Surface characteristics affecting active layer formation in palsas, Finnish Lapland

Size: px

Start display at page:

Download "Surface characteristics affecting active layer formation in palsas, Finnish Lapland"

Richard Austin
6 years ago
Views:

Permafrost, Phillips, Springman & Arenson (eds) 2003 Swets & Zeitlinger, Lisse, ISBN 90 5809 582 7 Surface characteristics affecting active layer formation in palsas, Finnish Lapland M. Rönkkö & M.

1 Permafrost, Phillips, Springman & Arenson (eds) 2003 Swets & Zeitlinger, Lisse, ISBN Surface characteristics affecting active layer formation in palsas, Finnish Lapland M. Rönkkö & M. Seppälä Department of Geography, University of Helsinki, Finland ABSTRACT: Active layer thickness was measured in palsas on two large mires close to Lake Ahkojavri in northernmost Finland in July and August, measurements were made on five types of palsa surface: bare peat, Empetrum nigrum ssp. hermaphroditum, Betula nana, moss and lichen covered surfaces. Palsa height was measured, and also the proportion of abraded surfaces and collapsed edges. Active layer thickness ranged from 29 to 74 cm. Statistical analysis showed that the active layer thickness correlated with the height of palsas, with their degree of erosion and collapse. The active layer was thickest on lichen-covered surfaces (mean 59 cm) and thinnest under Betula nana (mean 51 cm). A surprising result was that the bare peat surfaces (mean active layer thickness 52 cm) did not increase the thawing of the active layer. These observations are in clear disagreement with the vegetation change hypothesis proposed by Railton and Sparling (1973) according to which palsa formation depends on changes in surface albedo. Many of the studied palsas are strongly eroded but four new palsas (60 80 cm in height) were also found on the same mires. The present development stage of the palsas shows that winter wind activity is becoming stronger, causing surface abrasion and forming new palsas. In winter the former active layer on some palsas had not frozen totally. An unfrozen layer was observed between the permafrost layer and the thawing seasonal frost layer. 1 INTRODUCTION The aim of the study is to measure the active layer thawing depth and factors affecting it in the Paistunturit fell area, Utsjoki, Finnish Lapland. The characteristics observed are palsa height, vegetation cover, abrasion degree and the amount of block erosion. The amounts of thermokarst and cracking of peat on palsas and were also observed. Initial hypotheses were that: 1. The active layer on palsas is thinner under vegetated surfaces than under a bare peat surface, 2. the height and size of a palsa also affects the thickness of the active layer (the bigger the palsa, the thicker the active layer), 3. the much abraded and collapsing palsas have a thicker active layer than uneroded palsas. Figure 1. Lapland. Location of the studied palsa mires in Finnish 2 STUDY AREA The study area contains two palsa mires, Luovdijeäggi (12.6 ha in area) and Tsulloveijeäggi (8 ha), close to Lake Ahkojavri, in western Utsjoki, northernmost Finnish Lapland (about N E) (Fig. 1). The mires are located about 350 m above sea level. There are 48 palsas on Luovdijeäggi and 28 palsas on Tsulloveijeäggi. The height of the palsas varies between 0.4 m and 3.3 m. The palsas cover about 10 percent of the areas of the mires. The nearest weather station is at Kevo, some 40 km NEE of Lake Ahkojavri at 107 m a.s.l. There the coldest month is January (MMAT 16 C), the warmest July ( 13 C); the mean annual air temperature is 2.0 C (Climatological Statistics in Finland, 1991). The mean annual precipitation at Kevo is 395 mm. The snow free season is from 20 May to the end of September. The growth season is about 110 days and the thermal sum ( 5 C degree.days) varies between 400 and 900. Low mountains with deep river valleys characterize the topography; the elevation of the region is mostly between 250 and 400 m a.s.l. 995

3 METHODS Field measurements were made from 26 July to 17 August 1999. According to former studies (Seppälä 1976, 1983) the active layer on palsas does not increase much after the end of July.

