1900 1920 1940 1960 1980 2000 TIME, years

Fig. 41. Annual mean SAT record at the meteorological station Kapuskasing (Canada) for the 1918-2001 period. Data are shown as temperature anomalies from the base period 1961-1990. POM - pre-observational mean temperature.

1900 1920 1940 1960 1980 2000 TIME, years

Fig. 41. Annual mean SAT record at the meteorological station Kapuskasing (Canada) for the 1918-2001 period. Data are shown as temperature anomalies from the base period 1961-1990. POM - pre-observational mean temperature.

As a representative SAT record we have taken mean annual temperatures measured at meteorological station Kapuskasing (49.42°N, 82.38°W) (Figure 41). The homogeneous SAT series exists there from 1918. The record reveals certain warming with the mean rate of 0.015 K/year characteristic for the most of the twentieth century. In the last few decades, the general warming has been accelerated and its rate for the period 1970-2000 was almost triplicate (0.047K/year). Both Kapuskasing and Hearst boreholes are located in a bushed area that was formed after clearing of surrounding forests approximately 100 years ago. Large cleared fields are situated approximately 500m away from Kapuskasing and, according to Wang et al. (1992), may have only small effect of the temperatures. The larger effect from the closer deforestation may be at the Hearst site. The mean temperature anomalies corresponding to the 1918-2000 and 1970-2000 periods equal to 0.52 and 0.76K, respectively.

Temperature logging of the Hearst borehole was performed three times (for details see Section 1.3 (Chapter 1). Figure 17 (Chapter 1) compiles the results of these measurements. As shown, all temperature logs are quite similar with a weak but clear positive "U-shape" curvature in their uppermost parts that hints the presence of the recent warming. For inversion we used temperature-depth data only from below 20 m depth to exclude any seasonal temperature variations. The reducing parameters (T0 - surface temperature and G - geothermal gradient) were calculated by the linear regression of the deepest part of the T-z record. Reduced temperature obtained by subtracting background thermal field from the measured temperature log is shown in Figures 42 and 43. It is curved and systematically positive above 100-150 m depth indicating recent climatic warming. Chisholm and Chapman (1992) have demonstrated high sensitivity of the borehole temperature profiles to the POM-values. This statement is illustrated in Figure 42 that shows the observed and synthetic reduced temperatures for the three different POM-values. The degree of conformity between the real and simulated models is usually characterized by the sum of the squares of deviations between measured and synthetic temperature logs. We have calculated root mean square (rms) misfits for wide spectra of

Fig. 42. Combined meteorological and geothermal data were used to infer the POM-value for Hearst hole; reduced temperatures compared with synthetic transient temperature-depth profiles calculated for three choices of POM for the time prior to 1918 (see text). The inset shows the rms misfit as a function of POM and illustrates the best fit for POM = —1.98 K.

Fig. 42. Combined meteorological and geothermal data were used to infer the POM-value for Hearst hole; reduced temperatures compared with synthetic transient temperature-depth profiles calculated for three choices of POM for the time prior to 1918 (see text). The inset shows the rms misfit as a function of POM and illustrates the best fit for POM = —1.98 K.

Fig. 43. Left: Reduced temperature profile for Hearst hole compared with synthetic temperature profile computed for the best-fit POM-value. Right: "Left-over" temperature.

possible POMs to determine the best fit. Generally, preferred value of estimated parameter corresponds to the minimum of rms misfits. As seen in Figure 42, small differences in POM-value cause significantly poorer fit to the observed reduced temperatures. Even 0.5K difference in the POM-temperature is critical to obtain a good fit with the observed reduced temperatures. For the Hearst hole the best fitting reduced temperature is POM = — 1.98 ± 0.01K (rms misfit = 0.095 K). A sharp extreme in the misfit diagram (Figure 42, inset) indicates the character of the POM as a robust temperature estimate. Obtained POM-value is almost 2.5 K lower than both 1918-2000 and 1970-2000 temperature means, indicating significantly colder pre-1918 conditions.

As shown in Figure 43, there is a satisfactory coincidence between both the amplitude of warming and the depth of perturbed temperatures. The "left-over" temperature residuals, calculated as the difference between reduced and the synthetic best-fit POM-SAT temperature, does not exceed ±0.1K below 100m depth and reach 0.5K in the uppermost part of the borehole. In most cases POM coupled with the SAT measurements explains 80-90% of the transient borehole temperature signal (Harris and Chapman, 1997; Bodri et al., 2001). Larger "left-overs" were obtained, e.g. during the POM estimations from a suite of Cuban boreholes (Bodri and Cermak, 2001), where coupled POM-SAT data explained not more than 50-60% of the transient borehole temperature signal. This indicates that for definite sites, at least some portion of the borehole temperatures cannot be explained by the SAT origin and reflects also specific terrain effects. Larger magnitude of the "left-overs" in the uppermost part of the borehole Hearst can likely be attributed to the local different impact of deforestation detected by Wang et al. (1992) at the Hearst and Kapuskasing sites.

