The Cryosphere Discussions is the access reviewed discussion forum of The Cryosphere Mapping and morphometric analysis of glaciers in Jotunheimen , South Norway , during the “ Little Ice Age ” maximum

Mapping and morphometric analysis of glaciers in Jotunheimen, South Norway, during the “Little Ice Age” maximum S. Baumann and S. Winkler Department of Geography, Physical Geography, University of Wuerzburg, Am Hubland, 97074 Wuerzburg, Germany Received: 16 June 2009 – Accepted: 16 June 2009 – Published: 29 June 2009 Correspondence to: S. Baumann (sabine.baumann@uni-wuerzburg.de) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Investigations of the glacier maximum extent during "Little Ice Age" (LIA) in South-Norway have, until recently, mainly been carried out as locally focussed studies on selected glaciers (e.g.Bogen et al., 1989;Erikstad and Sollid, 1986;Faegri, 1948;Hoel and Werenskiold, 1962;Matthews, 1977Matthews, , 2005;;Winkler, 2002).These investigations included dating of moraines, e.g. by application of lichenometry, and mapping of selected glaciers.The focus of previous studies was the region of Jostedalbreen and a few individual glaciers in Jotunheimen, as e.g.Storbreen in Visdalen.
Glaciers offer a high potential to serve as key indicators for climate change (cf.e.g.IPCC, 2007).Therefore, glaciological data not only on selected individual glaciers, but on whole glacier regions are needed for a profound analysis (Hoelzle et al., 2007;Kargel et al., 2005;WGMS, 2008;Zemp et al., 2008), to reconstruct climate change and simulate future scenarios.By using a huge data base, regional averages can be Introduction

Conclusions References
Tables Figures

Back Close
Full detected to minimise the dangers of misinterpreting specific local behaviour of selected individual "key-glaciers" as common trend.Those regionally based results are more reliable.
Mapping the glacier extent during the LIA by conventional field work is time consuming and cannot be applied to investigate more than a few selected glaciers.To study a whole region with a large number of individual glaciers, many difficult to visit on foot, remote sensing provides an alternative.Previous studies in other regions have shown the potential of using satellite imagery as an efficient tool for mapping the maximum LIA extent of glaciers on the regional scale (Csatho et al., 2005;Paul and K ä äb, 2005;Paul and Svoboda, 2009;Wolken, 2006).
Especially in Norway knowledge of the behaviour of glaciers and their response to changes of climatic factors has a very practical meaning, because 98% of the domestic electricity is produced using hydropower and 15% of the exploited runoff is derived from glacierized river basins (Andreassen et al., 2005).In this context, a detailed as possible knowledge of glacial chronology during the later Holocene is important.It serves as opportunity to verify the forecast and simulation of future glacier behaviour.
The aim of this study is to reconstruct the glacier area during LIA maximum on a regional scale.Out of these results the associated inventory data (e.g.minimum and maximum altitude) can be determined by additional use of a digital elevation model (DEM).The data are used to analyze the area change since LIA maximum until AD 2003, and to detect spatial differentiation in glacier behaviour.

Study area
The study area of Jotunheimen is located in central South Norway (61.5 • N; 8.3 • E) (Fig. 1).It shows a high-alpine character and the highest points of Norway are located here (Galdhøpiggen 2469 m a.s.l.; Glittertind 2464 m a.s.l.).The present glaciers are mostly small individual valley-type and cirque-type glaciers, separated by steep rockwalls and range from 1300 to 2300 m a.s.l.(Andreassen et al., 2008).Long-term Introduction

Conclusions References
Tables Figures

Back Close
Full recorded mass balance along a West-East profile in Southern Norway reveals a strong gradient in mass balance with decreasing mass balance turnover towards the more continental drier interior (e.g.Østrem et al., 1988).Generally, glaciers in Norway are maritime influenced, but become more continental eastwards (Andreassen et al., 2005;Hoel and Werenskiold, 1962;Winkler, 2009).The study area lies in a transitional zone between maritime and continental glaciological regimes (Matthews, 2005) and represents the most continental glacier area in Southern Norway (Østrem et al., 1988).In previous studies the timing of the LIA maximum in Jotunheimen was dated, mainly by application of lichenometry (Erikstad and Sollid, 1986;Matthews, 1974Matthews, , 2005;;Winkler, 2002).

