Permafrost monitoring in the Hovsgol mountain region, Mongolia

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112,, doi: /2006jf000543, 2007 Permafrost monitoring in the Hovsgol mountain region, Mongolia Anarmaa Sharkhuu, 1 N. Sharkhuu, 1 Bernd Etzelmüller, 2 Eva S. Flo Heggem, 2 F. E. Nelson, 3 N. I. Shiklomanov, 3 Clyde E. Goulden, 4 and Jerry Brown 5 Received 2 May 2006; revised 21 March 2007; accepted 27 April 2007; published 28 June [1] The Hovsgol mountain region (49 52 N, E) of northern Mongolia contains widespread mountain permafrost and comprises the southern fringe of the Siberian continuous permafrost zone. In this paper, we report our initial monitoring of permafrost by means of measuring ground temperatures and active layer thickness in boreholes and some cryogenic processes under the influence of climate warming and human activities in the region. The average rate of increase in mean annual permafrost temperatures is from 0.2 C to 0.4 C per decade. Permafrost has been degrading more intensively during the last 15 years (since 1990s) than during the previous years (1970s and 1980s). Recent degradation of permafrost under climate warming in the Hovsgol mountain region is generally more intensive than in the Hentei and Hangai Mountain regions. Moreover, livestock grazing in some local areas accelerates degradation of permafrost due to loss of vegetation cover. Year-round temperature recordings by data loggers placed beneath different vegetation covers showed marked differences in active layer thickness and ground temperature. Citation: Sharkhuu, A., N. Sharkhuu, B. Etzelmüller, E. S. F. Heggem, F. E. Nelson, N. I. Shiklomanov, C. E. Goulden, and J. Brown (2007), Permafrost monitoring in the Hovsgol mountain region, Mongolia, J. Geophys. Res., 112,, doi: /2006jf Introduction [2] The Hovsgol mountain region (49 52N and E) is within the territory of Hovsgol Province, northern Mongolia (Figure 1). In the northwest, the region encompasses the Darhad depression, which is surrounded by high mountain ranges (>3000 m above sea level). The region also incorporates the mountain forest steppe plateau in the southwest, mountain taiga in the northeast, and the basin of the Eg-Delger River in the southeast. This area contains widespread mountain permafrost, and comprises the southern fringe of the Siberian continuous permafrost zone [Goulden et al., 2006; Sharkhuu, 2006]. [3] Global climate warming has induced degradation of permafrost at many locations in recent years [e.g., Jorgenson et al., 2001; Gavrilova, 2003]. According to Gavrilova s [2003] estimate the climate warming trend in central and eastern Yakutia (Russia) is 1.5 times as high as in southern Siberia and 3 times as much as in Mongolia. Particularly, the climate warming during the last years in the 1 Geo-Ecology Institute, Mongolian Academy of Sciences, Ulaanbaatar, Mongolia. 2 Department of Geosciences, University of Oslo, Oslo, Norway. 3 Department of Geography, University of Delaware, Newark, Delaware, USA. 4 Institute of Mongolian Biodiversity and Ecological Studies, Academy of Natural Sciences, Philadelphia, Pennsylvania, USA. 5 International Permafrost Association, Woods Hole, Massachusetts, USA. Copyright 2007 by the American Geophysical Union /07/2006JF Hovsgol region has been more intense than in other regions of Mongolia. According to data from the Hatgal weather station, since 1963 the mean annual air temperature has increased by 1.68 C. While the air temperature increase amounted to 0.61 C in the period from 1963 to 1986, it was 0.68 C between 1990 and The air temperature in winter has increased more than during the other seasons [Nandintsetseg et al., 2006]. [4] The impacts of intensive human activities on permafrost conditions are, however, pronounced in local areas and sites, and discriminating between the effects of climatic change and human impacts is a major challenge facing geocryologists [Nelson, 2003]. Most permafrost in the Hovsgol region is at temperatures close to 0 C at depths of m and is therefore vulnerable to climate change and human activities in terms of vegetation disturbances. Here, the effect of human activities on permafrost is caused mainly by livestock grazing, forest cutting, forest fires, mining exploitation and engineering works. [5] In recent years the impacts of climate warming and human activities on permafrost degradation in the region have been monitored within the framework of several international projects. Long-term (10 38 years) monitoring of permafrost in the region is conducted within the framework of the Circumpolar Active Layer Monitoring (CALM) and the Global Terrestrial Network for Permafrost (GTN-P) projects [Brown et al., 1997, available at Brown et al., 2000; Burgess et al., 2000; Nelson et al., 2004]. In addition, short-term (3 4 years) monitoring of permafrost is carried out by the Hovsgol Global Environment Facility/World Bank (GEF/WB, 1of11

2 Figure 1. Schematic map of permafrost and location of monitoring site in the Hovsgol mountain region. Acc. continuous denotes an accumulative type of landform (valleys and depressions), which in underlain by continuous permafrost, acc. discontinuous denotes the accumulative type of landform, underlain by discontinuous permafrost, acc. sporadic and acc. isolated denote accumulative type of land, underlain by sporadic and isolated permafrost, respectively. The denudation type of landform (watershed and slopes) is shown darker than the corresponding accumulative type. Project since The Hovsgol Project study area includes six valleys (Borsog, Dalbay, Sevsuul, Noyon, Shagnuul and Turag), which are aligned in parallel from south to north, between Borsog and Dalbay valleys in the south and Shagnuul and Turag valleys in the north, along the northeastern shore of Lake Hovsgol at 50 N 100 E (see Figure 1). At present, the Hovsgol GEF/ WB Project area is designated as part of the International Long-Term Ecological Research (ILTER) site at Lake Hovsgol [Tsogtbaatar and Goulden, 2000; Goulden et al., 2005a]. Cryogenic processes and phenomena are monitored in the project area, and in the Darhad depression to the west. [6] The objectives of the permafrost monitoring projects in the region are to estimate rates of degradation of permafrost during the last years under the influence of climate warming, and to document and interpret the impacts of human activities on permafrost degradation. In this paper, we present some initial results of permafrost monitoring implemented in the entire Hovsgol region. 