Essential requirements for the correct POM determination are: (1) the pure conductive regime in the subsurface and (2) the persistence of the land-atmosphere boundary layer conditions, thus a "constant" SAT-GST coupling mode. As known, the process of the heat exchange at the land surface is a complex function of the coupled atmospheric-plant-soil interactions; thus, in principle the response of the land surface to the atmospheric forcing may be time-dependent even at the annual and longer scales of aggregation. Drastic changes in the near-surface hydrology (evaporation and transpiration system), albedo,5 and even surface roughness change that accompanies extensive forest clearing, all have significant impact on the ground-air temperature coupling and therefore on the POM estimate. These problems will be discussed in the next section.

2.6 Ground-Air Temperature Coupling: Effect of Various Environmental Changes

2.6.1 Background

The above-mentioned investigations by Baker and Ruschy (1993), Putnam and Chapman (1996), Gosnold et al. (1997), and Harris and Gosnold (1999) have given the confidence that it seems reasonable to view borehole temperatures as the filtered versions of the

5Albedo is an important concept in climatology and represents a dimensionless measure of the surface/body reflectivity. It may be also expressed as a percentage from 0 to 100% and is determined as the ratio of total electromagnetic radiation reflected to the total amount incident upon it. The average albedo of the Earth is about 30%.

surface air temperature (SAT). Modeling of the GST-SAT coupling by González-Rouco et al. (2003,2006) using surrogate SAT simulations have shown that at long timescales the GST represents a good SAT indicator, and their variations practically repeat each other (for details see Section 2.4.4). However, observations do not support this conclusion unconditionally and at all timescales. Recent studies have revealed that in certain regions and under certain conditions the GST does not track accurately the SAT changes, especially at the short timescales. In the recent decade, the problem of the GST-SAT coupling represented the target of continuous study by several research groups.

Factors affecting ground temperature can be subdivided into three general categories: (1) meteorological, (2) terrain, and (3) subsurface thermophysical properties. Large spatial-scale GST differences are determined primarily by meteorological factors: solar radiation, air temperature, and precipitation through the processes of absorption/reflection/emission of solar short-wave and/or thermal infrared radiation and conductive coupling of ground-air temperature. Except for the factors mentioned above, thermal balance in the ground surface can be affected by numerous secondary processes. Basic agreement between reconstructed GST histories and available SAT records has been documented by numerous investigations at different spatial scales all over the world (see the references above). On the other hand, a strong correlation between both signals has been questioned by Majorowicz and Skinner (1997), Majorowicz and Safanda (2005), and Mann et al. (2003), especially for the northern locations with prolonged snow cover in the winter. The latter authors have argued that in such areas ground loses significant part of the information about air cooling in the winter months because snow insulates and reflects the incoming radiation. The GST-SAT decoupling can also arise due to latent heat effects of freezing/thawing processes. The GST-SAT differences at the daily and seasonal timescales are well documented. Except for the influences mentioned above, the one-by-one coupling between the GST and SAT can be also affected by such processes as the partitioning of moisture content between infiltration/evaporation/runoff, the biological processes, e.g. seasonal vegetation changes, chemical weathering and other long-term land surface changes, and other factors that are not directly connected to the climate. The summary of all processes creates the heat flux at the ground surface and thus affects the GST-SAT coupling.

The investigations of the ground-air temperature correlation are performed in two main related and/or complementary directions:

(1) Empirical site-specific observations of the GST-SAT coupling at specific locations using monitoring of the air/subsurface temperatures and other meteorological conditions. A comparison of soil and air temperatures provides a direct test of details of their coupling at shorter timescales (from daily to annual).

(2) Development of numerical models to simulate both short- and long-scale active processes at the level of air-ground interaction and in the subsurface.

(3) Collection of high-quality measurement data. The International Heat Flow Commission global geothermal data set ( contains over 10000 worldwide measured borehole temperature logs. GST reconstructions inferred from these data can be compared with the SAT measurements as well as with proxy sources available in the same locations during periods of overlap.

Such comparative studies can provide information on the longer scale GST-SAT coupling. It should be mentioned that because of suspected non-climatic disturbances, up to now not more than 10% of geothermal data set was used for paleo-climate reconstructions (Nitoiu and Beltrami, 2005).