Timing of the LIA maximum in Jotunheimen
In contrast to Jostedalsbreen lying further westwards, there are no historical documents or images of the glacier area of Jotunheimen, that would allow a distinct timing of the LIA maximum.ØYEN, 1893 andHoel andWerenskiold, 1962, respectively, report only a vague eyewitness story about the maximum extent of Storbreen in Visdalen (central West Jotunheimen), that also could not be dated precisely.Therefore, the reconstructions of glacial chronology in Jotunheimen prior to the first scientific measurements and historical photo-record around AD 1900 based mainly on lichenometry (e.g.Erikstad and Sollid, 1986;Innes, 1985;Matthews, 1974Matthews, , 1975Matthews, , 1977Matthews, , 2005;;Winkler, 2001).These studies reach a comparably high temporal resolution and relatively reliable results because of good ecological-methodological conditions in the area.
The existing lichenometric analyses for Jotunheimen allow a general statement about the LIA patterns in this region.The timing of the culmination of the LIA (LIA maximum extent) falls roughly between AD 1750 and 1800 with a distinct regional pattern (Matthews, 2005;Winkler, 2002).In West and Central Jotunheimen, the outermost moraines of the LIA often date from around AD 1750, whereas in East Jotunheimen, the related moraines are a few decades younger, i.e. date from around AD 1780/1800 (Fig. 2).Several glaciers, especially in West and Central Jotunheimen (e.g.Storbreen, Visbreen) have formed so-called "doubled" terminal moraines, i.e. the outermost position of the glacier during the "Little Ice Age" is represented by a double-ridged moraine.At the glaciers with such a phenomenon, the outer ridge always dates from the mid-, and the inner ridge from the late 18th century.Double-ridged terminal moraines do not occur in East Jotunheimen (Winkler, 2001).The existence of this spatial pattern and its glacialchronological interpretation was merged with a West-East gradient of the glaciological regime in Jotunheimen by Winkler, 2002, i.e. with a decline of the maritime influence eastwards within the mountainous region.The maximum around AD 1750 in West and Central Jotunheimen has therefore been interpreted as parallel to the maximum of the Western outlet glaciers of Jostedalsbreen.The climatological causes would have been analogously found especially in highly increased winter precipitation (De Jong et al., 2009;Nesje, 2009;Nesje and Dahl, 2003;Nesje et al., 2008a, b).Following the existing climate reconstructions (Nordli et al., 2005;Winkler, 2001), the second half of the 18th century was dominated by below-average summer temperatures.The reduced temperatures slowed down the glacier retreat of Jostedalsbreen since the LIA maximum, but had not enough impact to initiate a new maximum because of low winter precipitation (Bickerton and Matthews, 1993).In Jotunheimen and especially at the most continental glaciers to the east, these climatic conditions had been favourable for glacier growth.The related advance culminated around 1800 in East Jotunheimen and overrode all earlier LIA moraines.In the Western and Central part of Jotunheimen, glaciers advanced parallel to and near to the previous 1750 maximum position, but did not override and destroy the existing outermost terminal moraine and, hence, formed the double-ridged terminal moraines.
Even if the extent of the glaciers of Jotunheimen at AD 1750 and AD 1800 is only slightly different, and therefore for this study a uniform LIA maximum at AD 1750 as standard of comparison was chosen, the existing differences in the detailed timing of the LIA-maximum and its regional pattern have to be pointed out.(Fig. 3).All photos are cloud-free and taken at the end of the ablation season (August/September) when seasonal snow-cover is lowest.The actual year of the shot is unimportant for this study, because it focuses only on the LIA glacierextent; • Existing geomorphological-geochronological maps showing in total the LIA outline of 25 glaciers (Erikstad and Sollid, 1986;Matthews, 2005;Winkler, 2001) (Fig. 2); including timing of the moraines.These maps have no coordinate system and resemble partly more a sketch than a map; • GPS points of LIA moraine walls on eight glaciers.This data was collected AD 2008 during field-work by the first author (Fig. 2); • Digital topographic map (called "N50") in 1:50 000 (Andreassen et al., 2008)  • Digital elevation model (DEM), on the basis of N50 by Statens Kartverk with a 25 m resolution (Andreassen et al., 2008); • Glacier outlines from AD 2003 with clearly defined identification numbers (IDs) of the glaciers and basins (Andreassen et al., 2008); • Borders of hydrologic basins (called "Regine" watershed); The Regine watersheds were manually mapped with N50 as basis by NVE (Paul and Andreassen, 2009); • Glacier inventories from the AD 1980s (Østrem et al., 1988) and 1960s (Østrem and Ziegler, 1969), a listing from the AD 1930s (Hoel and Werenskiold, 1962), and digital glacier outlines from the 1980s by NVE.

