where GSTa and SATa are annual means of the GST and SAT, index "p" means the peak amplitudes, and indices "s" and "w" represent the summer and the winter, respectively. Including in this equation regression coefficients from Eq. (33), one can transform it into

GSTa - SATa = - a (SATa X AGST-SAT ) + I P I (SATA X AGST-SAT ), (35)

where SATa is the year-to-year amplitude of the annual SAT signal. Observations by Smerdon et al. (2006) have revealed 1.5-4.5K GST-SAT differences at Fargo (North Dakota) in 1981-1989 and 1993-1999 periods. Their variations were erratic and did not exhibit any significant linear trend for approximately two decades of observations. Correlation between differences observed and calculated by expression (35) was 0.86 (significant at the 0.0001 probability level). It was shown that calculated differences explain 73% of the variance in the observed values.

The above estimation has supported the hypothesis that meteorological conditions are the dominant causes for the occurrence of the GST-SAT decoupling and that the above empirical regressions represent a useful tool for the investigation of the GST-SAT differences on the longer scales. Application of the long meteorological records to the mul-tivariate expressions (33)-(35) provides the possibility of the Agst-sat calculation on decadal to centennial timescales. This idea was realized in the work by Pollack et al. (2005) (see Section 2.6.4).

While most of the relationships between ground temperature and meteorological variables generally represent the multivariate empirical regressions (like above-described Eq. (33)) and are not focused on the underlying physical processes, England et al. (2003) and Lin et al. (2003) have worked out numerical model that captures effects of rainfall on the GST temperature changes and provides controlled reliable simulations based on the quantitative description of the coupled moisture and energy transport through the air-ground interface. Their methodology uses the Land Surface Process (LSP) model that takes into account vertical energy and moisture transport in soil and vegetation. Its extensive discussion is presented in the works by England (1990),

Liou and England (1996), Judge et al. (2003), and Lin et al. (2003). Extensive calibration and validation of this model was performed through numerical field experiments (Judge et al., 1999, 2001).

The LSP comprises relatively complex and detailed description of the microphysical ground-atmosphere processes using multilayered soil and vegetation. The temperature and moisture profiles of the ground and canopy are determined by the coupled energy and moisture transport based on the changes in infiltration, evaporation, transpiration, and recharge fluxes over time. Infiltration is a positive term (water at the ground surface enters the soil) governed by the hydrological properties of the subsurface and by the difference between total precipitation and its share captured by the canopy. Effect of the runoff is generally not taken into account. It occurs when the precipitation rate exceeds the rate of infiltration, while most of the simulations do not include such extreme precipitation events. Because of an absence of runoff and because surface vegetation characteristics remained constant on the multiyear timescale, all changes in infiltration in the LSP model were caused exclusively by the changes in precipitation characteristics. Both evaporation and transpiration are negative terms that remove moisture from the soil and/or surface canopy by transport of vapor into the atmosphere. Finally, recharge fluxes occur due to the fluid flow at the water table boundary. Depending on the flow direction they can be positive or negative. It should be mentioned that the annual sum of recharge fluxes is small; thus, this latter factor is not as significant as the former processes.

In the works by England et al. (2003) and Lin et al. (2003) the 1-D model is developed for the multilayered soil with a two-layer vegetative canopy (grass and thatch) at the surface of the hypothetical location characteristic for the prairie grassland in the state of Kansas, belonging to the Great Plains area of the USA. This environment appears to be the most suitable to distinguish the influence of precipitation from other factors, because it is practically not subjected to snowfall and/or freezing and thus can illustrate well the pure effect of rainfall. Other advantages represent a good knowledge of the soil structure and an abundance of the meteorological data for the credible model forcing. Numerical simulations of the microclimatic ground-air interactions were performed over decadal timescales with minute resolution; thus, the authors have investigated both short- and long-term effect of precipitation on the GST changes. Thermophysical and hydrological properties of the subsurface layers in LSP model vary not only with depth, but also with the temperature and moisture content. The model forcing realizes from the surface and includes down-welling radiation, SAT, humidity, wind, cloudiness, and precipitation. An evaluation of the precipitation changes on the GST was a central goal of above researches. The study has concentrated on four primary characteristics of precipitation: its amount, intensity, frequency, and timing that were considered on the wide range of scales from diurnal to decadal. The SAT and precipitation data for model forcing were taken from the database of the U.S. National Climatic Data Center (NCDC; Meteorological time series have corresponded to the above-mentioned southern prairies region of Kansas.

