Embed Size (px)
Citation preview
Svalbard temperature and precipitation, late 19th century to the present. Final report on ACIA-funded project Jack Kohler1†, Øyvind Nordli2, Ola Brandt1, Elisabeth Isaksson1 Veijo Pohjola3, Tonu Martma4, Harald Faste Aas1 1Norwegian Polar Institute, Polar Environmental Centre, N-9296 Tromsø. 2Norwegian Meteorological Institute, Box 43 Blindern, N-0313 Oslo 3Dept. of Earth Sciences, U. Uppsala S-752 36 Uppsala Sweden 4Institute of Geology. Tallin Technical U. 10143 Tallinn Estonia. †e-mail: [email protected]
Table of contents Table of contents .............................................................................................................. 1 Summary........................................................................................................................... 1 Main conclusions:............................................................................................................. 2 1. Instrumental and ice-core temperature records ........................................................ 2
Introduction .................................................................................................................. 2 Data............................................................................................................................... 4
Expedition data......................................................................................................... 4 Ice core data.............................................................................................................. 5 Results ...................................................................................................................... 5
2. Instrumental and ice-core precipitation records ....................................................... 5 3. Mass balance record for Midre Lovénbreen and Austre Brøggerbreen, Svalbard ... 7
Introduction .................................................................................................................. 7 Methods ........................................................................................................................ 8 Balances as a function of elevation .............................................................................. 9 Results .......................................................................................................................... 9
4. Glacier geometry changes on Midre Lovénbreen, Svalbard .................................. 13 Introduction ................................................................................................................ 13 Calculation of glacier surface elevation and volume change ..................................... 14 Front position change ................................................................................................. 16 Relating front position change to climate................................................................... 16
5. References .............................................................................................................. 20
Summary The homogenized Svalbard Airport series is one of only a few long-term (>65 yr)
instrumental temperature series from the high Arctic. It is, therefore, an important record for interpreting present Arctic meteorological trends in terms of past behaviour. Some doubt remains, however, about the reliability of the early part of the Svalbard Airport record (1912-1920). In particular, the instrumental record seems to indicate a step change in Svalbard climate occurring around 1920.
We present the first results of extending the Svalbard Airport record through incorporating newly digitised meteorological observations from early 20th century over-wintering expeditions (eight winters in the interval 1899-1913). In addition, we compare
the extended instrumental record with the ice core oxygen isotope record, a temperature proxy, from the high-resolution 1997 Lomonosovfonna ice core.
Finally, we use newly digitised information from the Midre Lovénbreen glacier mass balance time series together with geometrical change data taken from older maps and front position records to estimate long-term changes in that glacier’s equilibrium line altitude (ELA), which are then interpreted in terms of temperature and precipitation changes.
The winter expedition data prior to 1911 indicate higher temperatures than the minimum values in the 1910s. For the limited periods in which there are both expedition data and Svalbard Airport data, the latter monthly mean temperatures are consistently colder than the former during the coldest months. Furthermore, the ice core record shows only slightly more negative isotope values in the period 1910-1920, suggesting that the winters in the early instrumental record may have been less cold than implied by the homogenized Svalbard Airport series.
The ice core record also indicates that there was a gradual warming trend starting at the end of the 1800s, rather than a step change in climate. The latter view is supported by analysis presented here of the advance of Midre Lovénbreen in the late 1800s to its maximum, attained sometime around 1910. The advance is interpreted to correspond to an equilibrium line altitude (ELA) lower by about 100 m relative to the modern; if this were due solely to a temperature change, then summer temperatures were colder by about 1 ºC for the period 1860-1900, compared to the present. If there were a concurrent precipitation increase then there would be a smaller temperature difference between the late 1800s and the present, and conversely.
The Svalbard Airport precipitation record is compared to the ice-core accumulation record determined using annual layer counting and datable volcanic and beta-activity peaks. Both the instrumental and the ice-core records indicate an increasing accumulation trend over the last 100 years or so. Much of the positive trend in the instrumental series has been attributed to the undercatch phenomenon (Førland and Hanssen-Bauer, 2000), in which the total amount of precipitation detected in a gauge decreases when the precipitation occurs as snow; in a warming climate, an increasing amount of the annual precipitation would occur as rain and there would be an apparent increase in the measured precipitation. However, the evidence from the instrumental record and from the ice core record suggests that while undercatch may explain part of the observed accumulation rise, there has been a real increase in accumulation during the latter half of the 1900s.
Main conclusions: • Gradual warming on Svalbard starts in latter half of 1800s, no step change in
temperature at around 1920. • Temperature in latter half of 1800s were ca. 1 °C colder than the present. • There has been a long-term trend toward increasing precipitation.
1. Instrumental and ice-core temperature records Jack Kohler, Øyvind Nordli, Elisabeth Isaksson, Veijo Pohjola. Introduction
The homogenized Svalbard Airport temperature record (Figure 1) is one of only a few long-term (>65 yr) instrumental temperature series from the high Arctic (Nordli and
others, 1996). As such, it is an important record for interpreting present Arctic meteorological trends in terms of past behaviour. The record comprises records from several different sources (Figure 3a)
-13
-12
-11
-10
-9
-8
-7
-6
-5
-4
-3
1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Tem
per
atu
r (
C)
Obser. Filt.1 Filt.2
Figure 1 Annual temperature at Svalbard Airport, in the Longyearbyen area. Annual values and smoothed values with two Gaussian filters (Filt.1 and Filt.2) that remove variations at time scales less than about 10 and 30 years respectively.