Despite that studies enumerated above represent different spatial and/or temporal scales, they complement each other. Empirical correlations established by comparison of the meteorological and geothermal data provide an experimental basis that could be simulated by numerical models primarily concentrated on the physics of various near-surface processes. The investigations by Zhang et al. (2001) represent typical example of the latter-kind research. The authors have examined records of soil temperature at several depths and have compared them with the main climatic variables (air temperature, precipitation, snowfall, and snow thickness data) at Irkutsk (Russia) over the 100-year long period from 1898 to 1995. The relationship between air temperature and soil temperature was proved to be so complex that, using the words by the authors, "changes in air temperature alone cannot explain the changes in soil temperatures in this region". This research has captured almost all important sources of uncertainties that subsurface temperatures could contain. One of the surprising observations was, e.g. that summer soil temperatures decreased by up to 4°C while summer air temperatures slightly increased. In other cases, when winter air temperatures have oscillated in the narrow range from 4 to 6°C, the rise of soil temperatures was even higher and reached as much as 9°C. Possible explanations for these phenomena suggested by the authors have included: (1) an increase in summer rainfall and (2) an increase in early winter snowfall coupled with an earlier increase in spring snow melt, respectively. The authors have concluded that the changes in soil temperature represent a combined complex output of the SAT and precipitation variations, especially of the snowfall and snow cover on the ground surface, and have warned that "when changes in soil temperature are used as the evidence of climatic warming, caution is required". They also emphasized that revealed surface warming of permafrost at high latitudes and subsurface ground warming in wide areas elsewhere in the world may be misleading and/or occasional because air temperature alone cannot explain such ground warming.

Similar studies by Baker and Ruschy (1993) and Putnam and Chapman (1996) have detected an air-soil temperature offset, when the ground was generally warmer by 1-3 K as well as the seasonal differences in the detected offset. These and other works on this topic have attracted attention for the possible serious shortcomings that the GST histories inferred from borehole temperature logs may contain. The next sections are devoted to the detailed discussion of this problem.

2.6.2 Snow cover and ground freezing

Winter snowfall as well as seasonal freezing and thawing cycles of soil can strongly influence the thermal and hydrological characteristics of the uppermost ground layers. Impact of these processes on the surface energy and moisture balance at least on the short scales may be quite serious. Latent heat exchanges at the ground surface and snow cover insulates it from air temperature variations. Because the thermal conductivity of frozen soil is larger compared to an unfrozen state, freezing significantly increase the soil heat flow.

Simultaneously it reduces hydraulic conductivity thus decreasing infiltration that can lead both to the more runoff and/or higher uppermost soil moisture content caused by restricted drainage (Williams and Smith, 1989).

Traditionally, investigations of seasonal freeze/thaw oscillations are performed by in situ measurements and numerical modeling of the ground-air temperature coupling, and thus reflect mainly site-specific and short timescale features of the process (Beltrami, 2001a; Schmidt et al., 2001). The effort to extend existing results to broader spatial regions and to investigate how precisely the GST and SAT signals track each other on the seasonal scale was undertaken in the work by Gosnold et al. (1997). The authors compared the GST record with the air temperatures along transect of the Northern Plains between southern Manitoba and northern Texas (approximately 33-49°N) and examined the nature of the ground-air coupling. Flat topography and geology of this area ensured favorable conditions for borehole temperature reconstruction free of potential topographic disturbances, microclimate, and groundwater effects. Criteria for the borehole screening also included surfaces as uncultivated grassland, shale bedrock, and sites remote from the regions of intensive anthropogenic activity that could result in transient variations of microclimate.

For the first test the set of 29 boreholes was selected. The GST reconstructions were performed using FSI method. All obtained GST histories indicated prominent warming trend over the last century. Its amount depended on the latitude; greater warming was detected for the northernmost boreholes. The comparison of the GST histories and SAT was performed using data from 55 stations of the United States Historical Climatology Network (U.S.HCN; situated in the same region. The HCN SAT data series are approximately century long and extend to at least 1994. The latest dates of borehole logging were 1994 and/or 1995. Similarly to the GST reconstructions the HCN temperature data have shown warming during the past century and strong latitudinal trend of its amount. Comparison of both datasets revealed coincidence between amplitude of GST and SAT warming south of about 45°N. On the other hand, the GST reconstructions have shown much stronger warming north of this latitude. The authors have performed modeling of the GST-SAT coupling by repeated calculations and used more than 100-year long SAT signal as a forcing to the 2-D conduction in the subsurface. The FSI of the generated synthetic T-z profiles has given an amount of GST warming similar to the temperature change determined by regression of the SAT records, and thus corroborated the one-by-one coupling between the ground and air temperatures under pure conductive regime of the heat transfer. To interpret obtained inconsistency of the GST and SAT north of the 45°N latitude Gosnold et al. (1997) have tested the data from a network of automated weather stations situated in the investigated area and the results of continuous monitoring of the SAT and soil temperatures at 10 cm depth at automated weather stations installed at three locations with different latitudes at Texas (33.1°N), South Dakota (43.7°N), and Manitoba (49.6°N). The results of this monitoring experiment have shown that the mean annual soil-air temperature differences arise primarily during the separation of both temperatures in winter.