2 3 METHODS Field measurements were made from 26 July to 17 August According to former studies (Seppälä 1976, 1983) the active layer on palsas does not increase much after the end of July. The depth of the active layer was measured with a metal rod (1 m long, 1 cm in diameter), which was pushed through the active layer to the frost table. The measurements were done with 1 cm precision. Active layer depths were measured on horizontal surfaces with different covers: Empetrum nigrum ssp. hermaphroditum, Betula nana, light lichen cover, moss and bare unvegetated peat (wind abraded surface) at 381 points, of which 250 were on Luovdijeäggi and 131 on Tsulloveijeäggi. The height of the Betula nana shrubs was also measured in order to investigate whether their height affects active layer depth. Each palsa was described according to the following criteria: depth of active layer under various surface covers, palsa height and diameter, amount of block erosion and abrasion, and degree of cracking. Palsa height was determined visually or with a staff. The diameter was measured using a measuring tape or by pacing. The direction of edges with block erosion was determined with a compass with 5 degrees precision. The orientation of the abrasion surfaces was also measured (Seppälä in press). The intensity of abrasion was estimated on a scale of zero to three (zero no abrasion, 1 palsa surface one third abraded, 2 one to two thirds of palsa surface abraded, 3 two thirds of palsa surface abraded). Block erosion was assessed on a similar scale (0 no block erosion, 1 a little block erosion, 2 some block erosion, 3 palsa much collapsed). The vegetation on palsas was also determined. The data were analysed using Excel and SPSS statistical charting, plotting, and data analysis programs. Average, standard deviation, median minimum and maximum were calculated for the data. The data were then stratified, initially into three classes, according to palsa height, degree of block erosion and abrasion. Statistical indices were calculated for the depths of the active layer under different surfaces separately. Height classes included palsas with the height 1.5 m, m and 2 m. Because few palsas scored a zero value, palsas were then divided into 3 groups, which guaranteed enough palsas in each class for statistical analysis to yield significant results. The aim was to find the most important factor determining the thaw depth of active layer on palsas. 4 ANALYSIS AND RESULTS A frequency diagram (Fig. 2) shows the depth of the active layer in the palsas studied for all data combined, Figure 2. Active layer observations classified according to their thickness. for each height class, surface type, block erosion type, and abrasion type. The data collected from two mires were analysed together as a single set. The mean depth of active layer on Luovdijeäggi was 53.9 cm and on Tsulloveijeäggi 53.2 cm. These data support the view of Seppälä (1983) that the depth of active layer varies little between palsa mires. The thinnest active layer was 29 cm and the deepest 74 cm. Standard deviation was 9.0 cm. The average was 53.7 cm and median 54 cm (Fig. 2). The cracking of palsas had no effect on the thaw depth; the active layer was equally deep in cracks and on the surrounding peat. 4.1 Active layer and surface characteristics The thinnest active layer (50.7 cm) was, according to the average and median, beneath dwarf birch (Betula nana) shrubs as expected and under moss cover (53.6 cm) but the median (52 cm) was the same in both cases. The second smallest mean of active layer thickness was beneath unvegetated peat surfaces. The median of the second thinnest active layer was under Empetrum nigrum surface, which had almost the same mean depth (53.1 cm) as the moss surfaces (53.6 cm) (Table 1). The mean active layer thickness was greatest under the lichen surface (59 cm) with the median (60 cm) ranging from 44 cm to 74 cm. A variance analysis (ANOVA) was carried out to investigate the significance of these differences. The deviation of the thickness of active layer is fairly large, about 8 cm, according to the differing surface characteristics. The smallest deviation is on the unvegetated peat surfaces and largest beneath the Betula shrubs (10.2 cm). The data are not normally distributed but almost bimodal. The explanation could be a residual thawed layer. The previous summer the active layer had melted so deeply that the seasonal frost of the following winter did not penetrate to the permafrost table (cf. Salmi 1970). 996