Questions have been raised about the reliability of the early part of the Svalbard Airport record, 1912-1920. During these years mean annual temperatures appear to increase dramatically (Figure 1), with an increase of over 3ºC indicated between the periods 1912-1917 and 1918-1923. From a technical point of view, no gross irregularity has been detected (Birkeland 1930; Steffensen and others, 1996). Nevertheless, modeling of mean sea level pressure fields and temperature data (Hanssen-Bauer and Førland, 1998) indicated that there might have been a much less dramatic temperature rise for the early period.
Met data from Svalbard
1898.00
1899.00
1900.00
1901.00
1902.00
1903.00
1904.00
1905.00
1906.00
1907.00
1908.00
1909.00
1910.00
1911.00
1912.00
1913.00
1914.00
1915.00
1916.00
1917.00
1918.00
1919.00
1920.00
1921.00
1922.00
1923.00
1924.00
1925.00
1926.00
1927.00
1928.00
1929.00
1930.00
Svalbard Airport (monthly) Ebeltofhamna (Robitzsch) AkseløyaKapp Lee Kapp Mitra (Robitzsch) SørkappøyaGreen Harbour (not digitized) Kvadehuken (not digitized)
Figure 2 Time extents of the available met data, including scientific expeditions and hunting expeditions equipped with meteorological instruments from DNMI during the period before regular observations where started at Svalbard.
Here we present the first results of extending the Svalbard Airport record through incorporating newly digitised meteorological observations from early 20th century over-wintering expeditions. Further, we compare (Figure 3) the existing and extended instrumental record with the oxygen isotope record, a temperature proxy, obtained from the 1997 Lomonosovfonna ice core (Isaksson and others, 2001).
Data Expedition data
Before regular meteorological measurements began on Svalbard in 1912, at Green Harbor, the Norwegian Meteorological Institute (DNMI) organized measurements in co-operation with hunting expeditions that normally arrived in summer for over-wintering, by equipping the expeditions with controlled meteorological instruments. In addition, several scientific expeditions took place within the time frame of interest. Here we also use temperature data published from the Robitzsch overwintering science expedition at Ebeltofhamna in Krossfjorden (Rempp and Wagner, 1916). For the Robitzsch data, daily mean temperatures were taken from report tables and averaged to create tables of monthly means. The expeditions always recorded temperatures at 0700 and 1900. These were averaged with no weighting to yield estimates of monthly means. In some years, measurements were made at 1300, and the daily minimum temperature was recorded; for these years, averages of all four measurements were used to form the monthly means. The mean difference for those years between using all measurements or just the 0700 and 1300 is less than 0.1 ºC. Table 1 gives an overview of the locations and periods of the hunting expedition data currently digitized, and Figure 2 shows graphically the time-extents of the digitized data. Figure 3b shows how the existing and extended instrumental record compare to the oxygen isotope record,
Table 1 Hunting expeditions equipped with meteorological instruments from met.no during the period before regular observations where started at Svalbard. The following abbreviations are used: T = temperature, Tn = daily minimum temperatures, D = wind direction, F = Wind force, N = cloud cover.
Svalbard Airport series: sources used
Interpolation
Barentsburg
Isfjord Radio
Svalbard airport
Longyearbyen
Green Harbor
-25
-20
-15
-10
-5
01890 1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Win
ter t
emp.
(deg
. C)
-22
-20
-18
-16
-14
-12
-10
δ18Ο
Deutsches Geophysikalisches Observatorium Adventbai 1911-1912 (ONDJFM) Akseløya expedition (ONDJFM)Svalbard Airport T (ONDJFM) Sørkapp (ONDJFM)Robitzsch Ebeltofthamna (ONDJFM) Kvalpynten (ONDJFM)Zieglerøya (ONDJFM) Kapp Lee (ONDJFM)Halvmåneøya (ONDJFM) Winter del 18O Lomonosovfonna
Figure 3 a) source data for the homogenized Svalbard Airport record. b) Comparison of the newly digizited expedition data, the Svalbard Airport record, and the winter Lomonosovfonna isotope record.
Ice core data
Ice cores are among the best archives of past climatic and environmental changes. While lower elevation glaciers like those found on Svalbard typically experience summer surface melt, recent analysis of Svalbard and other ice cores have shown that alteration of primary signals by melting is not as destructive as once thought (e.g. Koerner, 1997; Pohjola and others, 2002). The 1997 Lomonosovfonna core, taken on the highest ice field on Svalbard, yielded about 800 years of climate and environmental information (Isaksson and others, 2001). The core is dated using oxygen isotope and ion chemistry cycle-counting, identification of known volcanic horizons, and the location of peak beta-activity due to the 1963 bomb tests (Pohjola and others, in press). Dating errors are estimated to vary from ±1 years for the 1963 layer to about ±5 years for the 1800 layer (Kohler, unpublished data). Using the location of annual layers, it is possible to separate the oxygen isotope signal into peak winter and summer components (Pohjola, unpublished). The isotope data are not corrected for diffusion (e.g. Cuffey and Steig, 1998), so peak isotope values will change with time. This process is mitigated by summer melt, which leads to the creation of ice lenses, in which diffusion is greatly reduced.