This conclusion can be illustrated with the time series of temperatures recorded during similar monitoring experiment that was carried out by the Research Group of the Geophysical Institute of the Czech Academy of Sciences at the microclimate station Prague-Sporilov, the Czech Republic (50.04°N, 14.48°E, 274m asl). The monitoring has been running continuously since the summer of 2002. Four different surface types were investigated: bare soil, sand, grass, and asphalt. Air temperatures at 5 and 200 cm above the surface as well as the soil temperature at depth levels of 2, 5, 10, 20, and 50 cm were recorded at 5min intervals. The every 3-year (2003-2005) air temperature averages were surface dependent, but appeared lower than the soil temperature means for all four types of the surface. Thus, the differences between air temperature and soil at 2 cm depth amounted to 1.4-1.6K, 1.8-2.0K, 0.2-0.4K, and 4.1-4.8K for bare soil, sand, grass, and asphalt, respectively. This result hints that on the annual scale the soil is warmer than the air and corroborates similar observations mentioned above by Baker and Ruschy (1993) and Putnam and Chapman (1996) who have detected that the ground is generally warmer than the air by 1-3 K. The inter-annual variability of measured in Prague microclimatic station difference is also surface type dependent and ranges within the first tenths of degree Kelvin.

Figure 44 (See Plate 1 of Colour Plate Section) shows temperature variations for some shallow sensors in the Prague-Sporilov hole during the first quarter of the year 2005. It illustrates well the influence of the snow cover on the GST-SAT coupling. As seen, the magnitude of the GST-SAT difference exhibits significant variations. Subsurface heat conduction as well as the factors connected to the movements and/or diffusion of air and/or moisture masses (wind, evaporation/transpiration, vertical soaking of soil moisture, and precipitation) tend to equalize air and soil temperatures. Thus, soil temperatures generally follow the air temperature course when average SAT is above 0°C. The one-by-one GST-SAT coupling violates below zero temperatures in the presence of snow cover, because it insulates the ground surface and reduces heat loss (the condition at the measurement site was not enough cold for the soil freezing). It is noticeable that perfect coupling is restored almost immediately after snow cover was thawed (see, e.g. time interval between February 1 and 15). Similar monitoring experiment was performed at the station Potucky (the Czech Republic 50.43°N, 12.78°E, 864m asl). It is situated in the Ore Mts. forming the natural border between North Bohemia and Germany. The suite of boreholes is located on small territory in the close vicinity of the forested area of coniferous woods. The subsurface temperature monitoring at several shallow depths began in 2003 (for details see Section 4.2, Chapter 4). Figure 45 shows results obtained during autumn 2003 to spring 2004. Temperature was recorded at 2 cm depth and 5 cm height above ground surface to detect the effect of snow cover on shallow subsurface temperatures. The record completely corroborates the results of the Prague-Sporilov monitoring. The coupling of the temperatures is almost perfect in fall and spring and breaks down during most of the winter. The presence of the snow cover in the 2003-2004 winter and absence of really cold temperatures at Potucky station prevented occurrence of soil freezing. Thus, the winter decoupling of the GST-SAT that is seen in Figures 44 and 45 can be attributed exclusively to an influence of the snow cover.

Smerdon et al. (2004, 2006) have generalized results of above-cited and similar monitoring experiments. Except for the Czech records mentioned above, the authors have used temperature time series measured during monitoring experiments at Fargo (North Dakota), Cape Henlopen State Park (Delaware), and Cape Hatteras National Seashore (North Carolina). All sites represent different kinds of subsurface strata and/or climatic settings located within the mid-latitude zone from 35 to 50°N, and thus can be used also for the spatial decisions. Similarly to the Czech records, the North American time series

Sporilov, winter 2005

Sporilov, winter 2005

TIME, day

Fig. 44. Time series of air (at the height 2 m) and soil temperatures (at 2 cm depth) recorded under different surfaces at Prague-Sporilov station. Soil temperatures follow SAT at temperature above 0°C, but are decoupled when the surface is covered by snow. (See Plate 1 of Colour Plate Section).

Fig. 45. Time series of air (at 5 cm above surface) and soil (at 2 cm depth) temperatures recorded at Potucky microclimatic station during the 2003-2004 winter.

represent several years of simultaneous air and soil temperature monitoring at different heights/depths, and are particularly suitable to reveal the differences between annual GST and SAT signals. Thorough examination of the records has shown that on the annual scale GST signal (even somewhat attenuated and insignificantly phase shifted) follows well the SAT variations. The slight differences between annual GST and SAT signals may occur in both winter and summer seasons. Their amount depends on the site location and its climate as well as on the terrain characteristics. Thus, the study by Smerdon et al. (2004) has demonstrated that the GST-SAT decoupling at Fargo occurs mainly during the winter, whereas at Capes Henlopen and Hatteras observed attenuation of the GST signal has taken place during the summer season. The seasonal partitioning of the GST-SAT decoupling is caused mainly by the corresponding partition of the summer precipitation and snow. While the Fargo location is characterized by the modest rainfall and significant amount of snow, the Cape Henlopen and Hatteras stations have negligible or no snowfall. Similarly to the Czech monitoring results, the North American stations inferred influence of the snow cover on the GST-SAT coupling. According to the results by Smerdon et al. (2004,2006), in all investigated locations snow cover has affected heat transfer in the surface in such a manner that mean daily soil temperature under snow cover was warmer relative to the SAT.