3 Table 1. Mean thickness of active layers (in cm) according to the nature of the studied palsas. Surface quality Barren peat 52.1 Empetrum nigrum 53.1 Lichen 59.0 Betula nana 50.7 Moss 53.6 Height 1.5 m m m 58.0 Abrasion 1 1/ / / Block erosion 0 no some much 57.4 Total mean 53.7 Relationships between the depth of active layer and the height of the palsa were investigated with regression analysis. This confirms that average active layer depth does not differ between surfaces of different characteristics. According to the results of the variance analysis the hypothesis is rejected that palsa height is a significant control on active layer depth. 4.2 Active layer depth and block-erosion The depth of the active layer increases as the amount of block erosion increases on palsas as the insulating peat cover is eroded by collapse. Mean active layer depths are 49.8 cm, 55.1 cm and 57.4 cm in the three block-erosion classes. In palsas that are little collapsed the active layer is thinnest beneath the Betula nana surface according to the median (45 cm), but according to the average (46.6 cm) it is thinnest underneath moss-covered surfaces. On all kind of surfaces the active layer gets thicker as the block-erosion amount increases. 4.3 Active layer depth and abrasion Thickness of active layers The variance analyses indicates that active layer depth increases with greater abrasion, regardless of the surface character. The active layer is thinnest beneath the Betula surface and thickest under the lichen surface. The standard deviation is rather large and the distribution of data is bimodal. Table 2. Spearman s correlation coefficients of abrasion, block erosion and heights of palsas. Block Abrasion erosion Height Abrasion ** 0.291* Block erosion 0.487** ** Height 0.291* 0.558** Active layer depth and palsa height Mean and median active layer depths are smallest on palsas lower than 1.5 m and thickest on palsas over 2 m in height. On unvegetated peat surfaces the active layer is thinnest on the lowest palsas but thickest on the palsas m high. The difference compared to the highest palsas is only a few millimetres. Because the differences were minimal, variance analysis was done to compare the mean values. Variation is great in every palsa group regardless of height or surface characteristics. The height of a palsa does not by itself explain the depth of active layer. According to variance analysis there is a difference between the averages when all the height classes are compared. When the lowest palsas and palsas m high were compared the difference was statistically significant, but when the m and over 2 m high palsas were compared, no significant difference was observed. This same phenomenon was noticed when comparing the different abrasion and block-erosion amounts. The biggest difference between the active layer depths was between the classes 1 and the other two classes; classes 2 and 3 are more similar. Class 1 differs most from the others in height, abrasion and block-erosion amount. Frequencies show that active layer depth increases with height, amount of block-erosion and abrasion, but this is to be expected since they correlate with each other. 4.5 Relationships between palsa characteristics The relationship between observed palsa characteristics were investigated by Spearman rank correlation (Table 2). The relationship between palsa height and block erosion yielded r Thus, the higher the palsa, the more collapsed it is likely to be. The relationship between palsa height and abrasion was similarly tested and yielded r This shows a relationship that is almost statistically significant. The relationship between surface abrasion and block-erosion gives r 0.487, and is statistically significant. A low palsa can be very abraded because it is old and worn. Block erosion can be detected only from big palsas because the collapsed peat blocks are already sunk into the wet mire. 997