Mapping LIA area
All material has to be orthorectified and georeferenced in the same projection, using Universal Transverse Mercator (UTM) projection in World Geodetic System (WGS)1984 European datum.Jotunheimen lies in zone 32N V (StatensKartverk, 1985).Digitizing of the glacier outlines was done using a GIS-software (ArcGIS 9.2 by ESRI).The glacier outlines from the 1980s and 2003 served as basis for the minimum LIA glacier extent.The three available mapping backgrounds were a satellite image, aerial photos and geomorphologic-geochronological maps.The glacier outlines were digitized manually on all three sources.On the former glacier covered foreland, the vegetation cover is absent or sparse.Therefore, the spectral signal is different between the glacier foreland and the area beyond it (Albertz, 2007;Csatho et al., 2005;Richard and Xiuping, 2006).This circumstance is used for identification of the LIA areas on the satellite image.A bands 543 composite as red, green, and blue, respectively, has been applied (cf.Andreassen et al., 2008;Paul et al., 2004a).The visibility of the former LIA glacier area was adequate for manual mapping.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
The application of a classification to the glacier foreland failed because of overlapping spectral signals between foreland and bare rock sites numerously present in the area.Inclusion of other properties, e.g.DEM or slope, in the query could enhance the result, but not convincingly.
The aerial photos cover about 50% of the study area and 86% of the glacierized area (basis is glacier extent in the 1980s; Fig. 3), mostly available in stereo pairs.They were georeferenced based on the digital topographical map using the same projection and datum definitions as for the Landsat image and orthorectified in ERDAS Imagine 9.1 (by Leica Geosystems) using the DEM as altitude reference.For each photo, four to eight ground control points (GCPs) depending on the fitting of the resulting image were collected.Moraine ridges and the outline of the foreland were used to detect and map the LIA maximum extent of the glaciers.On most glacier forelands in Jotunheimen, there were fairly well detectable moraines.The outermost glacier wall could be dated as LIA maximum due to the almost complete absence of pre-LIA moraines in this area (e.g.Matthews, 1991).Photos in stereo-pairs were additionally examined stereoscopically and analyzed.On the geomorphological maps available for this study, topographical information was very sparse apart from the moraine ridges themselves.As a consequence, aerial photos had to be used to identify their actual location.Hence, orthorectification showed only limited satisfying results, but it was good enough to make a digitalisation possible.
In addition to the position, timing of the moraines was given.
To derive one applicable outline per glacier, the different outlines were compared.The outline from the map was assumed to be the most correct in any case and chosen every time as the final outline.If there were two outlines (satellite image/aerial photo), the one with the clearer visible boundary was chosen in case of close similarity.If they deviated remarkably, a closer re-check of the sources was made in order to detect any mapping errors.By applying this method, inconsistency among possible multiple versions of the outline was resolved, and one outline for each glacier was finally derived.
The GPS points of the moraine walls were collected during field-work in sum-Introduction

