Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat

(1990–2020)

Andrija Krtalić

, Ana Kuveždić Divjak

and Kristina Zeman Šteković

University of Zagreb Faculty of Geodesy, Kačićeva 26, Zagreb, Croatia

SOPRA PROJEKT LTD, Remetinečka cesta 5C, Zagreb, Croatia

Keywords: Glacier, Remote Sensing, Climate Changes, Sentinel 2, Landsat 5.

Abstract: Glaciers represent an important component of the cryosphere and are among the most sensitive indicators of

climate change and global warming. Over recent decades, climate change has significantly accelerated glacier

retreat, prompting the development of various monitoring methods, such as the Glacier Monitoring in

Switzerland (GLAMOS) program. Key parameters for understanding glacier dynamics include changes in

mass balance, length, surface area, and snow accumulation, all of which are closely tied to climatic variations.

These changes manifest as alterations in glacier morphology and mass, resulting in notable retreat compared

to previous decades when such trends were less pronounced. This study focuses on analysing changes in the

Great Aletsch Glacier, the largest glacier in the Alps, over the period 1990–2020 using remote sensing

data/techniques. Automated glacier detection and mapping methods were applied, utilizing optical satellite

data from Sentinel-2 and Landsat-5 missions. Glacier extents were delineated and analysed over the 30-year

period, integrating satellite-derived estimates with official GLAMOS data and climatological records from

the MeteoSwiss agency. The results reveal a reduction of approximately 5.2% in the surface area of the Great

Aletsch Glacier, providing valuable insights into the glacier’s response to ongoing climate change.

1 INTRODUCTION

Glaciers are massive, permanent ice bodies formed

over centuries through the accumulation and

compaction of snow into dense ice masses,

constituting a critical component of the cryosphere

(Hock and Truffer, 2024, Haeberli et al., 2007). While

most glaciers are concentrated in polar regions, they

are also found in temperate and tropical latitudes at

higher altitudes, where they play a crucial role in

sustaining human activities. Glaciers in these regions

act as vital freshwater reservoirs, support agricultural

practices, and often serve as significant attractions for

tourism.

On a global scale, glaciers are widely recognized

as one of the most visible and sensitive indicators of

climate change (Hock and Truffer, 2024); Winsvold

et al., 2016). This is particularly evident in alpine and

tropical glaciers, which are especially vulnerable to

climatic fluctuations. These glaciers are critical for

local water resources and human activities, yet they

https://orcid.org/0000-0002-9441-0179

https://orcid.org/0000-0003-1059-8395

have experienced significant retreat over recent

decades due to global warming. To address this,

numerous glacier monitoring programs have been

established, including Glacier Monitoring in

Switzerland (GLAMOS, 2025). Key glacier

parameters, such as changes in mass, length, area, and

snow cover, are directly linked to climatic variations,

with glacier length often responding to climate

change after a time lag (Nie et al., 2024, Pandey and

Venkataraman, 2012).

Given the importance of glaciers and the potential

hazards associated with their retreat, international

monitoring initiatives have been developed.

Programs such as the World Glacier Monitoring

Service (WGMS), Global Land Ice Measurements

from Space (GLIMS), the Randolph Glacier

Inventory (RGI), GLAMOS, and the Global

Terrestrial Network for Glaciers (GTN-G) provide

critical insights into the status and dynamics of

glaciers worldwide.

Krtali

c, A., Divjak, A. K. and Štekovi

c, K. Z.

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020).

DOI: 10.5220/0013481100003935

In Proceedings of the 11th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2025), pages 253-263

ISBN: 978-989-758-741-2; ISSN: 2184-500X

253

Monitoring glaciers poses significant challenges

due to their vast scale and the logistical difficulty of

collecting in situ data. Consequently, remote sensing

has become an essential tool for observing glacier

dynamics and detecting changes over time. Remote

sensing data/technologies enable the systematic

collection of Earth observation data, facilitating the

monitoring of temporal and spatial changes in

glaciers while also aiding in hazard assessments. The

literature identifies four primary remote sensing

methodologies for glacier monitoring, categorized

based on the type of data used: (1) optical and near-

infrared sensor data, (2) thermal infrared sensor data,

(3) microwave electromagnetic spectrum data, and

(4) synthetic aperture radar (SAR) interferometry

(Wen and Wang, 2024).