4 Figure 4. Regression between the means of active layer thicknesses and of heights of palsas. Linear regression with hatched line. Table 3. Palsa classes identified by cluster analysis. Figure 3. Relationships between the height of palsas and active layer thickness on different kinds of palsa surface. 4.6 Regression analysis Regression analysis was used to investigate whether palsa height controls the active layer depth. The regression model gives R , which means that palsa height explains only 19% of the thickness of the active layer. The correlation coefficient between palsa height active layer thickness is: r 0.43 (n 381, p 0.01). A polynomial regression model yielded R It was observed that vegetation cover causes great variance in active layer thickness (Fig. 3). The thinnest active layer is below the Betula nana shrubs and thickest on the lichen-covered surface. The effect of vegetation was eliminated by calculating the mean active layer depths of the different surface types for each palsa height class. The relationship between palsa height and active layer was then investigated. The linear regression model based on mean values shows that palsa height explains 60% of active layer thickness (n 381). The polynomial trend line gives R (Fig. 4). Palsas cannot be infinitely high nor the active layer infinitely deep (Fig. 3). 4.7 Cluster analysis Because the variables were both quantitative and qualitative, hierarchical cluster analysis was undertaken. The variables were palsa height, abrasion degree, amount of block-erosion and surface cover (unvegetated peat, Empetrum nigrum, lichen, moss and Betula nana). All the variables were measured for 37 palsas and used in the cluster analysis. Two main clusters were identified; they divide into 3 or 4 smaller clusters. Palsas in cluster one are all Abrasion/ Active Height block erosion layer (cm) I A B C , 2, II D E F below or equal to 170 cm high, with little or some deflation and collapsed and their active layer is fairly thin. Palsas in cluster two are over 180 cm high and little or much abraded and collapsed and their active layer is thicker than the palsas in cluster one (Table 3). A palsa which does not fit into these clusters is rather shallow (only 60 cm high) but its active layer is deeper than other palsas of the same height. Its abrasion and collapsing rate is in class 1. It obviously is an old palsa melting from its bottom. Fairly low and much abraded palsas belong to the second cluster; they are classified according to palsa height and active layer thickness. In the main cluster one there are three palsas that appear not to belong to any of the sub clusters. Even though the majority of old palsas at Luovdijeäggi mire melt there were found four new palsas with shallow active layers. They indicate that the climatic conditions are suitable for palsa formation. Their formation as well as the presence of largely abraded palsas mean strong winter storms and probably changing wind conditions (Seppälä, in press). 5 DISCUSSION AND CONCLUSION The thermal balance of the ground is a complex matter. This study concentrates only on the conditions 998