Conclusions References
Tables Figures

Back Close
Full The data came both from moraines, where a geomorphologicalgeochronological map was available, and also from unmapped glaciers.In this way, on the one hand the accuracy of the projection of the map in a GIS system, on the other hand the accuracy of the mapping itself could be tested.Neither at the already mapped glaciers nor on the unmapped glaciers were the differences between the fielddata and the mapping large.Therefore, georeferencing and mapping were concluded as good results.
The differentiation and calculation of the glacier area was done with the GIS-software ArcGIS 9.2.Since the LIA maximum, a large number of the study glaciers disintegrated into separate parts due to glacier retreat and became individual glaciers.For reconstructing the LIA glaciers, the present watersheds and basins were considered, but had to be adjusted to the formerly larger glacier areas especially at the lower part of the glacier.They had to include the recently separated glaciers in one basin that were one glacier at LIA times.Hence, the border of the single glacier units could be derived in this way.The value of the glacier area was calculated automatically with the GISsoftware.Finally, all glaciers smaller than 0.01 km 2 were deleted from the inventory.This process was adopted, because it seems hard to distinguish between glaciers and perennial snowfields of this size and, secondly, because the uncertainty of mapping is limited by the pixel size of the satellite image (Andreassen et al., 2008).That has also been the reason why the glaciers of this size class were not included in the inventory of 2003.

Mapping LIA central flowlines
Mapping of the central flowlines was done manually.The LIA glacier outlines and the altitude of N50 were the basis for this.Three criteria should be observed in the mapping: on every glacier the flowline should lie in the middle of the glacier, secondly, it should be perpendicular to the altitude contours, and, thirdly, it should run through the highest and lowest points of the glacier (Svoboda and Paul, 2009).On a glacier with different arms or a broad accumulation area, the flowline can consist of several Introduction

Conclusions References
Tables Figures

Back Close
Full branches (e.g.Memurubreen in Fig. 4).In these cases, the mean of all branches was calculated and taken as length of the flowline.

Mapping LIA inventory data
The inventory data of all glaciers were calculated automatically with ArcGIS.Therefore, the glacier areas had to be overlaid on the DEM.The inventory data include, among other parameters, minimum, maximum, and mean altitude, slope, and aspect.Except for aspect, all variables could be taken directly from ArcGIS.The aspect was calculated after Paul, 2003.

Sensitivity analysis
Glacier outlines were available from one or more of these sources: satellite images, aerial photographs, and maps.For the 18 glaciers for which all three sources were available, a sensitivity analysis showed that they gave comparable results (examples in Fig. 5).The coefficient of variation was 1.7%, and all areas were within the 95%confidencel interval.Therefore, the resulting LIA glacier outlines were judged to be sufficiently accurate.
In addition, a comparison was made between the glacier areas from the satellite image and the aerial photos.Even if the result for both sources was quiet good and the differences between them were small, mapping using the satellite image showed a better accuracy than mapping using aerial photos compared to the maps.In direct comparison between the 18 glaciers used for the sensitivity analysis, the resulting areas by the satellite image showed a plus of 1.3 km 2 or 0.01%, whereas these results by aerial photos showed a plus of 2.2 km 2 or 0.02%.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion

Results
During LIA maximum, there were 233 glaciers in Jotunheimen with a total area of about 290 km 2 (Fig. 1).The individual parts of composite glaciers and ice caps were counted as single glaciers and further analyses were also made with the individual parts.The mean area of a glacier was about 1.24 km 2 .The distribution of the glaciers according to size is seen in Table 1.Most of the glaciers (number) lay in the interval 0.1-<0.5 km 2 ; the largest part of the area was in the interval 1.0-<5.0km 2 .A large number (∼68%) of the smallest glaciers represents only a small percentage of the total glacier area (∼18%; cumulative values in Table 1) and less than 5% of the biggest glaciers represent more than 30% of the total area.But in general, the glaciers are not very large: only three glaciers are bigger than 10 km 2 , and the biggest glacier in Jotunheimen was Østre Memurubreen with an area of 12.35 km 2 .
The maximum altitude of the glaciers ranged between 1500 and 2500 m a.s.l. with a mean of 2010 m a.s.l.(standard deviation (σ)=170 m a.s.l.), the minimum one between 1000 and 2400 m a.s.l. with a mean of 1590 m a.s.l.(σ=206 m a.s.l.).The highest altitude difference was found on Styggebreen with a value of 1396 m.The plot of minimum glacier altitude against area shows a lower minimum altitude of bigger glaciers compared to smaller glaciers.The coefficient of determination of the best fitting straight line through all points is ∼0.36.
The length of the central flowline of the glaciers varied between 134 and 6818 m with a mean of 1554 m.About two thirds (65.6%) of the flowlines were shorter than the mean.The median value was 1064 m.More than half of the lengths (50.2%) could be found in the interval 1.0-<5.0km (Table 1).Only eight flow lines were longer than 5.0 km and no one longer than 10.0 km.Søre Veobreen had the longest flowline with a length of 6818 m.
The LIA inventory data were compared with the glacier data from AD 2003.Because of the separation of some glaciers after the LIA maximum, the present glacier basins had to be modified, to get the past catchment of each single LIA glacier.This meant  et al., 2008), the total area declined from about 290 km 2 to 190 km 2 (about −35%) and the mean flow-length from about 1.55 km to 1.03 km (about −34%).The relative area variation between LIA maximum and 2003 is slightly greater for smaller glaciers than for larger ones but shows no clear pattern.However, the range of change at the smaller glaciers is very large (0%-−100%).The largest mean change (−47%) is seen in the interval 0.1-<0.5 km 2 , the smallest (−28%) in the interval 5.0-<10.0km 2 .
As shown in Fig. 6, more glaciers smaller than 0.1 km 2 existed in AD 2003 but non larger than 10.0 km 2 .Except for these two outermost intervals, the relative distribution of glaciers was similar at both points of time.In total, 13 glaciers disappeared between LIA and 2003, and 34 glaciers separated into two or more parts.

Comparison of sources
For one third of the aerial photos the accuracy of the fitting was unsatisfying.In these cases the sources were examined closely, resulting in detection of altitudinal differences between the topographical map (N50) and the DEM (mean error=3.3m,σ=12.6 m), perhaps because of interpolation.To quantify the error concerning the resulting LIA-areas, the altitude was extracted from the DEM.Several aerial photos, which were at first orthorecitfied with N50, were orthorectified again with the extracted altitude lines from the DEM.The LIA area was digitized on them and they were compared.With the coefficient of determination close to 1 (r 2 =0.9997), the resulting error between the resulting LIA areas is small enough to be ignored.For this reason, errors in orthorectification of the aerial photos can be excluded as sources of uncertainties of Introduction