Oxygen isotopes are often used a temperature proxy (e.g. Dansgaard, 1964). Although the transfer function relating snow oxygen isotope content to local temperature is complicated and not fully elucidated, there appears to be a link between Lomonosovfonna and Longyearbyen; Figure 4 shows that 5-a smoothed records of summer (JJA) and winter (here taken as Sept-May) temperatures from the Svalbard Airport record are reasonably well correlated to the summer and winter values of δ18O from the Lomonosovfonna ice core. Results
Winter temperatures for the newly digitized expedition temperature series in the period 1899-1915 are colder than for the later period 1920-1930, consistent with a step-change in climate. However, the monthly data for 1912-1913, the period of overlap between the expedition and composite Svalbard Airport data, show that temperatures in the Svalbard Airport record are consistently colder during the coldest months (Figure 5). This is the case not only for the more distant records, like Sørkapp, which we would expect to be warmer due to its more southerly location, but also for the Adventbai record from further within Isfjord.
Furthermore, the part of the Lomonosovfonna oxygen isotope record that corresponds to the early 1900s does not show a step increase in temperature in the early 20th century, even allowing for dating errors.
We conclude, therefore, that the warming between the 1910s to the 1930s was less dramatic than indicated by the existing Svalbard Airport temperature record. The δ18O record from Lomonosovfonna, when converted directly to summer and winter temperatures through the regression equations depicted in Figure 4 show that the sudden warming around 1920 appears to be part of a warming trend that started already at the end of the 1800s.
2. Instrumental and ice-core precipitation records While precipitation data were digitised for the expeditions, where such
measurements were made, no attempt has been made to homogenize these or to incorporate them into the Svalbard Airport time series. It is simply much more difficult
to obtain reliable homogenous records of precipitation from disparate sources, as compared to temperature or pressure.
1993-1997
1988-1992
1983-1987
1978-1982
1973-1977
1968-1972
1963-1967
1958-19621953-1957
1948-1952
1943-1947
1938-1942
1933-1937
1928-1932
1923-1927
1918-1922
1913-1917
4.0
4.5
5.0
5.5
6.0
-17 -17 -16 -16 -15 -15 -14 -14 -13
Summer δ18O
Sum
mer
T (
JJA
)
1913-1917
1918-1922
1923-1927
1928-19321933-1937
1938-19421943-1947
1948-19521953-1957
1958-19621963-1967
1968-1972
1973-19771978-19821983-1987
1988-1992
1993-1997
-21
-19
-17
-15
-13
-11
-9
-7-19 -18 -17 -16 -15
Winter δ18O
Win
ter
T (
JFM
)
Figure 4 Comparison of 5-a smoothed records of summer (JJA) and winter (DJF) temperatures from the Svalbard Airport record to mean summer and winter values of δ18O from the Lomonosovfonna ice core. Reduced major axis regression (assumes equal error in both x and y) of the unsmoothed data (not shown) is indicated with the dashed line.
The precipitation record for Svalbard Airport (Figure 6), as it stands, shows an increasing trend during the measurement period that is statistically significant only at the longest time-scales (essentially a linear fit to the data showing an increase of about 0.5 mm per year over 90 years). The long-term trend in precipitation can be compared to the ice-core accumulation record determined using annual layer counting and datable volcanic and beta-activity peaks. Bearing in mind that the latter record must first be corrected for layer-thinning (otherwise there would be an automatic apparent increase in accumulation), the Lomonosovfonna accumulation record also shows a long-term increase during the 1900s, although the magnitude of the increase in terms of the observed accumulation’s variability is smaller than that of the Svalbard Airport precipitation record, as is the significance level.
1:1
-30
-25
-20
- 15
- 10
- 5
0
5
10
-30 -25 -20 - 15 - 10 - 5 0 5 10
Green Harbor/Longyearbyen monthly mean T
Ebeltofthamna 1912-1913
Kapp Mitra 1912-1913
Sørkapp 1911- 12, 1914-15
Deutsches Geophysikalisches Observatorium Adventbai 1911- 1912
Ny-Ålesund vs. Svalbard Lufthavn (modern: 1975-1997)
Figure 5 Comparison of monthly data from the Svalbard Airport composite record, and various monthly temperature records from the Ny-Ålesund area (1912-1913), and from Sørkapp (1911-1915). The modern Ny-Ålesund record is also shown, compared here to the Svalbard Lufthavn record. The older series are compared to the Svalbard Airport composite record, consisting of the adjusted Green Harbor record (Nordli and others, 1996).
0
100
200
300
400
500
600
700
800
1850 1870 1890 1910 1930 1950 1970 1990
Long
year
byen
pre
cipi
tatio
n (m
m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Lom
onos
ovfo
nna
accu
mul
atio
n (m
w.e
q. a
-1)
Longyearbyen monthly mean precip
5-a smoothed Longyearbyen annual mean precip
Lomonosovfonna accumulation
5-a smoothed Lomonosovfonna accumulation
Figure 6 Ice-core accumulation record determined using annual layer counting, based on core ion and isotope chemistry, and datable volcanic and beta-activity peaks. Accumulation is corrected for layer-thinning due to compression of ice. Also shown are the mean monthly precipitations for Longyearbyen, and the 5-a smoothed versions of both records.
3. Mass balance record for Midre Lovénbreen and Austre Brøggerbreen, Svalbard
Jack Kohler, Ola Brandt Introduction
Glacier mass balance is the amount of snow and ice lost or gained on a glacier. The annual net balance nb is the sum of the summer balance sb (roughly May-Sept., in the arctic) and the winter balance wb (Sept.-May). Balance varies over the extent of the glacier, but on most glaciers there is a high degree of correlation between balance and elevation.