Results of numerical experiments have proved that independently of other meteorological processes changes in various rainfall characteristics alone in principle can affect the GST temperature. The range for possible changes in precipitation characteristics was chosen in such a manner that simulation results were able to put upper limits on the expectable GST changes introduced by precipitation. The study has revealed the next principal influences of the precipitation on the GST change:

(1) To investigate the influence of the precipitation amount on the GST the baseline precipitation distribution was multiplied by factors from 0.5 to 2.0. Increase in the amount of daily precipitation manifests itself in the corresponding cooling and wetting of the ground and reduces mean annual GST. The reason is that enhanced precipitation intensity increases the amount of retained moisture and prolongs the time of the water storage in the ground. Both processes cause an increase of the average annual latent heat flux and a corresponding decrease in the sensible flux. Decrease of the precipitation amount and intensity has the reverse effect and causes warming in the ground. The 100% increase in the amount of daily precipitation in comparison with an average baseline value for the Great Plains area may cause annual GST cooling of ~0.5K, while its 50% decrease has resulted in the 0.6 K warming.

(2) To investigate an influence of the precipitation intensity on the GST its distribution was filtered to either decrease rate of occurrence and increase intensity (increasing variance) or to decrease intensity and increase the frequency of rainfall (decreasing variance). The possibilities varied from the constant drizzle all over the year (standard deviation equals to 0 mm) to the weekly precipitation amount that has fallen during one year (s.d. = 8.2 mm in comparison to the 6.2 mm for baseline). In all cases an amount of annual precipitation remained constant. Decreasing frequency and increasing intensity of the daily precipitation results in the cooling of the ground and increasing of the soil moisture content. On the contrary, increasing frequency and reduced intensity leads to warming and drying of the ground. In addition, when precipitation intensity decreases, significant part of the available moisture remains at the canopy and does not penetrate into the deeper soil. This causes an increase of the latent heat at the air-surface interface, because evaporation of shallow moisture occurs more rapidly than of that stored at deeper levels. Numerical simulations have shown that the 25% increase in the precipitation variance has cooled the ground by only 0.07 K, while a similar decrease has warmed the ground by ~0.3K. The factors described above thus have stronger impact on the GST than the changes in precipitation intensity.

(3) Experiments with diurnal precipitation timing (e.g. daytime or nighttime) have shown that it is not significant for the appearance of the GST changes on the longer timescales. On the other hand, precipitation is not equally distributed also over the year. Similar experiments with seasonal precipitation timing revealed more noticeable relationships between annual precipitation peaks (as presented in Figure 50), seasonal changes of the solar radiation, and the SAT. When precipitation maximum coincides with the maxima of the two latter variables (e.g. in Prague occurring in July), it reduces the mean annual GST and increases the annual soil moisture content. In those locations where precipitation peak is close to the radiation minimum in January precipitation causes the warming and drying of the soil with the corresponding reduced role of the latent heat and increased sensible heat. The physics of the process is the next. As known, recharge rates reach their maximum after precipitation events in the winter, when soil moisture is high and latent heat flux is low. When precipitation peak occurs in cold season and thus coincides with the maximum positive recharge rates, significant part of the moisture is removed to the phreatic9 zone as positive recharge. Moisture flow to deeper layers dries the uppermost soil. According to the estimates by England et al. (2003) and Lin et al. (2003), seasonal clustering of precipitation can in principle change the GST by 0.4-0.5 K. Detected influence of the seasonal distribution of precipitation on the magnitude of energy and moisture fluxes at the surface hints that the rough modeling of the precipitation influence on the long-term GST-SAT coupling based only on annual averages may not exactly reflect the consequences of seasonal patterns.