The experiments described above have also detected finer features of the GST-SAT decoupling during cold season, e.g. dependence of the temperature of the soil covered by snow on the thickness of snow layer, the snow quality, both air and ground temperatures before a snowfall, the presence of the vegetation cover as well as the thermophysical properties of the soil. Effect of the snow cover thickness is only of secondary importance. Numerical modeling by Gosnold et al. (1997) of the GST-SAT tracking in the presence of the snow cover has detected that the winter soil temperatures are more sensitive to the presence or absence of snow rather than to the variations in its thickness. Thus, the exact amount of the winter snowfall is not likely a decisive factor of the GST-SAT coupling during the winter. Snow pack control on the soil-air temperature tracking in other seasons was studied in the work by Grundstein et al. (2005). Annual coupling of the GST and SAT was investigated using the soil-air temperature measurements performed during 1990-2002 at Fargo (North Dakota) as well as numerical simulations based on the snow pack physical model. In accordance with the conclusions of the previously discussed studies, Grundstein et al.'s research has corroborated that the GST-SAT decoupling in the investigated location appears to be visible only during winters when dense, thick snow cover, and its long persistence cause strong insulation of the ground. In the late autumn and/or early spring the snow is thin and has a low density. It gives imperceptible thermal insulation and does not break one-by-one GST-SAT coupling. The Czech monitoring experiments described above have supported the influence of the type of surface on the ground-air temperature tracking. Thus, the grasslands preserve the snow cover longer than the bare surfaces, where the snow is not isolated from the ground heat flow. Combining the snow cover with the grass provides better insulation and the temperature under such surface remains above zero (see Section 2.6.3). The rate of snow melting was proven to be also surface dependent. The thickest snow cover is characteristic for the grass and the thinnest can be found in asphalt.

The monitoring results mentioned above are more representative of the mid-latitude seasonal GST-SAT relationships. As at the Prague-Sporilov station, winter temperatures at the investigated locations are generally not low enough for the soil freezing. At high-latitude regions where SAT temperature for a long time remains far below 0°C, the effect of freezing may even surpass the influence of the snow cover, Gosnold et al. (1997) have interpreted systematic northward increase of GST-SAT difference in the North America, which is revealed in their work, as the result of the northward increase in the duration of snow cover and often occurrence of the ground freezing. While on frosty days the air temperature may be significantly negative, latent heat released during freezing of soil moisture makes soil temperature remain at 0°C level until the whole moisture content has frozen. This is so-called "zero-curtain effect" that is caused by transfer of latent heat during freezing and thawing of water contained in the rock or soil. The degree of saturation and the thickness of the saturated soil represent the main factors controlling the duration of freezing process. Since near-surface soils often freeze before snow covers the land surface and durable soil freezing (sometimes for weeks to months) is a more often phenomenon than continuous snow cover, the freezing effect appears to be of greater influence on the GST-SAT decoupling. According to the observations by Gosnold et al. (1997), the onset of the strong soil-air temperature decoupling does not always correlate with the variations of snow cover; however, in all cases it coincides with the beginning of the soil moisture freezing.

Recent results of the continuous temperature monitoring at the Czech micrometeo-rological stations confirmed conclusions by Gosnold et al. (1997). Effect of the soil freeze/thaw events on the GST is reflected in the early section of the time series, presented in Figure 46. This diagram displays the air temperatures measured at 5 cm above the surface and the ground temperatures registered at depths of 2, 10, and 50 cm at Potucky station. The end of October and especially the beginning of December were characterized by the absence of snow and by the two episodes of the sharp fall of the air temperature well below 0°C. In the time intervals of strong air temperature decrease ground temperature at shallow depths of 2 and 10 cm have remained almost constant and close to 0°C and 1-1.5°C, respectively, illustrating the above-mentioned "zero-curtain effect" that occurs mainly due to latent heat released from the freezing of soil.

Fig. 46. Time series of air (at 5 cm height) and soil temperature changes at 2, 10, and 50 cm depth at Potucky station during October-December 2003.
Fig. 47. Behavior of the soil temperature at different depth levels below sand and grass surfaces at Prague-Sporilov station during freezing cycles of February 2006 (See Plate 2 of Colour Plate Section).