2. Permafrost Distribution [7] Siberian thick continuous permafrost with low temperatures embraces a considerable part of the Eurasian territory. Since Mongolia is located in the southern fringe of the Siberian permafrost zones, most of the permafrost in Mongolia is at a temperature close to 0 C, and thus thermally unstable. Permafrost occupies almost two thirds of Mongolia, predominantly in the Hentei, Hovsgol, Hangai, Altai Mountains and surrounding areas. The territory is characterized by mountain and arid land permafrost, from isolated (<10%) and sporadic (10 50%) to discontinuous (50 85%) and continuous (>85%) in its extent. Hovsgol is the region with the most prevalent permafrost in Mongolia. 2of11

3 Figure 2. Methodology of permafrost monitoring: (a) parameters of permafrost monitoring and (b) borehole instrumentation. [8] In parts of the Hovsgol region in which permafrost is widespread ( continuous or discontinuous, Figure 1), taliks are found on steep south facing slopes, under large river channels and deep lake bottoms, and along tectonic fractures with hydrothermal activity. In areas of sporadic and isolated permafrost, frozen ground is found only on north facing slopes and in fine-grained and moist deposits. Average thickness and mean annual temperature in the areas of widespread permafrost is m and 1 to 2 C in valleys and depressions, and m and 1 to 3 C on mountains, respectively [Sharkhuu, 2006]. Permafrost in the region has low to moderate ice content in unconsolidated sediments. Ice-rich (content is more than 20% by weight) permafrost is characteristic of lacustrine and alluvial sediments in valleys and depressions. In addition, ice wedges are found in the Darhad depression. The thickness of the active layer is 1 3 m in fine-grained soils and 4 6 m in coarse material. Cryogenic features such as frost heaves, cracks, icings, thermokarst terrain, solifluction lobes and sorted polygons are abundant in the region [Gravis et al., 1978; Sharkhuu, 2006]. [9] The project area is underlain by discontinuous permafrost, the thickness of which reaches 5 20 m in valley bottoms and more than 20 m on forested north facing slopes. Mean annual permafrost temperatures at depth of 10 m range from 0.3 C to 1.5 C. There is no permafrost on south facing slopes without forest [Etzelmüller et al., 2006; Heggem et al., 2006]. 3. Methods [10] Permafrost in the Hovsgol region has been monitored since At many locations, long-term CALM and GTN- P programs are based on ground temperature measurements in shallow to deep boreholes. Monitoring of frost heaving, thermokarst and icing is based on leveling measurement with Russian leveling equipment of model type HB-1. [11] Since 2002, permafrost conditions in the Hovsgol GEF/WB Project area have been studied based on data from permafrost temperature measurements in shallow boreholes, DC resistivity tomography measurements [Etzelmüller et al., 2006], ground surface temperature (GST) recordings by more than 30 miniature temperature data loggers (MTDs) [e.g., Heggem et al., 2006] and experimental observations for the role of vegetation on GST Boreholes [12] Altogether 17 boreholes at eight sites were drilled during the last 2 5 years in the Hovsgol region. Borehole depths are 5 10 m for five boreholes, m for seven boreholes, and m for five boreholes. In addition, in 2006 we have found 135 and 200 m deep geological boreholes, which were drilled in the mid-1980s in Burenkhan and Ardag mountain areas, and used them for permafrost monitoring. During the last 2 5 years, most of the boreholes with a depth of 10 to 15 m were redrilled and instrumented at points where previous deeper temperature measurements in the old ones were made years ago in Hovsgol region. Except for geological deep boreholes, all the new monitoring boreholes are prepared by dry drilling. All boreholes are situated in natural conditions without thermal disturbance. Each borehole is designed to prevent air convection and to protect against damage by passing people. All the redrilled boreholes are cased by parallel steel and plastic pipes of 3 5 cm diameter with the mouth of the pipe at a depth of about cm and covered by soil (Figure 2b). The empty space outside of casing pipes in all the boreholes is filled by fine sands Ground Temperature Measurements [13] The primary parameters being monitored are active layer thickness and mean annual permafrost temperature at the (10 15 m deep) level of the zero annual amplitude, as well as the temperature gradient of the permafrost (see Figure 2a). Temperature measurements in the all boreholes are made using identical thermistors at corresponding depths, and are carried out on approximately the same dates each year. We use movable thermistors string, prepared at the Geothermal laboratory of the Melnikov Permafrost Institute, Siberian branch, Russian Academy of Science. The accuracy of 3of11