The typically quoted mass balances ib , where i = n,s,w for net, summer, or winter, are areally-averaged quantities calculated from
dAzbAbA ii ∫−= )(1 , (1)
where A is the glacier area, and )(zbiare the various balances as a function of elevation.
NPI currently monitors mass balance on three glaciers (Figure 7), all near Ny-Ålesund: Austre Brøggerbreen, (1967-present), Midtre Lovénbreen (1968-present), and Kongsvegen, (1987-present). The first two are the longest continuous arctic glacier mass balance time-series.
The annual balances defined above reflect the loss or gain averaged for the glacier as a whole. From a climatic perspective, though, it is the balances as a function of elevation bi(z) that are more relevant, since they reflect more directly the regional-scale climate at a particular elevation band, as opposed to glacier-averaged balances, which by Equation 1 are influenced by the individual glacier’s geometry.
While the Norwegian Polar Institute (NPI) has always measured mass balance as a function of elevation on the three study glaciers, these data have never been reported in the literature. In fact, for the two older series, these data were never recorded in any systematic way, prior to 1987. Here, we present a reconstructed time-series of net, winter and summer mass balance as a function of elevation for these two glaciers, Midre Lovénbreen (MLB) and Austre Brøggerbreen (BRG).
BRGMLB
KNG
• Ny-Ålesund
0 10 km
BRGMLB
KNG
• Ny-Ålesund
0 10 km
Figure 7. Norwegian Polar Institute mass balance study glaciers near Ny-Ålesund. BRG:Austre Brøggerbreen;MLB:Midre Lovénbreen; KNG: Kongsvegen.
Methods We use original stake data from archived field notebooks and maps at NPI, graphs of
balance as a function of elevation taken from old NPI reports (NPI Årbok 1967-1982), and in those cases where either were lacking, regression analysis of reconstructed elevation-dependent balances against reported net balances.
The specific winter balance bw, that is, the winter balance at a particular point, is calculated from any of the following data: change in exposed stake height between and spring and the previous autumn; spring snow-depth sounding; snow depth measurements in snow pits located close to the stakes; or hand-contoured accumulation maps derived from snow soundings. Snow depth is transformed to water equivalent depth using the average of the particular years’ snow densities, where reported. When density information is missing, we assume the long-term average value of 0.37 g/cm3.
The specific summer balances bs can only be calculated from the change in exposed stake lengths above the snow/firn/ice surface between autumn and the previous spring. Conversion to water equivalent is trickier than in the case of the winter balance since the density depends on the amount of each component (snow, ice, firn) that is melted. We assume that on stakes well below the equilibrium line altitude (ELA), snow and ice (when appropriate) are melted, with the winter density used to determine the snowmelt and a density of 0.9 g/cm3 used for the ice melted. For stakes above the ELA, or in general, for cases in which not all the snow is melted, the snow remaining above the previous fall minimum is assumed to have a density of 0.55, a “typical” end of melt-season value, and the same value used for melted firn. We take no special account of superimposed ice or internal accumulation since these quantities have not been reported in any of the years prior to 1993.
Finally, the specific net balances are simply the difference between winter and the summer balance for each stake: swn bbb +=
Archived stake and sounding data are not available for some years. In such cases, the data are digitised from graphs of balance as a function of elevation presented in the NPI Årbok series. For MLB, data for 1973-74, and 1976-77 are derived from the NPI Årbok. The plots in the reports normally show the balance as a function of elevation. In some of the years, stakes are displayed, but in most cases only a hand-fit line is drawn. Similarly, BRG stake data for the years 1972-1974, 1976, and 1994 are missing and are digitized from figures in the old year books. Balances as a function of elevation
Re-calculated balances as a function of elevation are then derived from a robust linear fit of the specific balances and their respective elevations. Robust fitting is less sensitive to data outliers compared to ordinary least-squares regression.
The choice of a straight-line fit is deliberate; while the usual practice of the earlier mass balance workers was to hand-fit a curved line to interpolate and extrapolate the specific balance measurements, we feel that there is no evidence to suggest that a curved relation is statistically preferable to a straight line. A variety of questions arise when using a curved line: what sort of relation to apply (piecewise, polynomial, spline); how to prevent the relation from doing unreasonable things in the regions of extrapolation; and what to do when the original specific balances are missing and we use the digitized hand-fit curves? The records from MLB and BRG suggests that winter balances are probably safely modelled using a linear relation. Summer balances are more variable; in particular years, there are discernable kinks or bends in the summer balance relation, usually at the lower elevations. However, a straight-line fit does not lead to appreciable errors for re-calculating the glacier mass balances ib since most of the glacier’s area is confined to the intermediate elevation bands, where the bi(z) are adequately described by a linear fit. And any errors in individual years are minimized when considering long-term averaged balances as function of elevation.
There are no archive data at all for MLB in 1986 and BRG in 1985-86. The bi(z) for these years must instead be reconstructed from a regression of the reported annual net balances from 1967-2000 against our re-calculated bi(z) for the years in which we have data.
We estimate errors by bootstrapping (Efron and Tibshirani, 1991). The bootstrap is made for each year and each balance term by randomly re-sampling 100 new populations from the original distribution of n specific balance measurements. Re-sampled populations with less than four unique data points are discarded since the robust fit requires a minimum of three data points. For each of the new 100 populations a robust fit is made, from which mean and standard deviation can be calculated at the desired elevation bands. Results
Reconstructed net, winter and summer balances for MLB and BRG in 50 m wide elevation bands from 0-700 m asl are shown in Figure 8, and the balances for the two glaciers are compared with respect to each other in Figure 9c. The balances in individual years are not always similar for the two glaciers; this is particularly true for the summer and net balances. Comparing the balances averaged over the entire measurement period (Figure 9a-b), however, shows that the mean balances for the two glaciers over the measurement period are similar. The mean winter balances are essentially the same, although there appears to be a slightly less negative (and thereby
net) summer balance on MLB when compared to BRG at the higher elevations. Nevertheless, the difference is small and within the estimated error limits. This gives us some confidence in our reconstruction since the two glaciers are of similar size and orientation with respect to the local topography, and should therefore experience a similar microclimate.
