Snowlines And Glaciation Thresholds

In regions of permanent snow accumulation, it is possible to identify an altitudinal zone separating the lower region of seasonal snow accumulation from the upper region of permanent snow. The term zone is used because from year to year the actual boundary or snowline will vary in elevation, depending on the particular

FIGURE 7.9 Paleotemperature reconstruction for Europe based on periglacial features (right column) compared to other long proxy data records (Maarleveld, 1976).

weather conditions during the accumulation and ablation seasons. On a glacier, this would be equivalent to the firn line or (on temperate glaciers where there is no superimposed ice zone) the equilibrium line altitude (ELA). If observations were made over a period of time, the average elevation of the snowline would be apparent, enabling the regional or climatic snowline to be identified (0strem, 1974). Observations of modern glaciers indicate that the ELA approximates the height at which the accumulation area of the glacier occupies -70% of its total area. To estimate paleo-snowlines, the common practice is to reconstruct the paleo-ELA by mapping the former glacier area (from moraine position) and then determining where the ELA would have been, based on an accumulation area ratio of -0.7. This generally corresponds to the uppermost limit of lateral moraines. Mapping the difference between modern and paleo-ELAs constructed in this way can provide useful insights into past climatic conditions. For example, Pewe and Reger (1972) were able to demonstrate that the Arctic Ocean plays no significant role as a moisture source in the present-day glaciation of Alaska, because the snowline gradient along the north coast is not steep and the snowline does not fall to low elevations (Fig. 7.10). By contrast, snowline gradients in the south, adjacent to the Bering Sea, are very steep and increase in elevation rapidly away from the moisture source. A similar pattern of snowlines (though considerably lower) occurred in late Wisconsin times, indicating that the main moisture sources then were like those of today.

FIGURE 7.10 Modern and Wisconsin snowlines in Alaska (in meters). Snowlines today show the same pattern as glacial age snowlines, suggesting that no major change in moisture sources has occurred.The Gulf of Alaska, not the Beaufort Sea, was thus the most important source of moisture for glaciation (P6we and Reger, 1972).

FIGURE 7.10 Modern and Wisconsin snowlines in Alaska (in meters). Snowlines today show the same pattern as glacial age snowlines, suggesting that no major change in moisture sources has occurred.The Gulf of Alaska, not the Beaufort Sea, was thus the most important source of moisture for glaciation (P6we and Reger, 1972).

A similar index is provided by glaciation levels or glaciation thresholds that define the lowest limit at which glaciers or permanent ice fields can develop (Miller et al., 1975; Porter, 1977). This is usually determined by identifying the highest unglacierized, and the lowest glacierized, mountain summits in a region and averaging the two elevations. Both snowlines and glaciation thresholds can be used in paleoclimatic reconstructions if similar features can be mapped for periods in the past (Osmaston, 1975), though obviously they only provide information about (extreme) glacial periods and can add nothing to our knowledge of warmer intervals. Although much effort has been expended in mapping both modern and former snowlines, with only a few exceptions, paleoclimatic reconstructions have been simplistic and the results often equivocal. This is due mainly to the following problems:

(a) Present-day snowlines have not been adequately studied in relation to present climate; climatic "controls" on modern snowline elevations are not well understood and cannot be assumed to be the same in all areas. Furthermore, atmospheric lapse rates have not been well-documented in mountain regions and are problematic in paleotemperature reconstructions.

(b) Paleosnowline reconstructions are often based on features of varying age, possibly accounting for the large variations in estimates of past snowline lowering (Reeves, 1965; Brakenridge, 1978).