4 Figure 3. Topography map of Dalbay valley. Dots indicate locations of data loggers which are used for GST measurements in Dalbay valley. Some of them are overlaid on each other due to short distance between them, particularly a1 and a2: a1, grass-mowed plot; a2, plot with dense grass; b1, plot under 1.8 m high shrub; b2, plot under 0.5 m high shrub; c1, plot in sparse forest; c2, plot in dense forest; d1, plot without moss cover; and d2, plot with 10 cm thick moss cover. temperature measurements by Russian calibrated thermistor (MMT-4 model) is ±0.02 C. Thermistor resistance is measured by a multimeter (MB-400 model). In all boreholes the active layer thickness is determined by interpolation of ground temperature profiles, obtained by boreholes temperature measurements in late September and early October. In addition, temperature data loggers and thaw tubes are installed in the Hovsgol project (Borsog M12A, Borsog M12B, Dalbay M12, Noyon M12E, Shagnuul M12C), Hatgal (M11) and Sharga (M8), boreholes since In addition, we have installed HOBO H8 in the Tsagaan-nuur (M13) and Darhad (M13A) boreholes since Accuracies of measurement of ground temperatures by using UTL-1 miniature and four-channel HOBO H8 data loggers were ±0.25 C and ±0.5 C, respectively. The interval time of temperature recordings by data loggers is 90 min Ground Surface Temperatures (GST) in Experimental Plots [14] In order to study the thermal insulation effect of vegetation cover, year-round (October 2004 to October 2005) recordings from eight miniature temperature data loggers UTL-1 are used. The data loggers were set under the ground surface (at a depth of 2 5 cm) at eight plots which differed in vegetation cover; five of them (c1, c2, d1, d2 and b2) are on the north facing slope of Dalbay valley near the forest edge, but the remaining three (a1, a2 and b1) are in Dalbay valley bottom (see Figure 3). Temperature data loggers were placed in plots of which vegetation cover was kept in natural condition but in the a1 plot grass cover was mowed. Plots are set in pairs and can be compared with each other. The plots are grass-mowed bare plot (a1) and plot with dense grass (a2), plot (b1) under 1.8 m high dense shrub (Salix sp.) and plot (b2) under 0.5 m high dense shrub (Salix glauca), plots in sparse larch forest (c1) and in dense young larch forest (c2), and adjacent two plots, a plot (d1) without moss cover and another plot (d2) with 10 cm thick moss cover (Rhytidium rugosum and Aulacomnium. sp). All the paired plots are close to each other (distance is 150 m between c1 and c2, 2 m between d1 and d2, and 4 m between a1 and a2), and there is no significant difference in the terrain characteristics between pairs but the plot under 1.8 m high shrub (b1) and plot under 0.5 m high shrub (b2) are set 1200 m away and located in Dalbay riparian zone and Dalbay north facing slope, respectively. Therefore we cannot compare these two plots directly. In addition, we measured active layer thickness, ground temperature (at 1 m depth) and soil moisture (at 0.5 and 1 m depths) in the grass-mowed plot (a1) and the one under dense grass (a2). These are two experimental plots, 4 4 meter square, at the Dalbay study site. The plot a2 was allowed to remain under the natural vegetation cover while in a second plot, a1, vegetation cover was removed periodically by cutting the plants. Ground temperature was measured by the same movable thermistor described above. Percent of soil water content is determined on a dry mass or gravimetric basis. 4. Results and Discussion 4.1. Permafrost Temperature Changes Under Climate Warming [15] All observational data of the selected Sharga, Hatgal, Tsagaan-nuur, and Burenkhan long-term monitoring sites (Figure 1) show a clear trend of increase in mean annual 4of11

5 Figure 4. Long-term change in permafrost temperatures under the influence of climatic warming, measured in (1) Sharga, (2) Hatgal, (3) Tsagaan-nuur, and (4) Burenkhan monitoring boreholes. permafrost temperatures at a depth of m during the measurement period (Figure 4 and Table 1), varying between and Cyr 1. All borehole profiles show a deviation of the ground temperature from steady state, and toward warmer temperatures (see Figure 4). The rate of increase in mean annual permafrost temperatures in the study area varies from 0.02 C to 0.04 C yr 1, depending on local landscape and ground conditions such as thermal conductivity [Sharkhuu and Sharkhuu, 2005]. Estimated thermal conductivities of frozen ground are determined by mean of using tabulated nomogram for thermal conductivities which are calculated based on the relationship between ice content and partial density of ground [Melinikov and Tolstihin, 1974]. The highest rate, C yr 1 with an increasing trend (r 2 = 0.91, P < ) in permafrost temperatures during the last 11 years, is observed in the Burenkhan 1 borehole on a mountain slope composed of high thermal conductivity bedrock (Figure 5). Compared to Tsagaan-nuur borehole (measured since 1989), the Figure 5. Trend of increase in mean annual permafrost temperatures in Burenkhan monitoring borehole. relatively low rate of increase of temperature in the Sharga borehole (measured since 1968) shows that permafrost is degrading more intensively during the last 15 years than during the previous years ( s). Moreover, mean annual ground temperatures in the Burenkhan mountain area have increased by 0.27 C per decade on south facing slope, 0.19 C on north facing slope, 0.23 C in the upper watershed and 0.11 C in the valley bottom [Sharkhuu, 1998]. Thus a rate of increase in mean annual temperatures in a bedrock with higher thermal conductivity is more than in an unsolidated sediments with lower thermal conductivity, as well as the rate on south facing slope is more than on north facing slope. [16] The increasing permafrost temperature gradient with depth can be addressed as an indicator of recent and former degradation of permafrost [Harris and Haeberli, 2003]. Increasing temperature gradients are demonstrated in the Darhad and Burenkhan 2 deep boreholes (Figure 6 and Table 2). The ground of the Burenkhan 2 borehole is in general homogenous bedrock, composed of dolomite and limestone. The ground of the Darhad borehole is lacustrine deposit, composed of silt and clay, interlaying Table 1. Rate of Permafrost Degradation in Sharga, Hatgal, Tsagaan-nuur, and Burenkhan Monitoring Boreholes a Borehole Sharga Hatgal Tsagaan-nuur Burenkhan 1 Latitude, N Longitude, E Elevation, m asl Terrain wide valley bottom lake-river terrace lake terrace north facing slope Predominant ground gravelly sand and loam sandy gravel lacustrine silt and clay bed rock Ice content by weight medium (14%) low (9%) high (40%) low (4%) Thermal conductivity, W (m K) Particular density, kg m Depths that data loggers H8: at 0, 3, H8: at 1, 3, H8: at 0, 2, 10, and no data logger are installed 6, and 10 m 6, and 10 m 15 m since 2005 Measured dates 21 Sep 1968/4 Oct Oct 1983/20 Sep Aug 1989/5 Oct Aug 1987/5 Oct 2004 Active layer thickness h, cm 270/ /470 no data/ /420 Temperature t at depth 10 m 2.35/ / 0.49 NA/NA NA/NA 15 m NA/NA NA/NA 3.57/ / 0.35 Increase in temperature t per decade, C a Abbreviations: asl, above sea level; NA, not available. 5of11