Figure 8. Winter, net and summer balances for each 50 m elevation interval from 700 (upper lines) - 0 (lower lines) m a.s.l. for MLB, (left) and BRG (right)
Comparing the shorter period (1987-2000) for which the mass balance as a function of elevation is available at all glaciers (Figure 10), including Kongsvegen (KNG), a much larger, more exposed glacier lying to the east of Ny-Ålesund, shows more clearly differences between the glaciers. The difference between MLB and BRG is more pronounced during this period, particularly for the summer and net balances. KNG balances are decidedly more negative; this may be due to the fact that KNG is more exposed to the effect of wind, which both reduces the winter accumulation and increases the summer melt. In addition, KNG gets less shade from surrounding mountains than do MLB and BRG, so that the glacier surface is exposed to radiative melting for a greater part of the summer days, and finally.
As a final check, the reconstructed balances are used together with the distribution of area by elevation bands and Equation 1 to recalculate the balances ib . We present only results from MLB (Figure 11) because we have not computed the areal distribution. The results show that the recalculated balances ib are, for the most part, similar to those previously presented (e.g. LeFauconnier and others, 1999), again within the error limits.
We therefore are reasonably confident that we have not distorted the balance data as far as the original stake measurements are concerned.
Finally, we perform a multiple linear regression of summer temperatures Ts (July-August) and winter precipitation Pw (Oct.-June) from Ny-Ålesund against the Midre Lovénbreen ELA (Figure 12). Earlier work showed that these parameters did a reasonable job at predicting either ELA or bn (LeFauconnier and Hagen, 1990), and for the new extended period, we obtain an r2 of 0.56. The modelled ELA is related to the summer T (°C ) and winter P (mm) via:
ws PTELA 173.02.996.11mod −+−= (2) This relation implies that an increase of 0.6 °C (1 s.d. of the observed variation) in
the mean summer temperature results in an increase in the ELA by 60 m, while an increase in winter precipitation of 91 mm (again, 1 s.d. of the observed) leads to a
decrease in the ELA by just 16 m. Therefore, most of the variation in the ELA is due to changes in the summer temperature. This can also be seen from a regression of ELA against this variable alone, which yields r2 = 0.50.
This crude calculation of the glacier’s sensitivity to the climate variables gives values similar to those found by Fleming and others (1997) using a more physical mass balance model, who report an ELA sensitivity of 90 m °C-1, using mean annual temperatures.
a
b c
Figure 9 Mean of summer, net, and winter balances for (a) MLB and (b) BRG, as a function of elevation using all years up to 2000. (c) Balances as a function of elevation for MLB and BRG plotted against each other for all years, with separate balances in right hand plots (d-f), and best-fit lines.
1987-2000
0
100
200
300
400
500
600
700
800
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Balance (m w.eq.)
Bn BRG Bn MLB Bn KNG
Bw BRG Bw MLB Bw KNG
Bs BRG Bs MLB Bs KNG
Figure 10 Mean winter, net, and summer balances as a function of elevation for BRG, MLB, and KNG for the period 1987-2000. Mean for KNG is taken from observed data while means for BRG and MLB are from reconstructed data
Figure 11 Reconstructed winter, net, and summer balances for Midre Lovénbreen (dots and line), and estimated errors, compared with those previously reported (open circles).
0
100
200
300
400
500
600
700
1965 1970 1975 1980 1985 1990 1995 2000
Observed ELA
Regression-derived ELA using summer T andwinter P
Figure 12 ELA modelled using linear regression of summer temperatures (JA) and winter precipitation (Oct.-June) from Ny-Ålesund and observed Midre Lovénbreen ELAs.
4. Glacier geometry changes on Midre Lovénbreen, Svalbard Jack Kohler, Ola Brandt, Harald Faste Aas Introduction
Four topographic maps of Midre Lovenbreen based on terrestrial or aerial photography from summers 1936, 1962, 1969 and 1977 are compared with a newer digital elevation model (DEM) constructed from 1995 aerial photographs in a digital photogrammetric work station (Figure 13).
Coordinates of the older digitized maps are in ED50 and were transformed to WGS84 using the simple transformation:
NNN EDWGS ∆−= 5084 (3a) EEE EDWGS ∆−= 5084 (3b)
Offsets are based on four of the Norwegian Polar Institute’s fixed points around Ny-Ålesund whose coordinates are known in both systems. ED50 is converted to WGS84 using the mean offsets calculated at the Ny-Ålesund, Annekammen, Exilfjellet, Sarstangen stations: E∆ = -74.680 and N∆ = -205.715, with a standard deviation of less than 1 m for the four stations. In all subsequent figures, coordinates used are UTM with the WGS 84 projection.
Individual contours for the older maps were digitized and the change in glacier elevation relative to the 1995 reference surface was calculated by linearly interpolating the E and N positions of each digitised contour point into the 1995 DEM. The difference in elevation gives the elevation change during the time period for that specific position.