This study focuses on the use of optical and near-

infrared sensor data, specifically from the Landsat 5

satellite system, to analyse surface changes in the

Great Aletsch Glacier. By leveraging these methods,

the research aims to provide valuable insights into the

glacier's temporal dynamics and its response to

climatic changes over the last three decades.

2 REMOTE SENSING

APPROACHES TO GLACIER

MONITORING AND HAZARD

ASSESSMENT

Mountain glaciers in tropical and temperate regions

are very good indicators of climate change (Burns and

Nolin, 2014) and are key resources for drinking

water, maintaining agriculture and hydropower,

especially in dry seasons. The melting and retreat of

glaciers due to global warming leads to the creation

and enlargement of glacial lakes, which can pose a

significant threat in the form of glacial lake outburst

floods (GLOF). Globally, about 15 million people are

directly exposed to the impacts of potential GLOFs

(Taylor, Robinson, Dunning, 2023), and the most

endangered areas are Central Asia (India, Pakistan,

China), the Andes (Peru), and the European Alpine

countries (Switzerland, Austria, Italy), (Emmer et al.,

2022).

The continued loss of glacier ice and the

expansion of glacial lakes due to climate change,

particularly in temperate and tropical densely

populated areas, represent a global natural hazard that

requires attention to minimize loss of life and

destruction of infrastructure. Accordingly, the need to

monitor glacier changes has arisen, and global,

regional and national monitoring programs of

particular importance have been established. The

above topic is mostly dealt with by Indian and

Pakistani scientists with a focus on Himalayan glacial

phenomena (Nie et al., 2024, Pandeyand and

Venkataraman, 2016, Shafique et al., 2018),

European scientist with a focus on Alps glacials

(D’Agatha et al., 2018), and American scientists with

a focus on Andean and North American geospace

(Bolch and Wheate 2010, Romanov and Gutman,

2000).

Remote sensing-based observations of glacial

landforms have proven to be crucial for monitoring

changes, as routine in-situ data collection in isolated,

hard-to-reach mountainous regions is often very

difficult due to large spatial coverage and weather

variations (Nie et al., 2024, Wang et al., 2025). There

are significant advantages of remote sensing for crisis

management, such as continuous observations before

and after a certain event, the possibility of integrating

a wide range of different complementary sensors and

easy integration into geoinformation systems, the use

of adaptable and simple algorithms, a significant

reduction in fieldwork, archiving and reuse of images.

Potential disadvantages, some of which are being

addressed and reduced over time, include insufficient

spatial and temporal resolution, data incompatibility,

technical complexity, the high cost of access to

commercial images (sub-meter spatial resolution),

and the limited use of optical sensors depending on

weather conditions (Kerle and Oppenheimer, 2002).

The first satellite systems for remote sensing had

a spatial resolution of 80 m and a small number of

spectral channels (Landsat 2). The launch of the

Landsat 4 and 5 satellite systems with the Thematic

Mapper sensor (Landsat TM), with better spatial and

spectral resolution, enabled improved mapping and

monitoring of glacial areas. In addition to these data,

multisensor satellite data such as MODIS (Rigs et al.,

2006), ASTER (Wang et al., 2025), the upcoming

Landsat 7 and 8 missions and the new TM (Masek et

al., 2006, Shafique et al., 2018), Sentinel 2 (Frank et

al., 2016) and also radar images like Sentinel-1.

Satellite images are then integrated with global or

national meteorological data, digital elevation model

(DEM) data, and in situ data using various modelling

techniques (Ashraf et al., 2016).