The data indicate that the soil freezing at Potucky station during October-December 2003 did not actually achieve even 10 cm depth. The ground temperature at the uppermost "active" layer is a complex result of the heat transfer from the frozen upper and undisturbed lower layer as well as the heat release from advancing freezing front. That time there was no snow cover at the station; thus, time series in Figure 46 reflect pure influence of the freeze/thaw processes on the GST. Irregular monthly air surface temperature variations are significantly attenuated at the depth of 50 cm and occur with time delay of days. Temperatures at that depth are lower than the highest positive air temperatures by approximately 3-4 K and may be higher than the lowest negative air temperatures by 8-10 K.

Figure 47 (See Plate 2 of Colour Plate Section) shows the behavior of the ground temperature under sand and grass surfaces during February 2006 at the Prague-Sporilov station. Due to heavy frosts and absence of snow in January, the subsurface temperature below both surfaces has dropped below the freezing point. Temperature at 20 cm depth was quite stable at 0°C and -0.3°C under the grass and the sand, respectively. The higher temperature under the grass occurs due to an insulation effect of the vegetation cover, which is mentioned above. In the first half of February, when the SAT was relatively low slightly oscillating around zero, the GST under both surfaces remained practically constant. Its sharp decrease between 2 and 5 cm depths was observed only between February 14 and 15 and was given by a similar drop of the SAT. During the second half of February, when the air temperature increased above zero, the subsurface temperature change under the sand surface generally repeated the SAT course. However, the phase changes of soil water substantially reduced the GST variations. The surface temperature variations vanished at the interface of the frozen and thawed soil layers that remained at zero temperature. Temperature at 20 cm depth was practically constant, which hints that all heat coming from the surface was spent in melting the soil water between 10 and 20 cm (Figure 47, left). Under the grass, where insulation of the surface and low thermal diffusivity of the soil slowed down the penetration of the surface warming, at all measured depth soil temperature remained close to 0°C. (Figure 47, right).

It should be emphasized that the effect of soil freeze/thaw cycles on the GST-SAT decoupling is only of seasonal importance. On the longer timescales it probably does not violate the air-ground temperature correlation in such a strong manner as it seems at a first glance. The widespread investigations of the timing, duration, and areal extent of this phenomenon by Zhang and Armstrong (2001) over the contiguous USA territory using passive microwave remote sensing revealed high variability of the freeze/thaw variations. The onset of soil freeze occurred generally in October-November, while its termination was in March-April. However, it does not mean that the near-surface layer was continuously frozen during this period. Measurements by Zhang and Armstrong (2001) have shown that the number of days with real surface soil freezing varied from several days to as long as 5 months. Majority of the regions experienced less than 60 days of the actual freezing and their occurrence was quite sporadic. Because of the combined effects of the snow cover and latent heat released by freezing, the soil moisture significantly changes with time; thus, this kind of disturbance vanishes during averaging over large temporal/spatial scales and probably cannot create a false systematic secular trend in the GST.

Some recent critiques of GST reconstructions were presented in the works by Mann and Schmidt (2003) and Mann et al. (2003). These authors have compared the SAT, GST, and snow cover trends simulated for the latter half of the twentieth century for terrestrial regions of the Northern Hemisphere by means of the GISS ModelE kind of the GCM family, similar to the one described in Section 2.4.4 of this chapter, and argued that the interpretations of the past SAT trends using GST reconstructions could be significantly biased by an influence of the snow cover during cold season. According to the calculations of the above authors, air temperatures have a dominant influence on ground temperatures only during warm season, while during cold period snow cover has significantly insulated the ground surface from the SAT changes. This process tends to exaggerate the role of warm season, thus providing a source of possible bias when comparing GST and SAT series (see also examples in Section 3.3, Chapter 3).

This conclusion was rejected in the works by Smerdon et al. (2004) and especially by Chapman et al. (2004), who have argued that the statement by Mann and Schmidt (2003) completely contradicts with the results of their monitoring experiments. According to Chapman et al. (2004), the source of discrepancy is an artificial sharp division of the years and corresponding temperatures into cold and warm season performed in the work by Mann and Schmidt (2003), while the nature of the conduction process makes the ground temperatures sensitive to continuous rather than to rapid or even seasonal variations. Separated seasonal anomalies are thus inappropriate for detection of the GST-SAT coupling on the long scales that are used in the climatologic studies. An analysis by Chapman et al. (2004) has proved that the GST-SAT tracking is almost perfect (correlation coefficient equals to 0.97), when temperatures are assessed on at least annual scale and thus summarize/compensate both summer and winter effects. This result is confirmed by the recent millennium long simulation of ground temperatures performed in the works by Gonzalez-Rouco et al. (2003, 2006), who used simulated subsurface T-z profiles forced by the GCM as a substitute of the real climate and applied inversion technique to reconstruct GST histories from the calculated profiles (for details see Section 2.4.4 of this chapter). Modeling results of these authors have proved the fact that the air and ground temperature variations are practically identical on centennial and longer timescales.