In a similar manner, DGPS measurements made in the spring of 2002 were used to define the change in surface since 1995. However, all of the mapped surfaces are derived from photographs taken in summer. Therefore the underlying 2001 summer-fall ice/firn surface elevations are obtained by subtracting from the 2002 DGPS elevations the mean snow thickness measured in each elevation band as part of the mass balance program.
For the 1962, 1969, and 1977 maps, parts of the surrounding non-glaciated landscape
were digitized to evaluate the elevation quality of the older maps versus the 1995 DEM. We calculate the difference in elevation for points in the surrounding terrain lying on slopes less than 0.75, that is, with slopes comparable to those of the glacier. This analysis (Table 2) shows that the all but the 1969 map have the same mean elevation (Figure 14). Non-glaciated elevations are consistently higher in the 1969 map than the 1995 DEM and are therefore corrected by 2.3 m. Otherwise the means of the differences are within the theoretical and data-derived error limits. A similar error analysis for the 1936 map was not attempted due to the evidently rougher quality of the map construction, which was based on oblique aerial photographs, taken at greater distances than was the case for the later maps. For example, the locations of peaks in the surrounding topography do not align in a simple transformation relative to the reference 1995 DEM. Some impression of the mismatch can be obtained by comparing the 1936 glacier outline to later ones (Figure 15). Nevertheless, we can use the lowermost contours in the vicinity of the front since the surrounding peaks and the position of the outer moraine can be related by an approximately three-parameter linear transform (scaling plus E and N offset).
Figure 13 The 1995 DEM contoured according to the colorbar and the glacier outline marked thick blue. The 1962 digitized map is shown in black, with the thick black line representing the outline, and similarly for the 1969 map in green and the 1977 map in red.
Figure 14 Box plot showing the distribution of the elevation differences between non-glacier data
points lying on slopes <0.75 for the maps 1962,1969 original, 1969 corrected and 1977 and the 1995 DEM.
Calculation of glacier surface elevation and volume change Elevation differences are plotted as a function of the original elevation in the first
map of each pair (Figure 16), and show in all cases a significant elevation decrease
close to the snout, but no statistically significant difference in the firn area. The elevation change implied from map differences can be compared with the mass balance as a function of elevation summed over the appropriate time interval (thin lines, Figure 16). For the periods 1969-1995, 1977-1995 and 1995-2002, the reconstructed mass balance time-series described above are summed directly. For the 1936-1995 and 1962-1995 periods, the average balance is multiplied by the appropriate number of years.
4.33 4.34 4.35 4.36 4.37 4.38 4.39
x 105
8.7565
8.757
8.7575
8.758
8.7585
8.759
8.7595
8.76
8.7605
8.761
x 106
1907193619621969197719952002
Figure 15 Front positions and/or glacier outlines for Midre Lovénbreen
The two quantities differ (Figure 16) because we have not considered the ice flux gradients at the mass balance stakes. The change in thickness h and mass balance b are related by
eWbth
−=∂∂
(4)
where sse wxzuW +∂∂= is the emergence velocity, us and ws the surface horizontal and vertical velocities, and ∂z/∂x is the surface slope (Paterson, 1994). If the glacier were in perfect balance (∂h/∂t = 0), then the balance b at some elevation would be compensated by the emergence velocity. Evidently the upper glacier is in equilibrium, within the error limits. That the glacier is not in balance is evident from its continual retreat, and thus it is not surprising to find that on the lower glacier there is a discernable difference between the summed mass balance and the observed elevation change (Figure 16). The largest difference between mass balance and ∂h/∂t is for 1936-1995. This may be due to the lower quality of the 1936 map, to a longer-term change in the mass balance as a function elevation relation, or, most likely, to a change in
dynamics for the early part of the period, during which time the emergence velocity was presumably closer to the annual net balance at the front. However, no definitive conclusion can be drawn at his point. Front position change Front positions (Figure 15,Figure 17) are obtained directly from the maps for the years 1936, 1962, 1969, 1977, 1995. The 2002 front position was measured using DGPS in the field in September 2002. Distances along a centerline swath were averaged to give the front position for each year relative to the edge of the terminal moraine. The front position for 1907 is based on oblique photography showing the glacier at essentially its maximum position, and is therefore assumed to be 0. Positions relative to the front moraine from 1861 and 1892 made by earlier expeditions are taken from Hoel (1914), but there is some uncertainly in how well the latter measurements were documented. Liestøl (1988), for example, puts the culmination at around 1890 based on photographs taken by the Hamburg expedition. However, no measurements appear to have been made, relative or absolute, so we use the Hoel front position record for the early part.
Time period
Mean diff. for
slopes <0.75
Std. dev. for
slopes <0.75
95 % conf.
nterval for
mean
number points used
Mean diff for slopes 0.45-0.75
Std. dev. for
slopes 0.45-0.75
Mean diff for slopes 0-0.45
Std. dev. for
slopes 0-0.45
Mean diff.
for all
Std. dev.