Above numerical experiments have shown that estimated maximum GST changes, caused by corresponding changes in the main characteristics of precipitation, may reach tenths of degree. Even though such magnitudes are small, potentially they are not insignificant for detection of the real amplitude of the climate signal. Resulting subsurface temperature-depth profiles have a curvature similar to that caused by the climate change (so-called "U-shapes"; see Figure 20, Chapter 2). This opens the possibility of misinterpreting both effects. At a first glance the problem appears quite serious. However, numerical experiments by England et al. (2003) and Lin et al. (2003) have been performed for the extremely wide range of precipitation characteristics to put upper limits on the possible GST changes. Applied range of precipitation changes significantly exceeded really observed characteristics; thus, calculated amount of the GST disturbance can be taken only as acceptable upper limits. Precipitation influence on the GST does not appear so serious in the real nature. Numerical experiments by England et al. (2003) suggest that much less than half of the GST warming detected for the last five centuries could be credibly attributed to the overall changes in precipitation amount or its redistribution within the year. This conclusion was supported by the Lin et al.'s (2003) estimates, who have found that the GST response for really observed precipitation changes on the long scales will be relatively low. For example, during the twentieth century precipitation has increased by only 5-10% at the territory of United States. The Intergovernmental Panel of Climate Change (IPCC; has reported an increase of 0.5-1% total (including snowfall) decadal increase in the mid and high latitudes of the Northern Hemisphere. Some of subtropical areas have been subjected to only 0.3-0.5% decadal decrease in precipitation. The minimal increase/decrease factor used in model simulations by Lin et al. (2003) was 25% causing approximately ±0.2K GST change. It is more than twice larger than the actual precipitation increase estimated for North America and/or Northern Hemisphere. Extrapolation of the simulated GST change to the observed precipitation trends gives the values of the GST disturbance of only 0.05-0.10K. This temperature range is approximately an order of magnitude smaller than the amount of the twentieth century warming. Similar estimates have shown that the changes in the occurrence of extreme precipitation events that were also reported by the IPCC will cause very small changes in the annual GST of the order of hundreds of degree. And finally, no significant changes in seasonal timing of precipitation were documented over at least twentieth century; thus their contribution to the long-scale GST changes appears to be negligible.

9The phreatic zone represents permanently saturated with groundwater layers of soil or rock below the water table.

Taken together, described monitoring and modeling results help to understand real influence of the various effects of precipitation on the GST-SAT differences on seasonal and annual scales. An extrapolation of conclusions based on short-scale observations over much longer timescales is complicated. While short-scale GST-SAT differences may achieve several degrees of Celsius with significant and irregular inter-annual variations, the amplitude of the long-term trends is typically an order of magnitude lower. Because the amplitude of inter-annual GST-SAT differences will be smoothed on the long-scale averaging, it is obvious that only secular changes in the GST-SAT differences can violate the use of the GST history reconstructions as reliable estimates of long-term SAT variations. However, because of higher magnitude and irregular inter-annual oscillation of the GST-SAT differences their more weak secular variations may be hidden by the high variability of the short-term pattern. Results of numerical modeling by Lin et al. (2003) suggest that actually observed long-scale precipitation trends can only insignificantly break the GST-SAT coupling. Total effect of the precipitation on the GST is likely incomparable with the GST and SAT changes that occurred during twentieth century.

2.6.4 Effect of surface vegetation

Vegetation is the ground cover provided by plants. It may be regarded as the skin of the ground. It influences various processes in the biosphere at wide spatial and temporal scales. Besides that the vegetation regulates numerous biochemical processes (e.g. water, carbon,10 and nitrogen cycles),11 it also influences local and global energy balances that are important for the climate. In forested areas, e.g. not more than 5-20% of the shortwave solar radiation reaches the ground surface (Beltrami, 2001a; Nitoiu and Beltrami, 2005). Thus, the ground temperature exhibits weaker fluctuations in the regions with complete dense tree cover. Removing of this protection layer will be accompanied by a corresponding increase in solar radiation that reaches the ground surface and subsequent re-arrangement of all energy balance components (net short wave radiation, net long wave radiation, latent heat, sensible heat, and ground heat). Vegetation also strongly affects soil characteristics (e.g. soil volume, texture, and composition). Both processes can influence the GST-SAT coupling.

Investigations of the GST-SAT coupling by Smerdon et al. (2004, 2006) comprising results of the Czech and the North American monitoring experiments have shown that the differences between soil and air temperatures arise in both winter and summer seasons. While the snow cover/soil freezing are responsible for the decoupling of winter temperatures, summer precipitation reduces soil temperatures relative to SAT through evapotranspiration process. Observations have shown that except for the influence of the summer and winter precipitation and soil freezing/thawing on the GST-SAT coupling, it depends also on the type of the land cover. The above-mentioned monitoring experiment at the Prague-Sporilov site was performed under different surface types. Measurements have detected significant influence of the surface type on the GST-SAT difference. Four types of the surface were chosen for the experiment: the bare soil, the sand, the grass, and

10In the biosphere carbon cycle represents an exchange of carbon between living organisms and the nonliving environment.

uThe nitrogen cycle describes the transformation of nitrogen and its compounds in nature.