2.6.3 Effect of precipitation

In meteorology, precipitation means any form of water, whether liquid or solid, that falls from the clouds and reaches the ground. Precipitation is a major component of the hydro-logic cycle, and is responsible for depositing most of the fresh water on the Earth. It also represents the main component of weather. Precipitation occurs in a variety of forms, however, generally as rain and snow. An influence of the winter snow cover on the GST-SAT tracking has been discussed above. What about an influence of the summer rains? Summer soil temperature is controlled by the combined effect of air temperature variations and soil moisture content. An increase in rainfall during summer season would increase both the surface wetness and the soil moisture. This will result in more energy consumption for evaporation and thus cause cooling of the ground surface and soil. It is so-called soil moisture feedback (Yasunari et al., 1991; Matsuyama and Masuda, 1998). In principle, soil moisture feedback mechanism may explain soil cooling during summer, when air temperatures increase. Zhang et al. (2001) have detected clear negative correlation between monthly precipitation and soil temperature at 40 cm depth (greater monthly precipitation with lower soil temperature, and vice versa) in the 100-year long (1980-1990) meteorological time series measured at Irkutsk, Russia.

Existing monitoring experiments as well as the above-described numerical simulation by Mann and Schmidt (2003), Mann et al. (2003) have shown that anyhow the air temperatures have a dominant influence on ground temperatures during the warm season. Thus, revealed by the same monitoring experiments possibility of the GST-SAT decoupling during warm periods of the year likely represents the far weaker effect than the decoupling of both temperatures in the cold season. While winter snowfall as well as seasonal freezing and thawing cycles are the main reasons responsible for the breaking of the one-by-one GST-SAT tracking during cold season, the rainfall can produce definite air-ground temperature differences at warm conditions.

Precipitation is one of the main factors determining the subsurface thermal regime because it affects the amount of soil moisture and therefore the amount of energy removed from the soil by latent6 and sensible7 heat fluxes. Figure 48 illustrates the types of energy balance at dry and moist ground surfaces. Expression Q = H + LE + G combines the components of the total heat balance in the air-ground system, where Q is available net radiation, H the sensible heat, LE the latent heat, and G the subsurface (ground) heat. The latter three variables represent the major categories of the total energy use. Sensible heat is strongly conditioned by the temperature gradient between ground surface and air, while the ground heat flux depends on a similar gradient between the surface and the subsurface. When the evaporation of the water takes place, the positive latent heat flux (LE+) occurs in the ground surface. This means that the surface loses energy to the air above. Thus, evaporation is a cooling process for the ground surface.

6Latent heat flux is the flux of heat from the Earth's surface to the atmosphere that is associated with the change of states or phase, e.g. with evaporation of water at the surface. Term "latent" is used because this energy does not increase the temperature of water molecules and is only stored in molecules to be released later during the condensation process.

7Sensible heat is the heat energy transferred between the Earth's surface and air when there is a temperature difference between them. According to the direction of the temperature gradient, this flux can warm the ground or air.

Moist surface Moist surface Dry surface

Fig. 48. Types of energy balance at the ground surface during warm period.

Moist surface Moist surface Dry surface

Fig. 48. Types of energy balance at the ground surface during warm period.

Various processes operating in the vicinity of the ground-air boundary influence the heat flux balance that cause corresponding changes in both the SAT and GST. Thus, in some cases the GST may reflect the energy balance at the Earth's surface, rather than the SAT variations. Two of the diagrams in Figure 48 illustrate the differences between heat fluxes for dry and moist surfaces occurring in the daytime. In both cases equal amounts of incoming heat Q* are conducted down to the subsurface. No latent heat transfer occurs without available moisture content, which means the absence of the latent flux from the dry surface. Most of the energy Q* is transferred by sensible flux (H + ) that results in warmer air temperatures above dry surface. At moist surface the share of sensible heat is lower, while significant amount of the available radiant energy is used for evaporation of surface water, thus creating relatively cooler air than that above dry soil. It should be mentioned that latent heat flux always has priority. If moisture is available for evaporation, this process (LE+) takes preference over warming of the air (H + ) and /or warming of the ground (G+). At night the processes reverse. Because the thermophysical properties of the subsurface rock, such as thermal conductivity and heat capacity, depend on the water content, the rainfall can influence not only the energy balance of the ground surface-air system, but also thermophysical and/or hydrological characteristics of the ground. Regions with low porosity and permeability will likely not be significantly affected, while less consolidated medium will experience more pronounced changes.