for all
1962-1995 -0.07 4.47 ±0.16 3115 -1.05 5.12 0.60 3.83 -0.88 6.50
1969-1995 original 2.31 5.11 ±0.11 8327 2.70 5.91 1.72 3.53 4.52 6.88
1969-1995 corrected 0.07 5.11 ±0.11 8327 0.40 5.91 -0.5 3.53 2.22 6.88
1977-1995 0.07 4.75 ±0.17 3171 0.69 5.74 -0.82 2.53 0.97 7.26
Table 2 Mean and standard deviation for older map data points on surrounding mountain slopes
Midre Lovénbreen has until recently been classified as a surge-type glacier (e.g. Liestøl, 1988), due in part to the rapid advance in the 1800s, and to the generally convex appearance in pictures taken at about its maximum in 1907. However, Jiskoot and others (2000) performed a multiple linear regression of all Svalbard glaciers known to surge against various geometrical parameters, and found that Midre Lovénbreen, as well as several other small glaciers, had unusual geometry in regards to its surging classification, and suggested that the glacier be reclassified as a non-surging glacier. This permits the following analysis of the geometry changes in terms of climate; however, we urge caution in relying too much on the results, for if the glacier did surge in the late 1800s then the front position record cannot be ascribed directly to climatic factors. Relating front position change to climate
Front position data can be related to the climate for non-surging glaciers via a simple linear model (Oerlemans, 2002), which relates fluctuations in length L′ around a reference state to fluctuations in the ELA E around the reference value Eo through
+−=
dttdL
ttLc
EtE Lo)('
)('1
)( , (5)
where c is a constant relating steady-state glacier length to ELA, and tL is a characteristic length response time. Infrequent front position observations are converted into a continuous annual series by spline interpolation, the mean is subtracted to obtain a record of L´, which is then used in Equation 1 to yield a modelled ELA record that is dependent primarily on the free parameters c and tl.
0
100
200
300
400
500
600
700
-60 -50 -40 -30 -20 -10 0 10 20
1936-1995
1962-1995
1969-1995
1977-1995
1995-2002
1995-1936, 60 yrs avg. MB
1995-1962, 32 yrs avg. MB
Summed mass balance 1970-1995
Summed mass balance 1978-1995
Summed mass balance 1996-2001
Figure 16 Geometry changes for Midre Lovénbreen calculated from projecting the E and N coordinates of digitised contour lines into the 1995 NPI DEM of the area. Data for each map are binned as a function of original elevation, with only the mean in each bin presented here. Also shown are the elevation differences implied by summing the mass balance record over appropriate time intervals.
Choices of these parameters are critical to the type of record we can extract. Low values of Tl imply a glacier that can react relatively quickly. The fact that it doesn’t appear to do so implies then that the climate signal driving the glacier front variations must have been changing slowly. High values of Tl imply that the glacier is an effective filter; to get the observed relatively short timescale front position variations, ELA must have been varying significantly.
The front response time has a theoretical value given by tl =Ho/bf where Ho is the characteristic ice thickness and bf is the balance at the front. Using 100 m and –1.45 m w.eq a-1 gives a rough value of 70 years, which is reasonable within the range of values reported in the literature. A value of c of 0.35 is taken from Oerlemans (2002, Table 9.1), but needs to be investigated further. Finally, the value of Eo is adjusted so that the model and observed ELA have the same mean for 1970-2000.
The results suggest that the ELA was lower by about 100 m for the period 1860-1900 versus 1950-2000 (Figure 18). This is similar to values reported earlier by Liestøl (1988). Using the sensitivity analysis presented above, and attributing the lowered ELA solely to a temperature drop, the retreat history would indicate that summer temperatures were colder by about 1 ºC for the period 1860-1900, compared to the present (Figure 19). If there were a concurrent precipitation increase then there would be a smaller temperature difference between the late 1800s and the present, and conversely. However, the long-term trend in accumulation increase we infer above leads to a negligible contribution to a lowered ELA.
-1600
-1400
-1200
-1000
-800
-600
-400
-200
0
2001840 1860 1880 1900 1920 1940 1960 1980 2000 2020
Fron
t pos
ition
rel
ativ
e to
out
er m
orai
ne (m
)
-12
-8
-4
0
4
8
Vo
lum
e re
lati
ve to
199
5 (1
06 m3 )
Front position observations
Front position spline fit
Volume measured
Volume estimated
Figure 17 Observed front positions and volume changes calculated from historical maps of Midre Lovénbreen. Change in volume between 1907 and 1936 is estimated roughly by comparing the Hoel expedition maps to the Norwegian Polar Institute 1936 map.
Year
Front position relative to 1907
max (m)
Volume relative to 1995
(106 km3)
Change in front position since last measurement (m)
1861 -1400 1892 -200 1200 1907 0 8.5 200 1936 -148 7.9 -148 1962 -541 5.5 -394 1969 -621 4.0 -80 1977 -743 2.5 -121 1995 -938 0.0 -196 2002 -1027 -1.9 -88
Table 3 Front position and volume changes on Midre Lovénbreen.
100
200
300
400
500
600
1800 1850 1900 1950 2000
ELA
(m a
.s.l.
)
ELA for Midre Lovénbreen calculated from Lomonosovfonna data
ELA simulated with Longyearbyen met data
Modelled ELA from MLB front position
Observed ELA
Figure 18 Reconstructed and observed ELA for Midre Lovénbreen.
2
2.5
3
3.5
4
4.5
5
5.5
6
1850 1870 1890 1910 1930 1950 1970 1990 2010
Sum
mer
(JA
) te
mpe
ratu
re (
°C)
20- a smoothed Ny-Ålesund summer T from Longyearbyen record
Ny- Ålesund summer T from Longyearbyen record
Ny- Ålesund summer T
20- a smooth Lomonosovfonna summer d18O values turned into summer T
T implied from change in front position
Figure 19 Reconstructed summer (JA) temperatures for Ny-Ålesund.