the asphalt. The 3-year temperature averages indicate that the soil is warmer than the air for all surface types, but the soil (at 2 cm depth) and air (at 5 cm height above given surface) temperature difference was surface cover dependent and amounted to 1.5K, 1.9K, 0.3K, and 4.4K for the bare soil, the sand, the grass, and the asphalt, respectively. This pattern is valid also for the individual year averages. The inter-annual variability of the GST-SAT differences seems to be of the order of the first tenths of degree of Kelvin. New denser grass was seeded in spring 2004 and since that time the temperature above the grass cover appeared to be higher, probably due to decreased air circulation around the sensor that was partially protected by the grass. The highest difference for the asphalt can be explained by an extremely low albedo of this material that makes it very sensitive to incident solar radiation during the year. The GST of the asphalt in the "sunny" year 2003 was by ~0.7K higher than in more "cloudy" year 2004 and 2005, whereas the air temperature was higher by less than 0.3 K. During the winter, vegetation can give a similar insulating effect as a snow cover, protecting the ground from the weather extremes that induce high rates of heat transfer from and to the atmosphere. For example, studies have shown that forest soils do not really freeze in the winter due to the buffering capacity of forests. During January-February 2006, the weather in Prague-Sporilov site was characterized by heavy frosts and absence of the snow cover (see also Section 2.6.2). As a result, temperature under all surfaces has dropped to near the freezing point. Minimum temperatures at the depth 50 cm under the bare soil, the sand, the grass, and the asphalt were -0.29, -0.35, 0.26, and 0.046°C, respectively. The higher temperatures under the grass are given by the insulation of the vegetation cover and those under the asphalt by the above-mentioned low albedo of this material that helped to absorb sunshine during the frosty, but sunny days.

When relating the GST and SAT it is customary to assume that the soil-air temperatures coupling mode remained the same over long time intervals. This assumption could be questioned when the borehole sites were subjected to the drastic vegetation/land use changes. The vegetation changes and their causes are manifold. The processes that lead to the vegetation changes can be characterized as gradual or abrupt. Such processes can produce changes of vegetation structure and/or composition very quickly or for long time periods, respectively. Changes in land cover type may be direct, e.g. agriculture, forest clearing; or indirect as a result of altering disturbance processes, e.g. fire events, landslides, floods, etc. They may be either natural, such as germination, growth, death, or human-induced. All processes can operate over various temporal and spatial scales. Changes in the land cover influence all energy balance components. Exact responses to the land cover change are component specific. For example, both sensible and ground heat fluxes are reduced with an increase in tree canopy. On the contrary, conversion of a forest to short vegetation may raise surface temperatures due to increased sensible heat flux in relative to latent heat flux (Eltahir, 1996). In completely forested areas temperature at ground-air interface is lower than at grasslands or bare soils. Annual ground temperature under complete tree cover is also on average lower, while the soil moisture will be on average higher under such areas cover than under complete grassland. The lower ground and air surface temperature will lead to lower evaporation rates and to decrease the latent heat flux from the ground surface.

Among all possible land use changes the influence of the deforestation on the GST-SAT coupling represents probably the best-studied process. Deforestation is the removal of trees without sufficient reforestation. It may occur naturally as slow forest degradation or sudden extensive forest fires. Anthropogenic influence means conversion of forests to grassland and/or to arable land as well as urbanization and technological uses. Removal of significant tree masses influences all environmental characteristics, changes air-ground boundary as well as the surface hydrological regime, and thus seriously modifies the surface energy balance (Zeng and Neelin, 1999). It generally provokes noticeable changes in climate. Numerous studies have detected an increase in subsurface temperatures following deforestation (Murtha and Williams, 1986; Cermak et al., 1992; Majorowicz and Skinner, 1997; Zhang et al., 2001; Beltrami and Kellman, 2003; Nitoiu and Beltrami, 2005). Except for direct influences caused by the changes in the surface energy balance, climate changes may occur due to indirect feedbacks of altered bio/geo/chemical processes. Climate changes due to deforestation not only are of only local character, but generally embrace global scales as well (e.g. Chase et al., 2000). Betts (2004) and Betts et al. (2004) have compared the radiative forcing caused by the land use changes with the influence of the greenhouse gases, aerosols, and stratospheric ozone (see Section 3.4.2, Chapter 3) and have concluded that these effects have comparable magnitudes.