6 Figure 6. Change in permafrost temperature gradients with depth, measured in Darhad depression and Burekhan mountain boreholes. very thin layer of fine sand. We calculated the difference between two mean annual surface temperatures, extrapolated from the upper thermal gradient (20 50 m in Darhad and m in Burenkhan 2 boreholes) and deeper thermal gradient (50 80 m in Darhad and m in Burenkhan 2 boreholes). The Darhad borehole drilled in lacustrine sediments showed a MAGST increase of 0.67 C, while this value was 0.81 C at the Burenkhan 2 mountain borehole, which is drilled in bedrock (Table 2). Deep penetration of warming temperatures in the Burenkhan 2 borehole is caused by dry, dense and high thermal conductivity (3 WmK 1 ) of bedrock (mainly dolomite) in the borehole. The shallow penetration of warming temperature is characteristic for thick, fine lacustrine sediments with relatively lower thermal conductivity (<2 W mk 1 ), lower density, higher ice water content and thus considerable lower thermal diffusivity. The borehole temperature profile showed a major thermal gradient shift at 40 m and m depth at Darhad and Burenkhan, respectively. The thermal gradient change could be caused by a major change of thermal conductivity in the bedrock or subsurface material, which is supported by the rather abrupt changes of the thermal gradient in the borehole. The borehole log from the Burenkhan site showed a shift from dolomite to dolomite containing phosphate at 90 m depth; however, there are no independent measurements of possible conductivity changes. We have therefore calculated how fast a warming temperature pulse would penetrate into the ground within a given depth, applying the equation [from Williams and Smith, 1989] "!!# z z DTðz; tþ ¼ DT s erf erf ½4kðt lþš 1=2 ð4ktþ 1=2 where z is depth, t is time, DT s is the difference of a temperature pulse from a steady state at the surface, k is the thermal diffusivity, l is the period of the temperature warming pulse, and erf is the Gaussian error function. When assuming realistic values for thermal diffusivity of m 2 s 1 and m 2 s 1 for Burenkhan and Darhad, respectively, a similar pattern of temperature change with time are reached at m depth for Burenkhan and m depth for the Darhad depression, respectively. The maximum of temperature change is reached between 15 and 30 years after the start of the warming pulse in both boreholes in the Darhad depression. At Burenkhan, the same happens between 20 to 40 years (see Figure 7). This supports our hypothesis that the observed change in thermal gradient with depth in addition to lithologocal factors might be associated with a warming temperature pulse. The Darhad site would then represent a warming trend during the early 1980s, while the Burenkhan site would respond to warming trends during the early 1960s. Both warming episodes are recorded according to climate data from the weather stations at the Murun, Hatgal, and Rinchinlhumbe weather stations. [17] Comparison of increases in permafrost temperatures in Alaska and Siberian regions shows almost the same trend. The trends over the interval of late 1970s to middle 1990s are C yr 1 in Siberia [Pavlov and Perlshtein, 2006]. Trends of increase in permafrost table temperatures between 1987 and 2001 are 0.14 C yr 1 at the West Dock and Franklin Bluffs sites and 0.21 C yr 1 at the Deadhorse site, in northern Alaska [Romanovsky et al., 2003]. Estimated trends of the increase in mean annual permafrost temperatures over years in Mongolia [Sharkhuu and Sharkhuu, 2005] are C yr 1 in the Hangai and Hentei region, and C yr 1 in the Hovsgol region. Thus the comparison shows that permafrost in high latitudinal continuous permafrost zones (in Siberia and Alaska) is degrading more rapidly than at the southern fringe of continuous permafrost zone (in Mongolia). Meanwhile, the permafrost in the Hovsgol region is degrading more intensively than in the Hangai and Hentei regions Changes in Active Layer Thickness [18] Estimating the rates of active layer thickness increases in the boreholes is difficult and obviously depends on soil properties and moisture content at the sites. Nor- Table 2. Permafrost Temperature Gradients in Darhad and Burenkhan Deep Boreholes Borehole Darhad Burenkhan 2 Latitude, N Longitude, E Elevation, m asl Terrain dry plain of lake depression north facing 15 slope Predominant ground lacustrine silt and clay dolomites covered by 2 m thick debris Ice content by weight high (30%) low (2%) Depths that data loggers H8: 2, 5, 10 and no data logger are installed 15 m since 2005 Thermal conductivity, W/m K Particular density, kg/m Depth interval, m 20 50/ / Temperature gradient C/m 0.023/ /0.021 Mean surface temperature, 3.21/ / 1.75 extrapolated from each depth interval Warming temperature C of11