7.4.1 The Climatic and Paleodimatic Interpretation of Snowlines

It has commonly been assumed that snowlines are related to the height of the summer 0 °C isotherm, and, indeed, Leopold (1951) demonstrated a close correspondence between the two surfaces, in the western United States from 35° to 50° N. Paleosnowlines approximately 1000 m lower than today were thus interpreted as indicating lower summer temperatures. Using modern free air lapse rates of 0.6 °C per 100 m in summer months, Leopold concluded that July temperatures were 6 °C lower than today when snowlines were at the position of maximum depression. Other months were assumed to have cooled proportionately less, with midwinter (January) temperatures having remained the same as those today. As a result, he concluded that mean annual temperatures had been 4-5 °C lower. Using similar logic, but an assumed lapse rate of 0.75 °C per 100 m, Reeves (1965) concluded that the 1300 m lowering of the snowline in New Mexico was equivalent to a decrease in July temperature of 10 °C and in mean annual temperature of 5.1 °C. Considerable doubt was cast on these estimates by Brakenridge (1978), who found no strong relationship between present-day snowline in the American southwest and the July 0 °C isotherm, or indeed any July isotherm, as the latitudinal snowline gradient is considerably steeper. A better fit is obtained with the -6 °C mean annual isotherm; the "full glacial" snowline has a similar gradient, suggesting that there was a fall in mean annual temperature of 7 °C. In the central Andes (10°-30° S) snowlines are actually lower in areas with higher temperatures, because such areas have higher precipitation, which compensates for the warmer conditions; the highest snowlines are found in the cold, arid mountains of the western Altiplano (Sierra Occidental) (Fox, 1991).

These studies illustrate the basic problems of identifying how snowlines vary with temperature and, if they do, what lapse rate should be assumed in order to use former snowlines as an indicator of temperature change. It is probably not a reasonable assumption to use modern lapse rates in such calculations, as both temperature and moisture conditions in the past would have been different from today, perhaps resulting in lower lapse rates in many areas. However, this entire approach is extremely simplistic and neglects other important factors. Snowline is not only a function of lapse rate but also depends on the variation of accumulation with elevation (accumulation gradient), radiation balance, wind speed, humidity, and the variation of albedo with temperature (as temperature influences the frequency of snowfall vs rainfall events) (L. Williams, 1975; Seltzer, 1994).

The importance of precipitation in controlling snowline elevation and spatial variation has been noted for the Cordilleran mountains of both North and South America. In the Cascade Range of Washington, for example, 86% of the variance in glaciation thresholds is explained by accumulation season precipitation (though mean annual temperature is highly correlated with precipitation) (Porter, 1977). By assessing the amount of temperature change during the late Wisconsin (Fraser) glaciation from palynological evidence, Porter estimated that winter precipitation was 20-30% less than today, accompanied by ablation season temperatures 5.5 ± 1.5 °C lower than at present. Extending this approach to the Rocky Mountains, Porter et al. (1983) estimated that the lower late Pleistocene snowlines of that region required substantially lower temperatures than a simple lapse rate calculation would suggest (T = -10 to -15 °C vs -6 °C) because conditions were drier at the time. Unless both temperature and precipitation changes are taken into account, erroneous conclusions would be reached. However, such an approach is seen by Seltzer (1994) as only a partial solution because there were undoubtedly also changes in incoming radiation (due to orbital forcing) as well as regional effects due to changes in cloudiness and, hence, in the overall radiation balance. He proposes a model incorporating many of these factors, but estimating how the important variables may have changed remains a major challenge.

In the Andes, snowlines are highest (>6000 m) at the latitude of maximum aridity (Fig. 7.11). "Pleistocene" snowlines were lower by 650-1500 m, the depression being greatest in the hyperarid regions of southern Peru and northern Chile, and

FIGURE 7.1 I Modern (-) and Pleistocene (----) snowline elevations along west-east transects in the South American Andes. Principal zonal wind components today are shown by arrows at diagram on left. At the northern and southern locations Pleistocene snowlines were uniformly lower throughout the transect. In the central location (28° S) a reversal of snowline gradient is apparent between Pleistocene and modern times. This resulted from a shift in the subtropical high-pressure cell and associated wind fields, changing from predominantly onshore flow in Pleistocene time to offshore today (Hastenrath, 1971).

t6000

FIGURE 7.1 I Modern (-) and Pleistocene (----) snowline elevations along west-east transects in the South American Andes. Principal zonal wind components today are shown by arrows at diagram on left. At the northern and southern locations Pleistocene snowlines were uniformly lower throughout the transect. In the central location (28° S) a reversal of snowline gradient is apparent between Pleistocene and modern times. This resulted from a shift in the subtropical high-pressure cell and associated wind fields, changing from predominantly onshore flow in Pleistocene time to offshore today (Hastenrath, 1971).