Common effect of deforestation manifests itself as an increase of surface temperature in tropical and temperate regions (Betts, 2004; Betts et al., 2005). Such changes have been detected in numerous borehole temperature logs. Importance of the account for the deforestation disturbances during GST reconstruction from borehole temperature logs was emphasized in the work by Lewis and Wang (1992). These authors have measured temperature-depth profiles in 11 boreholes located at different Canadian environments. Repeated measurements have shown that average GST depends on the vegetation cover. Thus, in forested areas it is generally 4-5K cooler than at the bare surface. Similar values were measured in Atlantic Canada (Beltrami and Kellman, 2003) and in British Columbia (Plotnikoff et al., 2002). In the regions subjected to deforestation Lewis and Wang (1992) have collected the evidence that these areas have experienced subsequent GST warming. Numerous further studies have corroborated an increase in subsurface temperatures following tree cover removal (Bentkowski and Lewis, 1992; Majorowicz and Skinner, 1997; Skinner and Majorowicz, 1999; Bodri et al., 2001; Cermak and Bodri, 2001; Beltrami and Kellman, 2003; Lewis and Skinner, 2003; Nitoiu and Beltrami, 2005).

Beltrami and Kellman (2003) have performed the monitoring of the soil and air temperatures at three locations in Nova Scotia (Canada) to examine how they follow each other in "field" and forested areas. The "field" surfaces included a clay soil and the grass. High-resolution air-soil temperature monitoring over one-year time interval have shown that the maximum positive differences in soil temperatures between "field" and forested locations occur generally in the warm season (spring and summer) mainly because of direct solar heating of the surface at the "field" sites (the direct solar radiation in the forest amounts to only 5% of that detected at the "field" sites). Because of this effect, the spring thawing has occurred some 2 weeks earlier at the grasslands than in the forests. Differences may reach approximately 8K at the 0-20cm depth range, and are of ~6K in the 50-100cm depth interval. During cold seasons differences are significantly smaller. The authors have performed numerical modeling of soil temperatures. Their model has applied the 1-D conductive heat transfer regime and used air temperature as the surface forcing function. Calculations have indicated that during frost-free season (approximately 290 days in the spring-fall period) the soil thermal regime in the forest floor is directly coupled to the temperature at lower atmosphere. The discrepancies between measured and modeled data were insignificant. Decoupling of the measured and modeled temperature time series appears to be more noticeable at "field" sites, where assumption of conductive thermal regime driven by air temperature changes was not valid. Detected misfit clearly followed daily periodicity. The measured-modeled data differences were the largest in the daytime. The explanation of this phenomenon lies in the significant increase of the incident solar radiation on the field surface, such that air temperature forcing alone represents only small fraction of the energy driving thermal regime of the subsurface in the daytime. On the contrary, in forested areas, where direct solar radiation was much smaller than at the field sites, the SAT was the main forcing for the ground temperatures and the GST could be accurately simulated by a pure conductive model. In forested areas the short-scale GST-SAT coupling appears to be one by one, while in open fields GST changes are not properly represented by conductive models with SAT forcing.

Because deforestation is a widespread anthropogenic activity at all times and all over the world, many drilling sites may contain such kind of disturbance and need correction to separate influence of the non-climatic energy balance changes superimposed on the climate signal. Indeed, numerous temperature logs have been rejected from the GST history reconstruction because boreholes where they were measured were located in the regions of well-documented strong land use changes. The need to correct borehole temperatures for such perturbations was recognized from the very beginning of the borehole climatology. Lewis and Wang (1992) have suggested simple ramp/step model to correct effect of ground warming observed in several boreholes of British Columbia (Canada) after deforestation. This correction should be applied to the temperature log prior to its use for the GST history reconstruction. Obviously, this model was only a first-order approximation, and was not intended to account for all processes occurring in such areas.

Recently, Nitoiu and Beltrami (2005) developed a more detailed method for simulation of the effects of the GST changes caused by deforestation. One of the most influential studies in the history of forest ecology was that performed by Covington (1981), who described a pattern in organic matter storage as a function of the date of forest harvest. This so-called Covington's curve was based on the study of forest floors in series of northern hardwood stands of different ages in New Hampshire (USA). Since the energy balance at the forest floor is affected by the removal of trees and by the variations of the layer of organic matter at the forest flow, Nitoiu and Beltrami (2005) proposed a model based on the Covington's curve to describe GST variations following deforestation. For the case of total recovery of initial forest the model is formulated as

where TG is GST, t the time after deforestation (in years), and A, B, C, D, TG0 regression coefficients. Figure 52 shows GST variations caused by deforestation that occurred 50 and 100 years B.P., respectively. As seen, the GST sharply increases immediately after deforestation event, when environment is dramatically declined by the biomass removal and by the mixing of the forest flow into mineral soil during harvesting operations. Temperature disturbance reaches its maximum of 2K at approximately 15 years after harvest. As the forest floor organic matter recovers, temperature slowly returns to its original value TG0. Full recovery may take decades or even century-long periods. Eq. (36) assumes that the

Fig. 52. Response of the ground surface temperature to deforestation, two events considered which occurred 100 and 50 years B.P. (model by Nitoiu and Beltrami, 2005; A = 0.221, B = 1.24, C = -0.0649, D = 1.063, GST0 = 0K).

removed forest re-grows to its original state. This model can be improved for the more common cases when the new forest differs from its pre-harvest state or is transformed into bare soil/grassland. In the latter case temperature increase can be much higher than 2K, which is shown in Figure 52.