7 Figure 7. Temperature change with time, estimated at m in Darhad depression and at m in Burenkhan mountain boreholes. mally, active layer thickness varies about 10% between consecutive year measurements [e.g., Williams and Smith, 1989]. The active layer thickness measured in the Sharga and Tsagaan-nuur boreholes were very similar (Table 1), thus no major changes have happened. In contrast, the active layer depth in the Hatgal and Burenkhan 1 profiles changed with rates corresponding to cm per decade and appeared to be characteristic of areas with deep active layer. The low rate of change in active layer thickness in the Sharga and Tsagaan-nuur boreholes is probably due to icerich and fine grained sediments. However, it should be noted that we have no long-term monitoring data to estimate a real trend of increase in active layer thickness for this region. [19] In particular, by the ground temperature interpolation, the active layer thickness at the Hatgal monitoring site, composed of sandy gravel with relatively low ice content, was 3.6 m in 1969, 4.0 m in 1983, and 4.7 m This shows also relatively intensive degradation of permafrost during the last years (Table 1). The rate of increase in active layer thickness, determined by a thaw tube in the Burenkhan 3 borehole is estimated to be 3.5 cm yr 1 during the last six years (Table 3). The estimated average increase in active layer thicknesses are varies in the range of cm yr 1 in the Hangai and Hentei regions, and cm yr 1 in Hovsgol region [Sharkhuu and Sharkhuu, 2005]. Compared with Alaska and Siberia, active layer thickness in the Hovsgol region varies greatly from year to year due to deep seasonal freezing and thawing of the ground. A difference between active layer thicknesses of two years in bedrock and gravely sands with high thermal conductivity reaches 5 15 cm. Even a trend of increasing of active layer thickness is observed, due to variation of active layer thickness year to year and place to place, we cannot conclude that active layer thickness is increasing significantly in recent years. In Alaska and Siberia, significant positive trends of increase in active layer thickness have not been observed, too [Brown et al., 2000; Romanovsky et al., 2003] Some Cryogenic Processes as a Result of Permafrost Warming [20] The effect of increased ground ice melt often associated with climate warming is visible in the terrain. The widespread thermokarst lakes, depressions, hollows, and intensive thermoerosional riverbanks in the Darhad depression are direct indicators of ancient and recent degradation of permafrost due to climate change; there are no local or human disturbances here. The vertical extent of some thermokarst depressions and thermoerosional riverbanks at Sharga ganga reaches m, although the average is 3 7 m. Small subsurface cavities on the permafrost table are formed as a result of melting ice wedges. We have observed ice wedge polygons on the land surface and ice wedge bodies in steep exposures (outcrops) of thermoerosional river banks. Large animals (yaks and horses) have fallen into deep (3 m) surface cavities and died of exposure. This Table 3. Active Layer Thickness in the Burenkhan Borehole 3 a Parameter Value Latitude, N Longitude, E Elevation, m asl 1457 Terrain north facing gentle slope near valley bottom with poor grass Predominant ground 0 10 m, gravelly loam and sand Thermal conductivity, W/m K 1.56 Measured year 1987, 1997, 2000, 2001, 2002, 2003, 2004, 2005 Active layer thickness, cm 310, 330, 325, 340, 345, 348, 352, 342 a There is no data logger in the borehole; the active layer thickness is measured by thaw tube. 7of11

8 is evidence of subsurface cavities formed as a result of melting ice wedges or degradation of permafrost under climate warming. In the 1970s, Tsoidon Lake, a water body about 2 km in diameter, disappeared due to thermoerosional changes in the Hodon river channel leading to the lake. The swampy bottom of the lake hollow has dried progressively during the last years. At present, there is dry steppe bottom in the hollow, similar to that in Yakutian alases. We consider that intensive thermoerosion processes lead sometimes to large changes in local topography. [21] At present-day, active thermokarst processes are observed everywhere in the depression. Landforms or phenomena of the active thermokarst processes are characterized by thermokarst lake and hollow with steep shore of 1 5 m in height. Sometimes, we observe fresh cracks along the shore. The cracks are formed as a result of thaw settlement of ice-rich sediments. However, due to lack of data on thermokarst age, we are unable to document when the permafrost started to degrade. Thus we have started a monitoring campaign to estimate the rate of thaw settlement at some monitoring sites. [22] On the basis of visual observations, we revealed that in recent years there are many changes in dynamics of cryogenic processes within the valleys along the northeastern shore of Lake Hovsgol. The changes in dynamics of cryogenic processes in the eastern shore of the Lake Hovsgol are might caused by human activities in the context of climate change. At present, there are almost no active cryogenic processes in the Turag and Shagnuul valleys due to intensive degradation of permafrost under influence of livestock grazing, which appears to have reduced soil moisture and caused permafrost thaw with an increased active layer thickness. Livestock grazing is a major factor influencing the disappearance of some cryogenic processes and phenomena. We have observed some inactive solifluction terraces and lobes, which are traces of ancient solifluction processes. [23] In contrast, widespread permafrost and suprapermafrost waters, shallow active layer and wet land with dense vegetation cover in Borsog, Dalbay and Sevsuul valley bottoms create favorable conditions for developing active cryogenic processes and phenomena. These conditions are preserved owing to low impacts of humans on permafrost degradation. Active seasonal frost mounds (hydrolocalists) are characteristic of the valley bottoms. Results from monitoring of frost heave and thaw settlement indicate that the location and height of seasonal frost mounds are variable from year to year. Their height has reached m during the winter. The depth of summer subsidence and thawing on mounds varies from 0.3 to 1.0 m. Active thermokarst, thermoerosion, icing, and solifluction lead to considerable changes in terrain, landform, and hydrologic conditions. The formation of a 3 4 m deep thermoerosion ravine