During research, various multispectral channels

are most often combined in order to distinguish the

desired land cover classes by classification (Dozier,

1989, Rosenthal and Dozier, 1996), useful indices

such as the Normalized Difference Vegetation Index

(Rouse et al., 1974) and the Normalized Difference

Snow Index (Hall, Riggs, 2011) are calculated. and

today they are increasingly using machine and deep

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

254

learning methods (Chu et al., 2022). Many scientists

are engaged in measuring various parameters of snow

and ice such as the thickness and movement of

glaciers using LIDAR (Deems et al., 2013) to

estimate the melting of glaciers, or the thickness and

volume of snow and ice through aerial photographs

(Meyer et al., 2022), and to assess drinking water as

well as potential hazards.

The research presented in this paper aims to map

and analyse the spatial and temporal changes of the

largest Alpine glacier, the Great Aletsch, over a 30-

year period (1990–2020). This analysis exclusively

utilizes open-access datasets from the Landsat 5 and

Sentinel-2 satellite systems, employing selected

remote sensing methodologies to provide insights

into glacial dynamics and associated hazards.

3 MATHERIALS AND METHODS

3.1 Study Area

The object of the study is the Great Aletsch Glacier,

located in the Swiss Alps and representing the largest

Alpine ice mass with an area of 78.49 km

and a

volume of approximately 12 km

(data from 2017).

The Great Aletsch is one of 1400 Swiss glaciers

located in the Alps, and extends between 1650 and

4160 meters above sea level, with a pronounced

retreat of the lowest point (GLAMOS, 2025,

GLAMOS 1880-2021). The glacier consists of three

main tributaries – Aletschfirn, Jungfraufirn and

Ewigschneefeld (Figure 1) which merge at the

Konkordiaplatz with an ice thickness of 800 - 1000

meters and form a common 'tongue', which extends

approximately 15 km to the southwest (GLAMOS).

According to official data (GLAMOS 1880-

2021), the glacier area has decreased from 86.62 km

in 1976 to 78.48 km

(decrease of 9.4%) in 2016. The

Great Aletsch Glacier was at its peak around 1860. At

that time, it was about 3 km longer than today, and

the glacier edge was about 200 m higher in the

Aletsch Forest area, and this area stands out today as

a belt of relatively young vegetation in the landscape

(UNESCO, 2025).

Continuous monitoring of the length of the Great

Aletsch Glacier has been taking place since 1870, and

a reduction of -3459 meters in length has been

recorded in the period 1870-2021. More detailed

monitoring was established in 1918 with the

installation of the first measuring rod at 3350 m above

sea level at Jungfraufirn. The data are publicly

available on the GLAMOS website (URL4). Since

1990 (which was chosen as the reference year in this

study), the length of the glacier has decreased by -

1315.4 m (GLAMOS 2022).

The Greater Aletsch consists of two distinct areas.

The first is the accumulation area at higher altitudes,

where the annual increase in fresh snow during the

winter months is greater than the snowmelt in the

summer. The second area is the ablation area below

Konkordiaplatz, where the melting of ice is stronger

due to higher summer temperatures, and glacier mass

is lost in the lower part. It is assumed that this area

will see the largest changes in the results of remote

sensing data/methods.

Figure 1: Appearance and position of the main components

(Aletschfirn, Jungfraufirn and Ewigschneefeld) of the Great

Aletsch Glacier in the Swiss Alps (Source: GLAMOS

1880-2021).

3.2 Climatic Factors Affecting the

Great Aletsch Glacier

The development and maintenance of glaciers are

largely determined by climatological, orographic and

hydrographic features. The average annual

temperatures in the upper part of the Great Aletsch

Glacier are below freezing, while in the glacier

tongue area they are slightly above 0°C (Figure 2a).

Another very important indicator for glaciers is the

amount of precipitation. Figure 2b shows the

relatively dry climate in the glacier tongue, but the

highest precipitation in all of Switzerland is measured

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020)

255

in the Jungfraujoch glacier accumulation area, which

reaches an average of about 3000 mm per year,

influenced by precipitation from the north.