Subsurface temperature perturbations due to deforestation can then be simulated by 1-D purely conductive equation of heat transfer using synthetic GST histories, as presented in Figure 52, as the surface boundary condition. Correction for deforestation is performed by removal of the simulated T-z profiles from measured temperature logs. Calculations by Nitoiu and Beltrami (2005) have shown that disturbances due to deforestation propagate to some 150-200m depth. Correction is the largest in the uppermost 100 m depth interval. Its values depend on the harvest timing and for the recent century events can range between 0.1 and 0.6K. Effect of deforestation is more serious for the more recent events, while the magnitude of the disturbances caused by remote deforestation is significantly attenuated by subsurface heat diffusion process. In the case of remote deforestation at borehole site the GST histories inferred from corrected and uncorrected T-z profiles were practically indistinguishable. This hints that correction for deforestation is unnecessary in regions that experienced older deforestation. Both numerical simulations and analysis of field examples performed by Nitoiu and Beltrami (2005) have shown that the above method very effectively removes disturbances caused by deforestation. The shortcoming of suggested technique is that exact correction is only possible if timing and character of the land use change is known. In the real field situations the harvested and fully re-grown forest case is far not common. More often the deforestation represents a series of events, whose details are generally not well documented. Poor knowledge of the land use history may be a source of significant bias in applied correction.

The GST anomalies may affect not only areas that were really subjected to the land use changes, but also their wide surroundings. Ferguson and Beltrami (2006) have studied transient lateral effects of deforestation. Their numerical simulations have indicated that the subsurface temperature anomalies caused by deforestation of vast areas may extend far beyond the cleared region. Thus, for 500 m wide deforested area 0.1K temperature increase reaches approximately 40 m distance beyond the edge of the cleared ground in 50 years after deforestation and ~70m after 100 years, respectively. The authors have proposed effective technique for correction of the T-z profiles measured in boreholes drilled close to the areas affected by the land use change. Unfortunately, this method could not be applied at all locations. Similarly to the above-described correction the latter technique also requires at least some knowledge of the land use changes. Generally, applied models are based on an assumption of a homogeneous deforestation, while real deforestation takes place preferentially near areas that have already been deforested. This creates a mosaic of forested and deforested patches that further complicates the lateral effect of deforestation. An impact of such localized deforestation on the development of shallow temperature anomalies was studied by Bense and Beltrami (2007). The authors have used a suite of the 2-D models to illustrate thermal effect of the patch-like deforestation. Heat transfer can take place by both conduction and advection due to groundwater flow. Modeling results have shown that the patch-like pattern of deforestation can produce significant temperature gradients in the subsurface. Anomalous gradients can be intensified by the horizontal groundwater flow if its rate is above 10~8m/s. While in the case of pure conductive heat transfer maximum extent of the anomaly would not exceed ~50m (Ferguson and Beltrami, 2006), lateral advection of heat can extend the measured disturbance to several hundred of meters away from the deforested area during 100 years after forest clearing. The measured up- and downstream T-z profiles can exhibit contrasting features notwithstanding that both areas had undergone the same GST changes.

The vegetation increase can also affect GSTs. In their recent study Kaufmann et al. (2003) have applied statistical techniques to quantify effect of inter-annual variations in vegetation on the surface temperature for different types of land cover over Northern America and Eurasia. The database included satellite measurements of the surface greenness (interpreted as a proxy for photosynthetically active vegetation) and the ground-based meteorological observations for the years 1982 to 1999. Statistical analysis has shown that summer increase in terrestrial vegetation causes corresponding ground temperature decrease. Reductions in the extent of snow cover during the winter compel temperature to rise. Except for the seasonal vegetation increase, its long-term enlargement (e.g. reforestation process) can cause corresponding long-term decrease in the GST. For example, it is the case for many regions of North America over the past century, where subsistence farming was stopped and previous agricultural land was occupied by the forest (Ferguson and Beltrami, 2006).