on a gentle north facing slope on Borsog Mountain may be caused by human activities related to disturbance of vegetation cover during road construction Possible Permafrost Degradation Under the Influence of Human Activities [24] Even in the Hovsgol region climate change is a major factor influencing permafrost; but some changes are caused by localized human activities such as forest fire, road construction and livestock grazing. We carried out some observational studies in the Hovsgol Project area, including six valleys (Figure 1). In this area, human activities are limited primarily to pasture grazing. Dalbay and Borsog valleys in the south experience little or no pastoral use. Grazing is more intensive in Turag and Shagnuul valleys in the north [Goulden et al., 2005a, 2005b; Sharkhuu, 2006] (see Table 4). Such gradual grazing allows us to carry out studies of permafrost change induced by the effect of climate change, superimposed on impacts of grazing. [25] Initial data collected during short-term monitoring of permafrost in six valleys along the northeastern shore of Lake Hovsgol, illustrated in Table 4, indicates that active layer thickness varies from 1.4 m in Dalbay valley in the south to 4.8 m in Turag valley in the north. Permafrost temperatures at 10 m depth varies from 1.25 C in Dalbay valley to 0.42 C in Turag valley. Borehole measurements indicated that there was no permafrost on the Turag river floodplain in areas with sparse grass cover. Compared to Turag valley, the present shallow active layer and low permafrost temperatures in Dalbay valley are apparently insulated from thaw by the vegetation plant cover. This suggests that the loss of plant cover due to livestock grazing leads to an increase in active layer thickness and ground temperature and a decrease in water content. Consequently, we conducted preliminary experimental observations on the insulation effect of vegetation cover, based on differences in ground surface temperatures (GST). [26] In Dalbay valley, we developed eight plots with different treatments: grass-mowed bare plot (a1) and plot with dense grass (a2), plot (b1) under 1.8 m high dense shrub (Salix sp.) and plot (b2) under 0.5 m high dense shrub (Salix glauca), plots in sparse larch forest (c1) and in dense young larch forest (c2), and plot (d1) without moss cover and plot (d2) with 10 cm thick moss cover (Rhytidium rugosum and Aulacomnium. sp) (see Figures 3 and 8). [27] The experimental results from the grass-mowed plot (a1) and plot with dense grass (a2) demonstrates that at one plot under the natural vegetation cover (a2), conditions of a shallow active layer (56 cm in mid-july, 122 cm in early October), low soil temperature (0.15 C at 1 m depth in early October) are maintained and vegetation cover protects soil moisture from evaporative losses. Soil moistures were markedly different in the two plots at depths of 0.5 and 1 m. In contrast, the plot (a1) subjected to simulated grazing has a thicker thawing depth (73 cm in mid-july, and 155 cm in early October), higher soil temperature (0.5 C in early October), and lower soil moisture. Soil moisture content in the plot with dense grass (a2) was 45% and 176% by weight at depths of 0.5 m and 1 m, respectively, while in the grassmowed plot (a1) it was 37% and 31% at depths of 0.5 m and 1 m, respectively. [28] The thermal insulation values for different vegetation cover (Table 5) show the mean winter GST (December, January, and February) was coldest in plots in sparse forest (c1) and grass-mowed (a1). Snow cover is a very important factor in reducing the amplitude of changes and increasing the mean winter surface temperatures [Sharkhuu et al., 2006]. The plot with 1.8 m high shrub (b1) is the warmest plot in winter time, owing to thicker snow cover (11.33 cm). As compared to the grass-mowed plot (a1), which is the warmest plot in summer time, mean summer surface tem- 8of11

9 Table 4. Active Layer Thickness and Permafrost Temperatures a Site Borsog Dalbay Turag Shagnuul Latitude, N Longitude, E Elevation, m, asl Land surface dense grass dense grass poor grass poor grass Total biomass, g/0.25 m ± ± ± ± 3.4 Livestock number by sheep Land from swampy flood plain swampy flood plain north facing foot slope in valley north facing foot slope in valley Predominant ground gravel and sand gravel and sand gravelly sand gravelly sand Water-ice content, wet % no data Date of measurements 6 Oct Oct Oct Oct 2005 Depth that data loggers are installed H8 at 0.03, 1, 3 and 5 m H8 at 0.05, 1.5, 5 and 10 m UTL-1 at 0.03 and 10 m H8 at 0.03, 2, 5 and 10 m Temperature at depth of 10 m no data Depth of active layers, m Estimated depth of permafrost bottom, m no data a In the Hovsgol GEF study site boreholes at Borsog and Dalbay valley sites with dense grass cover and at Shagnuul and Turag valley sites sparse grass cover due to intense livestock grazing pressure. Data of total biomass, including necromass, and livestock number are provided by Ariuntsetseg (Impacts of nomadic livestock on a semi-arid boreal steppe plant community of northern Mongolia, in The Dynamics of Biodiversity Loss and Permafrost Melt in Lake Hovsgol National Park, Mongolia, unpublished report, Geo-Ecology Institute, Mongolian Academy of Sciences, Ulaanbaatar, Mongolia, 2005). Water-ice content is determined by gravimetric basis, and the value is an average value for soil column. perature (June, July, and August) differences are estimated to be 2.2 C under dense grass (a2), 3.6 C under 50 cm high dense shrubs (b2), 3.8 C under sparse forest (c1), 4.9 C under dense forest (c2) and 5.0 C under 180 cm high bushes (b1) at the Dalbay observation site. Besides, differences between mean summer surface temperatures under 10 cm thick moss cover (d2) and in the plot without moss cover (d1) is 6.4 C (see Table 5). The linear regression estimation of daily temperatures under different vegetation cover and air temperature at Dalbay shows that the presence of thick moss is a major factor in insulating the ground from the heat; the slopes of the regression of temperature under moss cover against air temperature, in summer and winter, are the lowest. The difference of slope of winter temperature regression (from November 1 to March 31) under different vegetation cover does not only reflect vegetation cover difference but is also affected by snow cover. Daily temperature data beginning May 1 until September 31, which is Figure 8. (top) Graphs demonstrating year-round temperature curves under different vegetation cover. (bottom) Illustration of ratio of daily TDD (thawing degree days) of surface and air. 9of11