Primarily influence of the precipitation on the GST-SAT coupling occurs on the very short timescales (directly during and after rain events) through, e.g. advective transport of heat by falling water that may significantly contribute to the development of shallow subsurface temperatures. For example, boreal forest sites in interior Alaska and NW Canada exhibited rapid but short soil warming of several degrees in response to summer precipitation events (Hinkel et al., 1997). Figure 49 displays temperature difference between the ground surface and 2 cm depth measured in dry and rainy periods at Prague-Sporilov. During 10-day interval with no rain the differences have shown quasi-periodic oscillations with maximum positive values in the daytime and negative values at night (compare with Figure 40 of this chapter). The range of variations reached ~9K. The temperature differences were negative after rain events both during

8-Oct 9-Oct 10-Oct 11 -Oct 12-Oct 13-Oct 14-Oct 15-Oct 16-Oct 17-Oct

Fig. 49. Time series of temperature difference between ground surface and 2 cm depth temperature at Prague-Sporilov station; comparison of rainy decade with dry period. Top panel also shows total rainfall amount.

8-Oct 9-Oct 10-Oct 11 -Oct 12-Oct 13-Oct 14-Oct 15-Oct 16-Oct 17-Oct

Fig. 49. Time series of temperature difference between ground surface and 2 cm depth temperature at Prague-Sporilov station; comparison of rainy decade with dry period. Top panel also shows total rainfall amount.

day and at night (air temperature at wet surface is lower than that at 2cm depth, e.g. June 30-July 2 and/or July 5-6; Figure 49, top). Its variations were significantly reduced and ranged within only ~3-4K. On the other hand, evaporation proceeds relatively quickly; thus, depending on the rain strength the "dry" regime was restored 1-2 days after rainfall.

The role of precipitation appears to be far more important on seasonal and/or annual scales because of its possible seasonal persistence. In the mid-latitudes snowfall and soil freezing (especially the latter process) represent generally sporadic events. As mentioned in the previous section, their effect on the GST-SAT decoupling is not perceptible already under decadal averaging. On the contrary, rainfall occurs more regularly during much of the summer and its annual distribution remains preserved for the long periods. The Prague site represents typical example of the seasonal timing of precipitation. Daily precipitation at Prague has no significant linear long-term trend. However, it has revealed a certain seasonal character that was preserved for a longer time (Bodri et al., 2005); the wetter season falls during May-August period and the precipitation minimum occurs in winter. This conclusion is confirmed by the meteorological observations in the nineteenth to twentieth centuries on the monthly scale of aggregation (Figure 50). The increase in precipitation in "wet" years occurs mainly due to its significant growth in summer period, when the actual monthly amounts of precipitation can be a few times higher than the average. Prevalence of summer precipitation is a specific feature of the hydrologic cycle in the Czech Republic and is preserved for at least 180-year long period. Similar persistence of the annual distribution seems to be common feature of the precipitation in many regions. According to Lin et al. (2003), there were no significant changes in the seasonal distribution of precipitation"/>
Fig. 50. Averaged monthly precipitation at Prague-Ruzyne. The 2214 months between 1805 and 1989 were used for averaging. (Data source: The Global Historical Climatology Network, GHCN 1;

in the USA and Canada for at least twentieth century. The influence of the precipitation on the GST-SAT relationship may be even more perceptible in the tropics, where evapo-transpiration8 is potentially significant year round.

Except for the cold season GST-SAT decoupling, the above-cited studies by Smerdon et al. (2004, 2006), which generalized the results of temperature monitoring at four microclimatic stations, have detected the GST-SAT discrepancies during warm period that occur as a result of the changes in surface energy balance caused by the rainfall. Precipitation spans a wide range of 52-115cm/year at four investigated locations with significantly different amounts of the rain and snow related parts. Thus, four data sets reflect the local climate conditions that may represent a base for comparison. Figure 51 shows time series of daily averages of air temperature (at 5 cm height) and soil temperatures at 5, 100, 200 and 500 cm depth measured at station Prague-Sporilov during the "rainy" year 2000. The amount of precipitation is presented on the histogram below. Detectable high-frequency oscillations of the air temperature record in summer (Figure 51) are caused mainly by the rains that change the moisture content of the soil and correspondingly both latent and sensible heat flow at the ground surface. General influence of precipitation at short timescales is to increase latent heat flux and to decrease sensible heat flux. As seen, rainfall events are accompanied by corresponding changes of both air and ground temperatures. The main observation about summer GST-SAT interrelation is that the rainfall does not cause total decoupling of both temperat-ures similar to that occurring in the winter due to snow cover and freezing/thawing cycles. The air

8Evapotranspiration represents the sum of evaporation and plant transpiration. The former process accounts for the movement of water to the air from the surfaces, while the latter process is responsible for the water movement within plants and for its loss through plant leaves. Types of vegetation and land use, percentage of soil cover, level of plant maturity as well as meteorological variables (solar radiation, temperature, humidity, wind) are among the factors that affect evapotranspiration.

Air temperature at 5 cm

Air temperature at 5 cm

Was this article helpful?

0 0

Post a comment