Other local terrain effects causing spatial and/or temporal variations in the land cover, such as forest fires, can also affect surface temperature and influence underground temperature field (Skinner and Majorowicz, 1999; Lewis and Skinner, 2003). Yoshikawa et al. (2003) have investigated an impact of wildfire on the ground temperature in the boreal forests of interior Alaska. Their experiment has detected significant increase of the near-surface temperatures in a short time after ignition. At 2 cm depth temperature has risen to more than 800°C already in about 10min after ignition. However, the ground temperature has increased only at the shallowest layer (<15cm). According to Yoshikawa et al. (2003), most of the heat from the fire is transferred into the subsurface by pure conduction that at the thermal diffusivity of the medium of ~10~6m2/s penetrates approximately 12 cm during one hour. This is the reason that no significant increase in temperature was detected below 15 cm. Longer term GST changes in burned areas occur mainly due to removal of vegetation and destruction of the organic matter layer covering the forest floor. According to the study by Yoshikawa et al. (2003), this layer is an important thermal insulator. Effect of both factors on the GST is similar to deforestation. The authors have monitored annual GST variations in the regions that were on fire. Comparison with the GST measured at the adjacent control unburned sites has shown that similarly to the case of forest clearing at all investigated locations summer ground temperatures were warmer than at the adjacent unburned control sites by 1-20K. The differences were less significant during the cold season. In case of fires that have burned during the last 10 years, the GST was higher at the burned sites than that at control sites for the early freezing period in autumn. This difference existed until the subsurface active layer became completely frozen.

Results of the above three sections concerning seasonal and terrain effects on the GST-SAT coupling can be summarized as follows:

(1) Monitoring results have shown that the SAT forcing represents the main cause for the GST changes. This finding generally supports the use of GST as an indicator of the SAT changes at times prior to the beginning of the instrumental record.

(2) Differences between annual GST and SAT signals are closely linked to the processes occurring in the shallow, approximately the upper meter, zone beneath the ground surface.

(3) Coupling of soil and air temperatures over a single year is complex. The winter snow cover and freeze/thaw effects represent the dominant influences causing GST-SAT decoupling. Because snow cover insulates the ground in the cold season, its systematic and persistent variations may distort one-by-one air-ground temperature coupling and hinder the direct comparison of both variables. In the regions with short-duration snow cover its random fluctuations tend to vanish in the longer period averages. The summer evapotranspiration likely have weaker effect on the GST-SAT decoupling.

(4) On the daily scale the GST may be warmer or cooler than SAT in the winter or summer, respectively. These regular seasonal differences manifest themselves as the attenuation of the annually averaged GST signal in comparison with the SAT. Percent of the GST attenuation may vary from approximately 7-8% to 20-25% (Smerdon et al., 2004, 2006). This effect may be progressively intensified as the seasons become more extreme. The seasonal decoupling have lesser influence on the phase shifts of the GST relative to the SAT. The temperature of the ground surface remains almost in phase with that of the air. Annual GST delay relatively to the SAT variations generally amounts to only 5-8 days and thus can be regarded as negligible for the long-term GST-SAT comparison (Smerdon et al., 2004, 2006).

(5) Normally the mean annual GST is higher than the SAT. For the most mid-latitude regions this difference amounts to 1-2K. This value is higher in the regions with deep, long duration snow cover and/or subjected to extreme soil freeze/thaw cycles.

(6) Within each region variations in the ground properties and surface characteristics due to environmental changes (vegetation, anthropogenic influence) can cause appreciable local variations. The GST-SAT comparison may be problematic in regions subjected to significant land use changes. Effective correction of the GST data is possible, if the timing and character of the terrain change are known. However, various events occurring on the land surface are generally not well documented. In this case their contribution to the heating/cooling of the surface cannot be readily separated from other surface effects.

(7) Fortunately, all detected processes that could break the GST-SAT coupling have only short-term effects on the heat transfer in the subsurface, particularly of daily to seasonal timescales. Most of them occur sporadically, while some of these factors tend to compensate each other under the averaging through the larger spatial scales. In general, detected short-term GST-SAT differences cannot violate the assumption on the tracking of both temperatures on the long timescale.

2.6.5 Long-term soil-air temperature coupling

All above details on the GST-SAT coupling concern short timescales (daily to annual). These short-term GST-SAT differences are well documented. They arise due to insulating effect of snow cover, freezinglthawing, evapotranspiration, seasonal changes in vegetation, etc. These effects are well interpreted and most of them can be taken into account by using the reduced thermal diffusivity on shallow subsurface. To jointly interpret results of the GST reconstruction with the SAT measurements andlor proxy data one should detect not only short-term GST-SAT tracking, but also their long-term coupling. If the differences between both kinds of the data are assumed to remain constant in time, the revealed GST-SAT decoupling will be preserved over the long timescales. In the opposite case of only randomly changing from year-to-year differences, detected on the daily to annual scales, GST-SAT departures signify nothing about the changes over centennial scale periods. It likely means that existing interpretation of the GST histories inferred from borehole T-z profiles as an estimate of the long-term SAT trends is correct. It therefore represents a challenge to extrapolate conclusions on the GST-SAT relationship observed on short timescales over much longer periods of time.