Conclusions References
Tables Figures

Back Close
Full the resulting LIA areas.Probably, the fitting difference is implicated in the big altitude range on each aerial photo (altitude range summit -valley often more than 1000 m).In comparison with other studies (e.g.Csatho et al., 2005), the determined deviation is in the normal range.
For orthorectification of the geomorphological-geochronological maps, aerial photos had to be used additionally because of too poor topographical information on the maps.That means that the independence of the maps is lost, when the sources are already mixed up in this fundamental process.Because of the GPS-data from field work that was done on five glaciers with a map as basis, a difference between the glacier outlines of the map and the GPS-coordinates is obtained.The mean error of the resulting LIA area is about 28.2 m 2 (σ=8.8m 2 ).This error is smaller than the pixel size of the satellite image (30×30 m 2 ).The RMSE of the satellite image itself is about 0.65 pixels, which are about 20 m (Andreassen et al., 2008).The determined error is therefore too small to be taken into account and can be ignored, because the comparison and mapping is made with all three sources.Thus, the independence of the orthorectification of map material is established in this case.
The question arises whether an error is already included when a comparison is done with sources of different resolutions.The pixel size of the satellite image is quite coarse (30×30 m 2 ) compared to the aerial photos (0.4×0.4 m 2 ; J. Borgeraas, TerraTec AS, personal communication, 2008).The proportion in resolution between the satellite image and the aerial photo is about 75:1.The error in detecting glaciers depends on the glacier size (Paul et al., 2003).Given σ=3% (standard deviation), glaciers of 0.2 km 2 represent the minimum glacier size in mapping for a 30 m resolution.Mapping sources with a resolution of 5 m can be used for all glaciers, taking into account the same standard deviation.Therefore, an error is included in comparing glaciers smaller than 0.2 km 2 .Other publications (e.g.Wolken, 2006) do not mention problems between sources of different resolutions.These sources were then more or less taken as complementary products and used to compensate the disadvantages of each other.Therefore, an error Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc

Printer-friendly Version
Interactive Discussion in comparing sources may exist, but is not taken into account in an ordinary mapping process because of this consideration.

Uncertainties in mapping
An assumption in mapping the LIA outlines was the use of the outermost moraine as glacier maximum during the "Little Ice Age".An analysis of the age of moraines is not possible via remote sensing, but timing could be taken from the literature (see Sect. 3).Therefore, it was possible to conclude that the outermost moraine represents the LIA maximum and pre-LIA moraines were absent.The only exceptions are icecored moraine systems at some high-altitude cirque glaciers in East Jotunheimen, e.g. at Gr åsubreen.The outermost ridges of those complex systems could well pre-date the LIA (e.g.Winkler, 2001;Shakesby et al., 2004Shakesby et al., , 2008)).But due to the short distance between the individual ridges of those complex moraine systems, the possible error in not mapping the ridge representing the maximum LIA extent is negligibly small compared to the whole glaciated area (coefficient of variance=4.5 %).Hence, this uncertainty resulting from not knowing the precise position of the LIA moraine at undated ice-cored moraine systems can be neglected.One factor of uncertainty is implied in topography.Because it is unknown at LIA maximum, the topographical map of today was used.Thus, an error e.g. in the calculation of slope is made, because the slope of the present glacier-free area and not the slope of the former glacier tongue was used.Furthermore, an error in the differentiation of the single glaciers is possible by using the recent watersheds and basins (even if they are enlarged).It is not assured that the former topography would have produced the same drainage pattern as the one today.The order of magnitude of the uncertainty concerning topography can unfortunately not be estimated because of missing data for comparison, and the available results have to be taken as the currently best possible solution.
As mentioned in Sect.4.2, all glaciers smaller than 0.01 km 2 were removed from the inventory because the uncertainties are too big.Areas of this size can hardly be distin-

TCD Introduction Conclusions References
Tables Figures

Back Close
Full guished in perennial snow or glacier (Paul and Andreassen, 2009).Therefore, it is not assured that these areas are "real" glaciers, because there is neither any differentiation of accumulation and ablation areas, nor any discernible flow, and therefore these areas are not glaciers by definition (e.g.Paterson, 1994).As with the new Norwegian inventory (details for compilation in Andreassen et al., 2008) is followed, in which glaciers of this size are also excluded, because of too high uncertainties in mapping.
Mapping should be an objective process that has clear rules and is reproducible.Because of different interpretations by different analysts, the results would never be the same (Andreassen et al., 2008;Paul and Andreassen, 2009).Therefore, it is worthwhile to include some objective method like a classification.Unfortunately, this method was not applicable in this study.Thus, objectivity was established by using a larger number of different sources of the same area, i.e. the comparison of different image sources and the verification with the ground truth data.
The mapping of glacier flowlines was shown in Sect.4.3.For the inventory, the mean of all branches was used in the case of branched flowlines.Applying this method, the result represents the glacier as a whole, but the flowline does not exist in reality.Additionally, the connection point of the flowlines of the different branches is only estimation.With different methods, e.g.choosing the longest flowline, a length can be chosen, that can be measured on the glacier.But in e.g.choosing this method, only a small part of the glacier is taken into account.Generally, it is not evident, from which branch the biggest mass flux is coming, if there have been no other measurements.At best, the main flowline should be there, where the biggest mass flux is expected.A coincidence of the biggest mass flux with the longest length does not necessarily occur, because that is more dependent on glacier area and volume.The glacier surface during LIA maximum is not known and, therefore, the value of the volume cannot be more than estimation.Furthermore, the flowlines are constructed on the present topography and may show a different course because of a different elevation distribution on the glacier surface.Because of these uncertainties, it is useful to calculate the length of the flowlines with a consistent and simple pattern.