10 Table 5. Mean Surface Temperatures in Single Plots With Different Vegetation Cover Located in Dalbay Valley Observation Sites a Mean Temperatures, C Summer Temperature Plots Data Loggers Under Winter Summer Annual Amplitude Difference, C FDD TDD TDD s /TDD a d2 10 cm thick moss b1 1.8 m high dense shrub c2 Dense forest c1 Sparse forest b2 50 cm high dense shrub d1 On moss surface a2 Dense grass a1 Grass mowed Air temperature at Dalbay a Measured by data UTL-1 loggers. The summer temperature differences are calculated on basis of difference between mean summer temperature beneath vegetation cover in each plot and temperature in grass-mowed plot. Here FDD is freezing degree day, and TDD is thawing degree days. TDD s and TDD a are surface and air thawing degree days. above 0 C, are used for summer temperature slope analysis (Table 6). [29] To compare plots with each other, we computed ratio of thawing degree days of surface (TDD s ) and air (TDD a ). The ratio of TDD of surface to air is consistent with the summer temperature reduction. The ratio of TDD of surface and air increases rapidly in the beginning of summer season and toward the end its rate does not change greatly (Figure 8). In the plot where vegetation cover was removed, the ratio was 1.14 because this site received direct solar radiation (see Table 5). The plots under 1.8 m high shrub (b1), 0.5 m high shrub (b2), in sparse forest (c1) and dense forest (c2) do not differ greatly from each other; their difference have resulted from the seasonal change (see Figure 8). In general, high soil surface temperature is also observed in heavily grazed places. These observations, though based on preliminary experiments, clearly demonstrate the importance of vegetation cover on ground thermal regime. 5. Conclusions [30] Permafrost in the Hovsgol mountain region is degrading, probably both due to climate warming and anthropogenic disturbances. Steepening of the near surface the temperature gradient in permafrost, and widespread occurrences of thermokarst and thermoerosion features, are direct indicators of recent and ancient degradation of permafrost in the region. [31] In general, permafrost in the Hovsgol mountain region is degrading twice as rapidly as in the Hentei and Hangai Mountain region [Sharkhuu and Sharkhuu, 2005], but less intensively than in eastern Siberia and the Trans- Baikal region [Gavrilova, 2003]. The rate of permafrost degradation in bedrock is greater than in unconsolidated sediments, in ice-poor sediments more than in ice-rich ones, and on south facing more than on north facing slopes. [32] Recent degradation of permafrost under the influence of climate warming leads to some changes in the ecological conditions. In particular, desertification and deforestation processes have been observed in the region s steppe and taiga zones, respectively. In general, our preliminary experimental results demonstrate that vegetation cover acts as an insulator in the summertime and reduces summer mean surface temperatures. Moss cover, dense grass, and forest are natural insulators, protecting soil moisture from evaporation and maintaining low soil temperatures. Thus the key to preserving permafrost and ecosystems, especially in the Hovsgol taiga zone, must be based on protection of the vegetation cover. [33] The program of monitoring and experimentation described in this paper represents an initial attempt to document permafrost evolution in the Hovsgol region. The extent and intensity of changes, their potential to affect agricultural and other economic activities in the region, and the interrelated nature of changes induced by climatic and Table 6. Coefficient of Determination and Parameter of the Linear Regression for the Relation of Air and GTS a Plots Data Loggers Under r 2 Value Annual Summer Winter Slope of the Regression Plots r 2 Value Slope of the Regression Plots r 2 Value Slope of Regression d2 10 cm thick moss d d b1 1.8 m high bushes c b c2 dense forest b d b2 50 cm high shrubs d c a2 dense grass a a d1 on moss surface c b c1 sparse forest b c a1 grass mowed a a a Data from October 2004 to October 2005 is used for calculation for the slopes (annual), from May 2005 to September 2005 is for slope of summer temperature, and November 2004 to March 2005 is for slope of winter temperature. Probability for r 2 and slopes of all regression analyses were significant (<0.0001). 10 of 11

11 human activities, indicate an urgent need to continue and expand permafrost observation programs in the region. [34] Acknowledgments. The first two authors of the paper thank the International and U.S. Permafrost Associations for their assistance in implementing CALM and GTN-P programs in Mongolia, including in the Hovsgol region. Permafrost studies in the Hovsgol GEF/WB Project area are supported financially by a grant from the Global Environmental Facility, implemented by the World Bank. The contribution and support of Department of Geosciences, University of Oslo, has been valuable. The first author is grateful for her visits to the universities of Oslo and Delaware and the 2005 AGU Fall Meeting in San Francisco, provided by a grant from the Netherlands Governmental funds and sponsored by the Hovsgol GEF/WB Project. The comments from reviewers were crucial for the improvement of the manuscript. The authors also thank to Robert Anderson, who undertook substantial editing of the