5 prevailing at the end of the thawing season. It is therefore important to remember that the conditions in winter and spring also affect thawing. Snow cover, for example, has a strong influence on the temperature system in the ground and also the hydrological conditions in spring. According to several studies an active layer beneath a vegetation cover does not thaw so deeply as beneath bare ground. Removal or thinning of the insulating peat cover normally increases the thaw depth to cause thermokarst (Brown 1970; Luthin & Guymon 1974; Smith 1975; Allard et al. 1996). It was presupposed that the active layer would be thickest on an unvegetated surface, but was not found in the present study. One reason for the fact that the mean active layer depth on the unvegetated surface is almost the same as on Empetrum nigrum and moss surfaces could be evaporation during the springtime. Melt water evaporates more quickly from a bare peat surface than from a vegetated surface because the vegetation holds moisture well. A dry peat layer protects permafrost below. Other explanations for the relatively thin active layer could be that the unvegetated peat surface is free of snow during winter. Cold can penetrate deeply into the peat. The dense or high vegetation collects snow but also protects the permafrost from spring warming and causes a thinner active layer than sparse or low vegetation (Smith 1975). The active layer was thinner under Betula nana shrubs even though the difference compared to other surfaces studied is not significant according to variance analysis. There is more snow on the high vegetation and therefore the heat loss is less compared to other vegetation or bare ground (Kershaw & Gill 1979). This could explain why the active layer beneath the Betula nana bush is not notably thinner than on the lower vegetation or unvegetated peat surface. Betula nana shrubs transpire more during the summer so peat underneath it dries and protects the permafrost. The ground also remains cooler due to evaporation. The growing conditions for Betula nana are more favourable when the active layer is thicker. The active layer was thickest under the lichen-covered surfaces regardless of the size of palsa or abrasion or block-erosion. The peat under the lichen cover is fairly porous and provides a good insulator but on the other hand porous peat infiltrates meltwater and rain easily and that enhances thawing. Also the low evaporation might explain the high depth of active layer under lichen cover. According to Rouse et al. (1977) evaporation is least on the lichen heaths, so the ground cools less than other vegetated surfaces and peat. It also remains moist and its insulating effect is therefore weak. According to variance analysis lichen affects the thaw depths a little more than other surface types. This study does not support the views of Railton and Sparling (1973) that palsas evolve because of the high albedo under a light coloured lichen-cover, which should decrease the thickness of active layer. They assumed that the vegetation succession from Sphagnum fuscum cover to Cladonia sp. dominant vegetation (surface albedo) causes the formation of palsas. This study also does not support the theory that melting of palsas starts when the abraded dark surface of peat absorbs more radiation during the summer. Cummings and Pollard (1990) claimed that the low albedo of unvegetated peat surface causes the thicker active layer. Low palsas have shallow active layers regardless of the development stage. Generalizing it could be said that the active layer becomes thicker as abrasion, block erosion and the height of the palsa increase. It seems that palsa height is the controlling factor on active layer thickness in the Paistunturit area. ACKNOWLEDGEMENTS Professor Derek Mottershead kindly revised the English of the manuscript. Kirsti Lehto made the final drawing of the figures. Hilkka Ailio finished the layout of the paper. Field investigations were financially supported by Seth Sohlberg s Delegation. REFERENCES Allard, M., S. Caron & Y. Bégin (1996). Climatic and ecological controls on ice segregation and thermokarst: the case history of a permafrost plateau in northern Quebec. Permafrost and Periglacial Processes 7, Brown, R.J E. (1970). Permafrost as an ecological factor in the subarctic. In: Ecology of the Subarctic Regions. Proceedings of the Helsinki symposium, Unesco, Paris. Cummings, C.E. & W.H. Pollard (1990). Cryogenetic categorization of peat and mineral cored palsas in the Schefferville area, Quebec. Collection nordicana N 54, Proceedings of the 5th Canadian Permafrost Conference, Centre d`etudes nòrdiques, Quebec. Kershaw, G.P. & D. Gill (1979). Growth and decay of palsas and peat plateaus in the Macmillan Pass Tsichu River area, Northwest Territories, Canada. Canadian Journal of Earth Sciences 16: 7, Luthin, J.N. & G.L. Guymon (1974). Soil moisturevegetation-temperature relationships in central Alaska. Journal of Hydrology 23: 3 4, Railton, J.B. & J.H. Sparling (1973). Preliminary studies on the ecology of palsa mounds in northern Ontario. Canadian Journal of Botany 51: 5, Rouse, W.R., Mills P.F. & R.B. Stewart (1977). Evaporation in high latitudes. Water Resources Research 13: 6, Salmi, M. (1970). Investigations on palsas in Finnish Lapland. In: Ecology of the subarctic regions. 999

6 Proceedings of the Helsinki Symposium 1966, UNESCO, Paris. Seppälä, M. (1976). Seasonal thawing of a palsa at Enontekiö, Finnish Lapland, in Biuletyn Peryglacjalny 26, Seppälä, M. (1983). Seasonal thawing of palsas in Finnish Lapland. In: Proceedings of the fourth international conference on permafrost, July 17 22, Fairbanks, Alaska, National Academy Press, Washington D.C. Seppälä, M. Surface abrasion of palsas by wind action in Finnish Lapland. Geomorphology (in press). Smith, M.W. (1975). Microclimatic influences on ground temperatures and permafrost distribution, Mackenzie Delta, Northwest Territories. Canadian Journal of Earth Sciences 12: 8,

Permafrost & climate change in northern Finland Dr Steve Gurney

Permafrost & climate change in northern Finland Dr Steve Gurney Senior Lecturer in Geomorphology University of Reading, UK Docent in cold climate geomorphology University of Turku, Finland Topics Introduction