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion

Problems in remote-sensing methods
For mapping of the glacier outlines on basis of the aerial photos, some restrictions had to be made.Firstly, as already mentioned in Sect.4.2, not the whole study area is covered by aerial photos, but only a certain fraction (Fig. 3).Secondly, due to distortion and displacement towards the edges of the photos, the effective working size was even smaller than the photo itself.And, thirdly, it was difficult to detect the moraine ridges in those cases, where the pictures were not available as stereo pairs.Another disadvantage of the aerial photos is the time required for preparation.Depending on the size, and therefore on the number of the photos, the process of orthorectification and georeferencing of the aerial photos took much longer in total compared to the satellite image.Each single photo had to be referenced separately.Even if GCPs could be used several times, i.e. on other photos, the process itself took nearly the same time for each photo and therefore lengthened total preparation time.Though, a big advantage is the high resolution of the pictures (0.4×0.4 m 2 ), because objects could be identified very precisely, whereas the pixel size of the satellite image (30×30 m 2 resolution) is a limiting factor for mapping.
Possible sources of error in using remote sensing in high altitudes for glacier detection are cast shadow and debris-covered glaciers.Both topics are already discussed widely (Bolch and Kamp, 2006;K ä äb et al., 2002;Paul, 2001;Paul et al., 2002Paul et al., , 2004b;;Sidjak, 1999).Problems due to cast shadow were tested in the satellite image and on the aerial photos.Shadowed areas were mostly in the high-altitude accumulation areas and not on the low-lying glacier forelands important for LIA mapping.Additionally, considering cast shadow is easier in manual mapping than in classifications, because no systematic error is produced during the mapping process and each single shadowed area is controlled manually.Therefore, problems with cast shadow were only of little importance for the calculation of the LIA area.Besides, shadowed areas on the aerial photos were in many parts not totally black, and especially snow was still visible.
In the study area, there are no debris-covered glaciers.But on the satellite im- age, the glacier foreland shows the same structural and spectral signal as a debriscovered glacier.Because of manual mapping, all glacier forelands could be detected correctly without the still unsolved problems in (semi-)automatic mapping of debriscovered glaciers in remote sensing (Bolch and Kamp, 2006;Bolch et al., 2008;Paul et al., 2004a).Therefore, neither shadow nor by debris cover was a source of error.