Figure 2: a) Average annual temperature in °C and b)

average annual precipitation in mm in Switzerland in the

period 1991-2020 (MeteoSwiss, 2025). The location of the

Aletsch Glacier is marked with a yellow ellipse.

The amount of snow cover is another important

factor for glacier maintenance. Official data from

MeteoSwiss show a statistically significant decline in

the number of days with fresh snow, as well as in

annual fresh snow depths, since the 1960s in many

areas of Switzerland, which is accelerating glacier

melting (MeteoSwiss, 2025).

Under the influence primarily of temperature and

precipitation, the glacier is divided into an area of

accumulation (the upper part of the glacier) and

ablation, i.e. erosion/landing (the tongue of the

glacier), which will later be important for the analysis

of the results of this research.

The Jungfraujoch station has been active since

1933 and is located at 3571 m above sea level on the

top of the glacier, i.e. in the accumulation area. The

climate temperature average for the period 1991-2020

is from -13.3°C in the coldest month of the year

(February) to 0.6°C in the warmest month (August).

From the annual temperature average monitored until

the 1930s, it is evident that the average annual

temperature is increasing slightly.

Insolation has been increasing since 1980, and

some of the sunniest years occurred after 2000. This

development is less pronounced in areas with higher

altitudes, namely alpine areas.

3.3 Input Data and Methods

All data and software used for the realization of this

research are publicly available and free of charge.

QGIS, version 3.28, with the Semi-Automatic

Classification (SCP) plugin was used as the primary

software for the realization of the work.

3.3.1 Landsat 5 and Sentinel-2 Data

This research is based on the use of images from the

Landsat 5 (launched in 1984) and Sentinel-2

(launched in 2015). The study used only images from

the Landsat 5 satellite system, in order to use data

from the same sensor for the entire time period.

Exceptionally, images from the Sentinel-2 satellite

system (launched in 2015) were used due to better

spatial resolution. The period for which Landsat 5

data is available is 1984-2012, and for Sentinel-2

from 2015 to 2020. Landsat 5 carries the passive

sensors Multi-Spectral Scanner (MSS) and Thematic

Mapper (TM) and orbits at an altitude of

approximately 705 km with a temporal resolution of

16 days. The TM sensor data were used, whose spatial

resolution per multispectral channel is 30 m. Sentinel-

2A and Sentinel-2B (launched in 2017) are equipped

with optical-electronic multispectral sensors (MSI)

that create a set of 13 spectral channels (image): 4

channels with a spatial resolution of 10 m, 6 channels

with a spatial resolution of 20 m and 3 channels with

a spatial resolution of 60 m. Sentinel-2A and

Sentinel-2B provide a temporal resolution of 5 days,

and a radiometric resolution of 12 bits. The images

are available in L1C and L2A processing modes, with

the L2A product being fully atmospherically

corrected, and the images with this level of processing

were used in this work.

For both systems, L2/L2A preprocessing level

images were downloaded in GEOtiff format. Satellite

systems and dates when the images were created:

- Landsat 5 - 18/08/1990, Scene cloud cover: 4%,

- Landsat 5 - 20/08/2000, Scene cloud cover: 11%,

- Landsat 5 - 09/08/2010, Scene cloud cover: 12%

- Sentinel-2 -07/08/2020, Scene cloud cover:

0.5%

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

256

3.3.2 Glamos Data

Glacier Monitoring in Switzerland (GLAMOS) is a

glacier monitoring program in Switzerland that

systematically documents and monitors long-term

changes in glaciers in the Swiss Alps. GLAMOS is

jointly managed by ETH Zurich and the Universities

of Freiburg and Zurich and the CC Cryospheric

Commission. The GLAMOS database contains basic

data and facts on more than 1,400 Swiss glaciers

covering an area of 641 km2, as well as

inventories/databases on changes in glacier length,

mass, volume, etc. The inventories contain

representations of the outer edges, surfaces and debris

covers of all Swiss glaciers.