manuscript. References Brown, J., O. J. J. Ferrians, J. A. Heginbottom, and E. S. Melnikov (1997), International Permafrost Association circum-arctic map of permafrost and ground ice conditions, U.S. Geol. Surv. Circum Pac. Map Ser., Map CP- 45, scale 1:10,000,000. Brown, J., K. M. Hinkel, and F. E. Nelson (2000), The Circumpolar Active Layer Monitoring (CALM) program: Research designs and initial results, Polar Geogr., 24, Burgess, M., S. L. Smith, J. Brown, V. Romanovsky, and K. Hinkel (2000), Global Terrestrial Network for Permafrost (GTNet-P): Permafrost monitoring contributing to global climate observations, Geol. Surv. Can. Current Res., 2000-E14, 8 pp. Etzelmüller, B., E. S. F. Heggem, N. Sharkhuu, R. Frauenfelder, A. Kaab, and C. E. Goulden (2006), Mountain permafrost distribution modelling and validation using a multi-criteria approach and DC resistivity tomography in continental central Asian mountains: Example from Hovsgol area, northern Mongolia, Permafrost Periglacial Processes, 17, , doi: /ppp.554. Gavrilova, M. K. (2003), Recent climatic change in Asia along Yakutia- Mongolia Transect, Pap. Meteorol. Hydrol., 25, Goulden, C. E., L. Ariuntsetseg, and B. Nandintsetseg (2005a), Is the steppe getting drier?, paper presented at First International Symposium on Terrestrial and Climate Change in Mongolia, Inst. of Meteorol. and Hydrol. of Mongoia, Ulaanbaatar, Mongolia. Goulden, C. E., B. Etzelmüller, L. Ariuntsetseg, B. Nandintsetseg, O. Avirmed, O. Batkhishig, A. Sharkhuu, and N. Sharkhuu (2005b), Permafrost thermal properties and thaw and its relationship to soil and plant cover, Lake Hovsgol, Mongolia, Eos Trans. AGU, 86(52), Fall Meet. Suppl., Abstract C31A Goulden, C. E., O. Tumurtogoo, E. Karabanov, and A. Mongontsetseg (2006), The geological history and geography of Lake Hovsgol, Mongolia, in The Geology, Biodiversity and Ecology of Lake Hovsgol (Mongolia), edited by C. E. Goulden et al., pp.1 20, Backhuys Publisher, Leiden, Germany. Gravis, G. F., S. I. Zabolotnik, A. M. Lisun, and V. L. Sukhodrovskiy (1978), The geocryological characteristics of the Mongolian People s Republic and some characteristics of permafrost development in the past, in Second International Conference on Permafrost: USSR Contribution, edited by F. J. Sanger and P. J. Hyde, pp , Natl. Acad. of Sci., Washington, D. C. Harris, C., and W. Haeberli (2003), Warming permafrost in the mountains of Europe, World Meteorol. Org. Bull., 52, Heggem, E. S. F., B. Etzelmüller, S. Anarmaa, N. Sharkhuu, C. E. Goulden, and B. Nandintsetseg (2006), Spatial Modeling of ground surface temperatures in the Hövsgöl area, northern Mongolia, Permafrost Periglacial Processes, 17, , doi: /ppp.568. Jorgenson, M. T., C. H. Racine, J. C. Walters, and T. E. Osterkamp (2001), Permafrost degradation and ecological changes associated with a warming climate in central Alaska, Clim. Change, 48, , doi:0.1023/ A: Melinikov, P. I., and N. I. Tolstihin (Eds.) (1974), General Permafrost Science (in Russian), Nauka, Novosibirsk, Russia. Nandintsetseg, B., C. E. Goulden, and J. S. Greene (2006), Recent climate change study near Lake Hovsgol, in Changes in Climate, Ecology and Patterns of Pastoral Nomadism in Lake Hovsgol National Park, Mongolia, pp , Admon Publ., Ulaanbaatar, Mongolia. Nelson, F. E. (2003), (Un)frozen in time, Science, 299, , doi: /science Nelson, F. E., N. I. Shiklomanov, K. M. Hinkel, and H. H. Christiansen (2004), The Circumpolar Active Layer Monitoring (CALM) Workshop and the CALM II Program, Polar Geogr., 28, Pavlov, A. V., and G. Z. Perlshtein (2006), Monitoring and Modeling the permafrost dynamics under climate changes, paper presented at Asian Conference on Permafrost, Chin. Acad. of Sci., Cold and Arid Reg. Environ. and Eng. Res. Inst., Lanzhou, China. Romanovsky, V. E., D. O. Sergueev, and T. E. Osterkamp (2003), Temporal variations in the active layer and near-surface permafrost temperatures at the long-term observatories in northern Alaska, paper presented at Eighth International Conference on Permafrost, Int. Permafrost Assoc., Zurich, Switzerland. Sharkhuu, A., N. Sharkhuu, and C. E. Goulden (2006), Permafrost condition and changes at eastern shore of Hovsgol Lake, Mongolian Geosci., 29, Sharkhuu, N. (1998), Trends of permafrost development in the Selenge river basin, Mongolia, paper presented at Seventh International Conference on Permafrost, Natl. Res. Counc. of Can., Yellowknife, Canada. Sharkhuu, N. (2006), Geocryological conditions of the Hovsgol mountainous region, in The Geology, Biodiversity and Ecology of Lake Hovsgol (Mongolia), edited by C. E. Goulden et al., pp , Backhuys, Leiden, Netherlands. Sharkhuu, N., and A. Sharkhuu (2005), Recent degradation of permafrost in Mongolia, paper presented at First International Symposium on Terrestrial and Climate Change in Mongolia, Inst. of Meteorol. and Hyrdrol. of Mongolia, Ulaanbaatar, Mongolia. Tsogtbaatar, J., and C. E. Goulden (2000), Mongolia: Long term ecological research, in The International Long Term Ecological Research Network, edited by J. R. Gosz, C. French, and P. Sprout, pp , Acad. Printers, Albuquerque, N. M. Williams, P. J., and M. W. Smith (1989), The Frozen Earth: Fundamentals of Geocryology, Cambridge Univ. Press, Cambridge, U. K. J. Brown, International Permafrost Association, P.O. Box 7, Woods Hole MA 02543, USA. (jerrybrown@igc.org) B. Etzelmüller and E. S. F. Heggem, Department of Geosciences, Physical Geography, University of Oslo, P.O. Box 1047, Blindern, N-0316 Oslo, Norway. (bernde@geo.uio.no) C. E. Goulden, Institute for Mongolian Biodiversity and Ecological Research, Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Philadelphia, PA 19103, USA. (cgoulden@acnatsci.org) F. E. Nelson and N. I. Shiklomanov, Department of Geography, University of Delaware, Newark, DE 19711, USA. (fnelson@udel.edu; shiklom@udel.edu) A. Sharkhuu and N. Sharkhuu, Geo-Ecology Institute-301, Mongolian Academy of Sciences, Baruun Selbe-13, Ulaanbaatar, Mongolia. (anarmaash@yahoo.com) 11 of 11