5. References Ahlmann, H.W. 1934. Scientific results of the Swedish-Norwegian Arctic Expedition in the
summer of 1931. Stockholms Universitet. Birkeland, B.J. 1930: Temperaturvariationen auf Spitzbergen. Meteorologische Zeitschrift. 47,
234-236. Cuffey, K.M., and E.J. Steig. 1998. Isotopic diffusion in polar firn: implications for
interpretation of seasonal climate parameters in ice-core records, with emphasis on central Greenland. J. Glaciol. 44(147), 273-284
Dansgaard, W. Stable isotopes in precipitation. Tellus, 14(4), 436-468 Efron, B. and R. Tibshirani, 1991. Statistical data analysis in the computer age. Science, 253,
390-395 Fleming, K.M., J.A.Dowdeswell, and J. Oerlemans. Modelling the mass balance of northwest
Spitsbergen glaciers and responses to climate change. Annals Glaciology 24, 203-210 Førland, E.J., I. Hanssen-Bauer. 2000. Increased precipitation in the Norwegian Arctic: true or
false? Climatic Change, 46, 485-509. Hanssen-Bauer, I., E.J. Førland. 1998: Long-term trends in precipitation and temperature in the
Norwegian Arctic: can they be explained by changes in atmospheric circulation patterns? Climate Research 10, 143-153 pp.
Hoel, A. 1914. Résultats des campagnes scientifiques accompliés sur son yacht par Albert I}er{ Prince Souverain de Monaco, publiés sous sa direction avec le concourse de Jules Richard. Fascicules 40-44. Gouvernement à Monaco, 42(3). 68 p.
Isaksson, E., Pohjola, V, Jauhiainen, T., Moore, J., Pinglot, J-F., Vaikmäe, R., van de Wal, R.S.W., Hagen, J-O., Ivask, J., Karlöf, L., Martma, T., Meijer, H.A.J., Mulvaney, R.,Thomassen, M.P.A. Van den Broeke, M., 2001. An ice core record from Lomonsovfonna, Svalbard: viewing the isotopic, chemical and structural data from 1920-1997 in relation to instrumental records. J. Glaciol. 47 (157), 335-345.
Isaksson, E., Kohler, J., Moore, J., Pohjola, V., Igarashi, M., Karlöf, L., Martma, T., Meijer, H.A.J,. Motoyama, H.,.and van de Wal, R.S.W. Climate and sea ice variability around Svalbard - results from two ice cores. Submitted to The Holocene June 2002.
Jiskoot H., T. Murray T, P. Boyle. 2000. Controls on the distribution of surge-type glaciers in Svalbard. J. Glaciol. 46 (154), 412-422
Koerner, R.1997. Some comments on climatic reconstructions from ice cores drilled in areas of high melt. J. Glaciol. 43 (143), 90-97.
Lefauconnier, B, and J.-O. Hagen. 1990. Glaciers and climate in Svalbard, statistical analusis and reconstruction of the Brøgger glacier mass balance for the last 77 years. Annals Glaciology 14, 148-152.
Lefauconnier, B., Hagen, J-O, Ørbæk, J.B., Melvold, K., and Isaksson E., 1999. Glacier balance trends in the Kongsfjord area, western Spitsbergen, Svalbard, in relation to the climate. Polar Research 18(2), 307-313
Liestøl, O. 1988. The glaciers in the Kongsfjorden area, Spitsbergen. Norsk Geografisk Tidsskrift. 42(4): 231-238
Nordli, P.Ø., I. Hanssen-Bauer, E.J. Førland. 1996: Homogeneity analyses of temperature and precipitation series from Svalbard and Jan Mayen. Norsk Meteorologisk Institutt-Klima. Report No. 16/96. 41 pp.
Oerlemans, J. 2002. Glaciers and Climate Change. A.A. Balkema, Lisse. Paterson, W.S.B. 1994. The Physics of Glaciers. Third edition. Oxford, etc., Elsevier, 480 p. Pälli, A., J. Kohler, E. Isaksson, J. Moore, F. Pinglot, V. Pohjola, H. Samuelsson. Spatial and
temporal variability of snow accumulation using ground-penetrating radar and ice cores on a Svalbard glacier. 48(162), J. Glaciology.
Pohjola, V, Moore, J., Isaksson, E., Jauhiainen, T., Martma, T., Meijer, Vaikmäe, R., van de Wal, R.S.W. 2002a. An ice core record from Lomonsovfonna, Svalbard: investigation of depositional signals with respect to melt. Journal of Geophysical Research, 107 (D4), 10.1029/2000JD000149.
Pohjola, P., T. Martma, H. A. J. Meijer, J. Moore, E. Isaksson., R. Vaikmäe,. and R.van de Wal. 2002b: Reconstruction of annual accumulation rates back 200 years based on the water isotope record from the ice field Lomonosovfonna, Svalbard. Annals Glaciol. 35, in press.
Rempp, G. and A. Wagner. 1916. Die Temperaturverhältnisse uber Spitzbergen nach den Erhebissen der Aufstiege mit freien Registrierballonen Fesselballonen und Drachen während der Überwinterung n Advenbay 1911/12. Veröffentlichungen des Deutschen Observatoriums Ebeltofthafen – Spitzbergen. Herausgegeben von H. Hergesell, Lindenberg Erstes Heft. Druck Von Friedr. Vieweg & Sohn, Braunschweig.
Steffensen, E. P.Ø. Nordli, I. Hanssen-Bauer. 1996. Stasjonshistorie for norske meteorologiske målinger i arktisk. Norsk Meteorologisk Institutt-Klima Report 17/96.