t6000

least in the equatorial zone. This suggests a general temperature decrease over the entire region, but substantially higher precipitation amounts in the desert zone (Hastenrath, 1967). Further insight is provided by east-west transects across the mountain barrier, which indicate a reversal of modern snowline gradients between 28° and 32° S (Fig. 7.11). This is related to the prevailing circulation around subtropical high-pressure cells centered at ~30° S; prevailing winds are easterly to the north and westerly (onshore) to the south. During glacial periods the lower snowline gradients generally parallel modern snowlines, but at 28° S this is not the case. "Pleistocene" snowlines show a reversal of gradient, suggesting an equatorward shift of -5° in the boundary between temperate latitude westerlies and tropical easterlies in the lower troposphere. Where westerly wind regimes became established, a marked eastward rise of Pleistocene snowline resulted (Fig. 7.11). Similar reversals of firn line gradient, from modern times to glacial periods, have also been noted in parts of East Africa (Hamilton and Perrott, 1979).

From this brief survey it is apparent that snowlines in different regions are controlled by different climatic parameters and that these must first be understood in order to use paleosnowlines in paleoclimatic reconstructions. However, some assessment of the relative importance of different climatic variables to snowline lowering can be made by the use of an energy balance model that takes into account many of the relevant variables and the interactions between them. For example, such a model has been applied to the question of what climatic conditions were necessary to bring about extensive glacierization of northern Canada (Williams, 1979). By calculating the regional snowline for varying climatic conditions it was possible to determine where perennial snow cover is most likely to have developed in the past (i.e., which areas are most susceptible to glacierization) and the extent of glacierization brought about by different changes in climatic conditions. Interestingly, Williams's model indicates that a substantial fall in summer temperatures (1012 °C) is necessary to extensively glacierize Keewatin and Labrador, though smaller changes in temperature are sufficient to glacierize Baffin Island, to the north (Fig. 7.12). This confirms the view that Baffin Island is particularly sensitive to climatic fluctuations, and is likely to have been a primary site for ice-sheet initiation in the past (and probably in the future also) (Tarr, 1897; Bradley and Miller, 1972; Andrews et al., 1972). It is also of interest that the model results indicate that increased snowfall has little impact on the areal extent of glacierization; temperature changes seem to be of most significance for regional snowline lowering and glacierization in this region (Williams, 1979).

7.4.2 The Age of Former Snowlines

In the preceding section, reference has frequently been made to "former" or "Pleistocene" snowlines without qualification. However, not all paleosnowlines are defined in the same way, further confusing the paleoclimatic picture. A common method is to estimate the average elevation of cirque floors occupied by glaciers during a particular glacial period (Pewe and Reger, 1972). Alternatively, paleosnowlines may be located at the median altitude between the terminal moraine of a given

amounts of uniform spring and summer temperature decrease, assuming "normal" (1931-1970) snow accumulation and Earth-orbital parameters of 116,000 yr B.P. for 31 March. For example, with a spring-summer lowering of 6-8 °C the perennial snow cover would have been confined mainly to the Arctic Islands and the northern tip of Labrador-Ungava; with a 10-12 °C temperature lowering extensive areas of Keewatin and Labrador would have been glacierized (L. D.Williams, 1979).

amounts of uniform spring and summer temperature decrease, assuming "normal" (1931-1970) snow accumulation and Earth-orbital parameters of 116,000 yr B.P. for 31 March. For example, with a spring-summer lowering of 6-8 °C the perennial snow cover would have been confined mainly to the Arctic Islands and the northern tip of Labrador-Ungava; with a 10-12 °C temperature lowering extensive areas of Keewatin and Labrador would have been glacierized (L. D.Williams, 1979).

advance and the highest point on the cirque headwall (Richmond, 1965). In both approaches, orientation of the cirque is an important factor, not only in terms of radiation receipts, but also in terms of prevailing winds and enhanced precipitation catchment in the lee of mountain barriers. Markedly different paleosnowlines may result from studies based on different cirque populations of varying orientation. Furthermore, the use of cirque floor elevation without knowledge of the age of glacial deposits associated with the feature may lead to a mixed population of paleosnowline estimates. Thus, Hastenrath (1971) is only able to describe the paleosnowlines he mapped as "Pleistocene" and Pewe and Reger (1972) describe their paleosnowlines as "generalized Wisconsin age" features. Such factors may explain the large variations in snowline depression commonly reported; in New Mexico, for example, estimates of regional snowline lowering vary from 1000 to 1500 m (Brak-enridge, 1978) and even larger variations are noted in other studies (Reeves, 1965). This imprecision in dating, and uncertainty over climatic controls on snowline elevation, severely limits the value of most snowline studies for paleoclimatic reconstruction. However, careful study of modern conditions and of well-dated glacial deposits can provide important insights into paleoclimatic conditions of mountain regions, as shown by Porter (1977). Indeed, there would appear to be considerable potential in a re-evaluation of this subject for many mountainous parts of the world.