For the purposes of this work, data on inventories

from 1976 (Glacier Inventory, 1976), 2010 (Glacier

Inventory, 2010) and 2016 (Glacier Inventory, 2016)

and 2022 (GLAMOS, 2022) were downloaded in

order to create a mask. with glacier boundaries for

cutting individual rasters and for accuracy control.

The downloaded data were used in the QGIS software

as a shapefile, and were further processed.

The GLAMOS Glacier Inventory data for the

years 1976, 2010, 2016 and 2022 were downloaded

in a form suitable for processing in QGIS software,

i.e. as shapefiles with associated formats (shp, shx,

qpj, prj, dbf, cpg).

3.3.3 Copernicus DEM

Digital Elevation Model (DEM) used as a basis for

display and analysis of results. The DEM (DEM

GLO-30 Public, 2021) was downloaded from the

Copernicus Global Digital Elevation Model platform

through the OpenTopography portal. The required

area is downloaded in the form of GeoTIFF format

with 30 m spatial resolution. Through the QGIS

software, a hypsometric display with shading and

color scale was created for a realistic display.

3.3.4 Input Data Transformation

The downloaded data had different reference systems

and datums (Table 1). Therefore, it was necessary to

transform all input data into a single project geodetic

datum and coordinate system so that the data could be

combined. GLAMOS databases store data in the

system used in Switzerland and Liechtenstein, while

DEM raster data, as well as satellite images, were

downloaded in global systems. Finally, the EPSG

code 3857 Pseudo Mercator system, which is most

commonly used for applications such as Google

Maps, OpenStreetMap, ArcGIS, ESRI, was chosen

for the research so that the analysis results could be

verified with real images widely available to the

public (Google maps, etc.).

Table 1: Reference systems of input data in research.

Down

loaded

data

GLAM

1976,

2010,

2016

Coper

nicus

DEM

Landsat 5

Sentinel-2

Data

analysis

EPSG

code

2056 4326 32632 3857

CRS CH1903

WGS8

4 GPS

WGS 84 /

UTM 32N

WGS 84

Pseudo-

Mercator

Further processing of all raster and vector data

was performed with EPSG code 3857. Due to their

size and the slowness of the processing process, the

satellite images were cropped to the area of interest

using the SCP tool (Clip option), and Landsat 5

images were resampled to Sentinel-2 spatial

resolution. Originally, the Landsat 5 and Sentinel 2

images were in the EPSG 32632 reference system, i.e.

WGS84, UTM zone 32N, and all images were

cropped to a smaller area, where the glacier is visible.

For further mapping and analysis, a glacier mask

was created using the GLAMOS database, with the

assumption that publicly available data is of the

highest accuracy in demarcating individual glaciers

from the Great Aletsch 3b).

3.3.5 Normalized Difference Snow Index

The Normalized Difference Snow Index (NDSI) has

been shown in the literature to be a reliable algorithm

for mapping glaciers due to the fresh snow that covers

most of the surface of each glacier (1). Reflectivity

varies by snow and ice category - freshly fallen snow

has very high reflectivity in the visible and infrared

regions. Firn (granular one-year-old snow) has 25-

30% lower reflectivity than fresh snow, and glacial

ice has high reflectivity in the blue (0.4–0.5 μm) and

green (0.5–0.6 μm) wavelength bands that drop

sharply to almost zero in the red (0.6–0.7 μm) band.

Debris and debris on the glacier surface significantly

reduce reflectivity (Pandey et. al., 2016). This

algorithm is commonly used in snow/ice mapping

applications as well as glacier monitoring. Pixels with

a value greater than 0.0 are considered to contain

snow cover (Riggs et al., 2016).