Glacier change
The total decrease in glacier area of ∼34% (∼1.3% per decade) in Jotunheimen is lower than in other parts of the world since the LIA maximum.In the European Alps, the glacier area decreased since 1850 (LIA maximum) until 2000 by ∼49% (∼3.2 per decade) (Zemp et al., 2008).A strong trend of higher relative loss from smaller glaciers is noticed there.This pattern of the whole mountain area is mirrored in the Swiss Alps.
The area decrease there since 1850 (LIA maximum) until 1999 is ∼3.4% per decade (Paul et al., 2004b), or ∼51% in total.A higher relative loss from smaller glaciers is also visible there (Paul et al., 2007).A different relative decrease of glaciers depending on size is not visible in Jotunheimen.
LIA maximum glacier extent in the Canadian Arctic was detected by a trimline and moraine survey via remote sensing (Paul and K ä äb, 2005;Paul and Svoboda, 2009).Contrary to Jotunheimen, the number of glaciers on Cumberland Peninsula, a part of Baffin Island, at LIA maximum (there about AD 1920) is more or less uniform for all area classes less and larger than 5 km 2 .The relative area change since LIA maximum until 2000 shows also a different pattern from that observed in Jotunheimen.On Cumberland Peninsula, there is no scatter towards smaller glaciers and a dependency of relative area change on glacier size.The scatter on all glaciers of Baffin Island shows a slight increase towards smaller glaciers, but not as much as in Jotunheimen.Since LIA maximum, the glaciers in Baffin Island have lost ∼13% (∼1.6% per decade), on Cumberland Peninsula ∼12.5% (∼1.4 per decade).The decadal loss corresponds with the values of Jotunheimen, but the total time span there since LIA maximum is about three times as large as for the Canadian Arctic.The development of the glacier area in New Zealand (Southern Alps) is observed for the whole region only until 1978.The area decrease since about AD 1850 (when many glaciers remained close to their LIA maximum position) until 1978 is ∼49% (∼3.9% per decade) (Hoelzle et al., 2007).In the 1980s and 1990s, many debris-free glaciers advanced (Chinn et al., 2005), but retreat continued at most of the large debris-covered valley glaciers, often by calving over proglacial lakes.Those mainly contribute to the massive mass loss during recent years (Chinn et al., 2008).Between 1981 and 2003, the larger glaciers in Jotunheimen showed nearly no change, but there is a negative trend in Jotunheimen for the glaciers in the Eastern and Southeastern part (Andreassen et al., 2008).The relative area reduction in Jotunheimen and Breheimen together is ∼4% since the 1980s until 2003.Hence, the total area reduction in Jotunheimen since LIA maximum is lower than in New Zealand until the beginning of the 1980s, also if taken in account, that the whole area reduction of about 4% since then has taken place in Jotunheimen.

Conclusion
The application of satellite imagery and aerial photos for manual mapping LIA outlines in Jotunheimen was successful on a regional scale.By an automatizised analysis, the inventory data of these glaciers could be determined.The glacier flow lines were also mapped manually for all glaciers.All outcomes can be seen as good results of the mapping.Some uncertainties remained in the mapping process, but their influence is not big enough to call into question the reliability of the mapping results.
Only a few glaciers vanished between LIA and 2003.Overall, the glaciers were relatively bigger in size at LIA maximum, especially remarkable at the upper and lower end of the range.The relative glacier retreat is not as great as in other regions.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / EscPrinter-friendly Version Interactive Discussion Qualifikationsprogramm f ür Wissenschaftlerinnen an der Universit ät W ürzburg.The study is connected with the DFG-funded project MaMoGla (Grant WI-1701/3) and NVE.TCD Wolken, G. J.: High-resolution multispectral techniques for mapping former Little Ice Age terrestrial ice cover in the Canadian High Arctic, Remote Sens. Environ., 101, 104-114, 2006.Zemp, M., Paul, F., Hoelzle, M., and Haeberli, W.: Glacier fluctuations in the European Alps 1850-2000: an overview and spatio-temporal analysis of available data, in: The darkening peaks: Glacial retreat in scientific and social context, edited by: Orlove, B., Wiegandt, E.

4 Material and methods 4.1 Study material
Especially, if this information is kept in mind in the search of climatic causes of the "Little Ice Age" and Introduction of several basin outlines of 2003.Therefore, one LIA glacier can consist of several single 2003 glaciers.To compare both points in time, each glacier of the LIA inventory was taken as single glacier, and e.g.not the whole ice cap including the outlet glaciers was compared.In comparison to the glaciers of 2003 (Andreassen

Table 1 .
Classification of the glacier area during LIA maximum into size intervals.