An interesting example is provided by Dahl and Nesje (1996) in their study of Holocene paleoclimate in the mountains of southern Norway. They estimated changes in the size of Hardangerjokulen from glaciofluvial and glaciolacustrine sediments, relying on the fact that as the outlet glacier advanced and retreated certain thresholds were passed, which were recorded as diagnostic signatures in downstream sediments. This enabled the ice cap area to be estimated, and from that the paleo-ELA could be calculated (Fig. 7.13). Changes in the upper limit of pine (Pinus sylvestris L.) provided an assessment of variations in summer temperature, and from these estimates of temperatures at the ELA could be obtained (using a lapse rate of 0.6 °C per 100 m). They then used a well-established relationship between winter accumulation (A) and temperature (t) at the equilibrium line of Norwegian glaciers (A = 0.915 e0339t) to reconstruct the change in winter accumulation at the ELA (assuming this relationship has remained constant over time) (Fig. 7.13). Their approach reveals that at certain times in the past (9500-8300, 7200-6200, and at ~5500 and -4500 [calibrated 14C] yr B.P.) the ice cap grew in size due to higher winter precipitation, in spite of warmer summer temperatures. The major "Little Ice Age" advance (culminating in the eighteenth century) was unusual in resulting from both higher winter precipitation and lower summer temperatures, which perhaps explains why the ELA at that time was lower than at any other time in the Holocene.

7.5 MOUNTAIN GLACIER FLUCTUATIONS25

Glacier fluctuations result from changes in the mass balance of glaciers; increases in net accumulation lead to glacier thickening, mass transfer, and advance of the glacier snout; increases in net ablation lead to glacier thinning and recession at the glacier front. A glacial advance therefore corresponds to a positive mass balance brought about by a climatic fluctuation, which favors accumulation over ablation (Fig. 7.14). However, there are many combinations of climatic conditions that might correspond to such a net change in mass balance (Oerlemans and Hoogen-dorn, 1989) so that evidence of a formerly more extensive ice position does not provide an unequivocal picture of climate at the time. However, if a check on one important variable (such as temperature) is available from an independent source of paleoclimatic data, it may be possible to calculate, or model, the overall change in climate which resulted in the advance or retreat of the glacier terminus (Allison and Kruss, 1977).

Changes in mass balance are not transformed immediately into changes in glacier front positions. There may be a period of down-wasting, during which the glacier

25 The growth and decay of continental ice sheets, and associated changes in sea level and oceanic and atmospheric circulation are fundamental characteristics of the Quaternary Period (for a global overview, see Quaternary Science Reviews, 5, 1986 and 9, 2/3, 1990). Such studies provide important regional information on ice extent in response (implicitly) to large scale changes in climate.

FIGURE 7.13 Variations in equilibrium line altitude (ELA) on Hardangerjokulen, south central Norway (top panel), summer temperature based on subfossil wood (Pinus sylvestris L.) found above modern tree line (middle panel) and winter precipitation (accumulation) derived from modern relationships between accumulation and temperature at the ELA (lower panel: note scale is inverted). Values are expressed relative to modern conditions. Ages are given in calibrated and uncalibrated l4C years at top. Summer temperatures were above present levels for most of the Holocene; glacier advances were associated with periods of higher winter accumulation. Only the "Little Ice Age" experienced both lower summer temperatures and higher winter precipitation, leading to the largest ice advance of the entire Holocene (Dahl and Nesje, 1996).

