectable to a depth of about 100 m. As such, climatic reconstructions from geothermal data can provide a robust complement to the existing paleoclimatic database as well as providing the long-term records of temperature change needed for the validation of General Circulation Models (GCMs) for future climate change estimates.

Recently, several attempts have been made to reconstruct a pre observational mean (POM) in order to extend the meteorological record into the past and provide a reference on which to base the changes inferred from meteorological data (Harris and Chapman, 1997) and to calibrate proxy climatic indicators Beltrami and Taylor, 1995, Beltrami et al., 1995. However, although the Earth's response to the energy balance (or imbalance) at the surface is related to the surface air temperature, the temperature of the ground is an integral of the effects of air temperature variation, vegetation cover and snow cover variations, phase changes (freezing and thawing) and solar radiation changes at the ground surface. The interaction of all these variables determines the temperature of the ground in a complex and complicated series of processes. It is thus important to attempt to clarify the long-term effects of variations in energy exchanges at the air–ground interface on the subsurface thermal regime.

Within this context an experimental air–ground station operates in Pomquet, Nova Scotia, monitoring meteorological variables, soil thermal conditions, snow cover and vegetation cover, in order to examine some of the processes involved in the energy exchanges within the first meter of the soil and at the air–ground interface.

This note provides some of the results arising from these observations and carries out some simple numerical calculations to illustrate the observations. It is shown that during spring, summer and fall, while air temperatures remain above the freezing point, the heat transfer within the soil is dominated by conductive processess, and that during this time the ground temperature tracks air temperature variations closely, whereas during a period of time in winter, freezing of the upper soil introduces non-conductive mechanism (latent heat) such that the ground does not respond to air temperature variation until the upper layers of soil are completely frozen. That is, the ground does not appear to record air temperature variations during this time period in winter. We compare the heat content of the first 5 cm of the soil column as determined from data with the modeled situation where soil freezing is absent in order to estimate the magnitudes of non-conductive processes.