The above-discussed GST-SAT differences occur because the fact that subsurface temperatures integrate combined surface signal and do not retain the air temperature variations alone. Studies described above dealt with the GST-SAT coupling on short (from daily to annual) timescales at single locations. Comparison revealed differences in the amplitude and phase between two signals that occur both in the winter and in the summer and vary with meteorological conditions. Downward propagation of the surface temperature signals is affected by the snow cover, subsurface freezing and thawing, rainfall, water infiltration and its subsurface migration, vegetation, evapotranspiration, etc. These processes have primarily short-term effect, and observed GST-SAT differences can be successfully approximated using daily meteorological observations (Pollack et al., 2005). On the contrary with short-scale temperature monitoring experiments, borehole temperature-depth profiles are used for estimation of the global and hemispheric temperature on timescales of centuries or longer (see Section 3.2, Chapter 3). What is about GST-SAT tracking on the large spatial/temporal scales characteristic for the paleoclimatic studies?

All existing interpretations of merged GST and SAT records were based on the assumption that both databases are closely related on the long timescales. Because of detection of the short-term GST-SAT discrepancies, the question about a long-term GST-SAT tracking has arisen and an assumption of their longer scale coupling also has become a subject of discussion.

To better understand the long-scale GST-SAT coupling and to document the details of the penetration of the surface signal into the ground, a climate and ground temperature observatory was installed in arid northwest Utah in 1994, and over a decade-long ground temperature monitoring has been performed at Emigrant Pass Observatory (EPO), Utah (Bartlett et al., 2004; Davis et al., 2006). The EPO (41.50°N, 113.68°W, 1750m asl) represents a standard weather station situated on exposed granitic rock at the top of a 150 m deep borehole (GC-1) drilled in 1978. Results of its repeated temperature logging are presented in Figure 16 (Chapter 1). Inversion of measured T-z profiles inferred surface temperature changes that are closely coherent with those observed at the nearby meteorological station 40 km to the northeast (Chisholm and Chapman, 1992). Observatory consists of an array of thermistor strings in the subsurface. Ground temperatures are monitored at several shallow depths from 2.5 cm to 1m. Meteorological and shallow ground variables are recorded simultaneously. All data from the EPO since November 2004 is available and can be found on the web site epo/EPO_data. The file is automatically updated daily. The combined database gives the opportunity to observe the GST-SAT dependence in near real time and to test theoretical models of the GST-SAT interactions. Over decade-long continuous temperature monitoring has shown that GST variations are influenced mainly by the surface air temperature that explains 94% of the GST variance, by incident solar radiation (accounts for 1.3% of the GST variance) and snow cover. Ground temperatures are generally higher than air temperatures. Daily averaged GST-SAT differences range between +14 and — 10K. They are much lower on the annual scale and vary between only 2.3 and 2.5 K. These differences occur due to the solar radiation effect in the summer and the insulating effect of snow cover in the winter. Much of the inter-annual variations in the GST-SAT difference occur due to the changes in solar radiation. It was shown that incident solar radiation is more important during the summer. On the long scale there is a linear relationship between the GST-SAT difference and solar radiation with the slope of 1.21 K/100W/m2 and the intercept of 2.47K (Davis et al., 2006). Because of its low thermal diffusivity snow attenuates surface temperature variations in the winter, but its insulating effect is of only minor influence on the annual GST-SAT coupling at the EPO site (accounts for only 0.5% of the annual GST variance). Using EPO monitoring results Bartlett et al. (2004) have developed two-layered forward numerical model of snow-ground interactions. The model is based on three characteristics of snow cover: (1) the onset time, (2) duration of the snow cover, and (3) its thickness. These parameters are generally available from meteorological and remotely sensed data; thus, the authors have validated their model using the century-long National Weather Service data from numerous sites over North America (see above). Their calculations have verified the applicability of the developed model for the broad spectrum of snow conditions and have confirmed its suitability for the prediction of the GST changes in different environments.

I- 1960 1970 1980 1990

Was this article helpful?

0 0

Post a comment