FIGURE 7.13 Variations in equilibrium line altitude (ELA) on Hardangerjokulen, south central Norway (top panel), summer temperature based on subfossil wood (Pinus sylvestris L.) found above modern tree line (middle panel) and winter precipitation (accumulation) derived from modern relationships between accumulation and temperature at the ELA (lower panel: note scale is inverted). Values are expressed relative to modern conditions. Ages are given in calibrated and uncalibrated l4C years at top. Summer temperatures were above present levels for most of the Holocene; glacier advances were associated with periods of higher winter accumulation. Only the "Little Ice Age" experienced both lower summer temperatures and higher winter precipitation, leading to the largest ice advance of the entire Holocene (Dahl and Nesje, 1996).

1880 <890 1900 1910 1920 1930 1940 1950 i960

FIGURE 7.14 Mean departures (from 1851-1950 averages) of summer (June, July, and August) precipitation (%) and temperature (°C) at seven stations above 2000 m in the Swiss Alps. Shaded areas indicate times when precipitation was above average and temperatures were below average.These correspond closely to the times of principal glacier advances in the area (bars at bottom) (Hoinkes, 1968).

1880 <890 1900 1910 1920 1930 1940 1950 i960

FIGURE 7.14 Mean departures (from 1851-1950 averages) of summer (June, July, and August) precipitation (%) and temperature (°C) at seven stations above 2000 m in the Swiss Alps. Shaded areas indicate times when precipitation was above average and temperatures were below average.These correspond closely to the times of principal glacier advances in the area (bars at bottom) (Hoinkes, 1968).

loses mass but does not recede. For example, the lower Khumbu Glacier, in the Mount Everest massif, thinned by -70 m between 1930 and 1956, but the glacier snout remained stationary (Miiller, 1958). Even when down-wasting is not a significant factor, glacier front positions will lag behind climatic fluctuations. Different glaciers have different response times to mass balance variations (Oerlemans, 1989). An increase in net mass results in a kinematic wave moving down the glacier at a rate several times faster than the normal rate of ice flow. The time it takes for this wave to reach the glacier snout is the response time of the glacier and depends on a number of factors including the glacier length, basal slope, ice thickness and temperature, and overall geometry of the glacier itself (Nye, 1965; Paterson, 1994). The South Cascade Glacier (Washington) for example, has a response time of only about 25-30 years, whereas large ice sheets have response times on the order of millennia.

There is some evidence that an excess of ablation over accumulation may cause a more rapid response, with ice recession lagging very little behind the climatic fluctuation that caused the mass loss (Karlen, 1980). Glacier front variations are thus a rather complex integration of both short- and long-term climatic fluctuations, so that one should not be surprised to see some larger glaciers advancing at the same time that smaller glaciers, with shorter response times, are retreating. Indeed, many Arctic glaciers are still advancing today in response to cooler conditions of the last century, at the end of the Little Ice Age, whereas smaller mid-latitude alpine glaciers over the same interval have advanced, receded (due to the early twentieth century world-wide increase in temperatures), and subsequently readvanced in response to cooler conditions over the last two or three decades. Glaciers of different sizes and response times in different areas may therefore be undergoing synchronous advances, but in response to different climatic events; the largest glacier systems respond to low-frequency climatic fluctuations whereas the smaller systems respond to higher-frequency fluctuations.

7.5.1 Evidence of Glacier Fluctuations

The complexity of climatic conditions and differing glacier response times makes glacier fluctuations, as viewed through the misty window of the paleoclimatic records, a rather complicated source of paleoclimatic data. This is compounded by the difficulties inherent in dating former glacier front positions in environments that generally have very little organic material for dating, and where weathering rates are extremely slow.

A record of changes in glacier front positions is generally derived from moraines produced during glacial advances; periods of glacier recession, and the magnitude of the recession, are much harder to identify in the field, though the record of down-valley glaciofluvial or glaciolacustrine sediments may assist in the interpretation of former glacier front positions (Dahl and Nesje, 1996). The problem with the record of moraines is that they are commonly incomplete, with recent advances (which were often the most extensive) obliterating evidence of earlier, less extensive advances. However, detailed stratigraphic studies of glacial deposits may reveal buried soils and weathered profiles indicative of former subaerial surfaces, subsequently buried by more recent morainic debris (Rothlisberger, 1976; and Schneebeli, 1976).

By far the greatest difficulty in the use of glacier front positions as a paleoclimatic index is the problem of dating the glacial deposits. Radiocarbon dates on organic material in soils that have developed on moraines may be obtained, but they provide only a minimum age on the glacial advance. There may be a delay of many hundreds of years between the time a moraine becomes relatively stable and the time it takes for a mature soil profile to develop. If a soil has been buried by debris from a subsequent glacial advance, a date on the soil would provide only a maximum age on the later glacial episode because the organic material in the soil may be hundreds of years older (at least) than the subsequent ice advance (Griffey and Matthews, 1978; Matthews, 1980). Past variations in radiocarbon production rates further impede accurate dating, especially in the last 500 yr or so, encompassing the critically important "Little Ice Age" glacier advances. For older glaciations in volcanic areas, tephrochronology (see Section 4.2.3) or the dating of interstrati-fied lava flows may assist in the interpretation of glacial events (Loffler, 1976; Porter, 1979). Elsewhere, recent advances in cosmogenic isotope dating of exposed surfaces have opened up the possibility of dating the time that has elapsed since an area was ice-covered, or since which a moraine became stabilized (Phillips et al., 1996).

Lichenometry is commonly used to date moraines (see Section 4.3.1) but the technique is uncertain and probably no more reliable than ±20% before 1000 yr B.P.; in most cases, lack of a well-dated calibration curve, or the derivation of a curve based on observations in an environment totally unlike that of the glacial valley or cirque in question, makes the uncertainties even larger. In short, the chronology of glacier fluctuations is quite uncertain in most parts of the world.

7.5.2 The Record of Glacier Front Positions

Studies of glacier fluctuations have been conducted in virtually all mountainous parts of the world (Field, 1975). In spite of the difficulties of dating glacial deposits, mountain regions have provided some of the most comprehensive records of repeated glaciation throughout the Quaternary Period (Richmond, 1985). Glacial deposits from mountain glaciers indicate that there have been at least 11 episodes of major glaciation in the last 1 M years that were (broadly speaking) of global significance (Fig. 7.15) as evidenced by a corresponding enrichment of the oceans in lsO. However, because of the forementioned dating problems, the most detailed work has focused on Postglacial (Holocene) glacier fluctuations. Early work in the Rocky Mountains led Matthes (1940, 1942) to suggest that many alpine glaciers disappeared during a mid-Holocene warm and dry period (termed the Altithermal by Antevs, 1948) only to be regenerated during subsequent cooler and/or wetter periods ("Neoglaciations"; Porter and Denton, 1967). Evidence that mountain glaciers and small ice caps may have disappeared entirely in the early to mid-Holocene is often circumstantial, but nevertheless quite compelling.

The most comprehensive studies have been carried out in Scandinavia, based on lake sediment studies and 14C dated subfossil wood from above modern treeline that provide estimates of summer temperature for much of the Holocene (see Section 8.2.2). It seems likely that many alpine glaciers in the region could not have survived the warmest episodes from 8000 to 4000 yr B.P. (Karlen, 1981; Nesje et al., 1991). Similar conclusions have been reached by other workers elsewhere (e.g., Brown [1990] for the equatorial glaciers of New Guinea). A post-mid-Holocene climatic deterioration is apparent in many areas, resulting in renewed glacierization and glacier advances between 5000 and 4000 yr B.P. (Fig. 7.16). Several episodes of glaciation occurred over the following few thousand years, culminating in the most recent Neoglacial episode (which occurred between the fourteenth and early nineteenth centuries) and which generally resulted in the most extensive Holocene glacial advances. This period is commonly referred to as the Little Ice Age (Bradley and Jones, 1993) and is particularly well-documented in western Europe (Grove, 1988). In the European Alps, glacier advances can be traced through historical records, paintings, and sketches, and this has greatly facilitated the interpretation of glacial deposits in the field. Detailed reconstructions of glacier front positions over the past 300-400 yr have been constructed for several Swiss and Austrian glaciers, most notably the Grindelwald Glacier, Switzerland (Zumbiihl, 1980). Using a variety of such sources, periods of both glacier advance and recession can be determined (Fig. 7.17; Messerli et al., 1978). However, such detailed records are rare and the advance/retreat record would have been impossible to construct with only geo-morphological evidence to go on.

Because the Little Ice Age advances were most extensive (in many areas greater than at any time since the end of the last major glaciation) the record of u

"TT H

Late

Pleistocene Middle

Early

Pliocene

0 0

Post a comment