NDSI is calculated according to the formulas:

DSI = (Green-SWIR) / (Green+SWIR) (1)

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020)

257

3.3.6 The Normalized Difference Vegetation

Index

The normalized difference vegetation index (NDVI)

was created in order to later classify the land cover

more easily. NDVI (2) is the most used indicator for

detecting changes in vegetation cover. NDVI is

calculated according to the formulas:

DVI = NIR - RED / NIR + RED (2)

3.3.7 Supervised Land Cover Classification

Glacier mapping was conducted through supervised

land cover classification of the broader Great Aletsch

Glacier region. This analysis utilized satellite imagery

from the Landsat 5 system for the years 1990, 2000,

and 2010, as well as Sentinel-2 imagery for 2020.

Supervised classification, a digital image processing

technique, was employed to distinguish land cover

types based on their spectral characteristics. The

primary goal was to produce a thematic land cover

map, specifically isolating the glacier surface from

the surrounding rocky terrain and vegetated areas.

For each analysed year, one set of rasters was

created and a training set of samples for classification

was created. The more "homogeneous" areas of the

image belonging to the same land cover class were

selected, according to the judgment and experience of

the operator. The selection of ROI polygons is

facilitated by the application of the NDSI and NDVI

indices for spotting and extracting snow and

vegetation areas. RGB images created with

combinations of channels 5-4-3 for Landsat 5, and 11-

7-4 for Sentinel 2 were also used. Such combinations

of channels qualitatively distinguish snow from

vegetation and the surrounding rocky area.

Based on the selected samples for 1990, land

cover classification was performed using three

methods: Minimum Distance, Maximum Likelihood

(classical approaches), and Random Forest (a

machine learning method), implemented through the

specialized SCP plugin, for the purpose of examining

the most appropriate method. After multiple attempts,

the Minimum Distance and Maximum Likelihood

methods produced unrealistic results and were

therefore excluded from further analysis.

Random Forest (Hansch, 2025) is a classification

based on machine learning. More samples result in a

longer, but better-quality process because there are

more samples on which it "learns" how to create a

decision. Patterns are created exclusively by

manually drawing polygons for compatibility with

SNAP and SCP tools. Random Forest is a special

machine learning technique based on iterative and

random creation of a decision tree, i.e. a set of rules

and conditions that define a class.

First, the input features are defined, in this case

spectral bands by selecting samples of individual

classes. Random Forest calculates several decision

trees based on the following parameters: (1) number

of training samples: the number of pixels from the

training sample that are randomly used to train the

model; 5000 was chosen according to the instructions

from the SPC, and (2) number of trees: the number of

decision trees; the higher the number of trees, the

higher the accuracy of the model. 100 was chosen.

The separation of trees is defined according to the

principle of the Gini coefficient. A pixel is classified

as, for example, class 1 if the majority of decision

trees evaluated it as class 1.

The causes were selected along the glacier to

represent the entire class of “glaciers”, i.e. different

types of glaciers cover such as fresh snow, granular

snow and glacial ice. The class contains a very wide

range of values because glaciers consist of uneven

deposits of different types of snow, ice, debris i.e.

moraine material and the like. The class of “rocks”

has a uniform, relatively low reflectance across all

spectral channels and is marked with a light brown

colour. The class of “vegetation” has an increased

reflectance around 4.0 µm. The class of “shadows” is

separated because without it the rock and glacier

classes cannot be properly classified.

4 RESULTS AND ANALYSIS

4.1 NDSI and NDVI

The NDSI was calculated for all weather data sets,

and the one for 1990 served as a reference value for

comparisons. By analysing the image, it is possible to

conclude that the higher pixel values (darker colour)

are in the upper, accumulation part of the Great

Aletsch Glacier, which has more precipitation, i.e.

where snow is retained throughout the year. The

lower part of the glacier has lower pixel values

(weaker colour) where the snow cover is thinner 3).

Only values above 0.0 are selected and shown in

the image. Several “lines” running from

Konkordiaplatz along the entire length of the glacier

tongue can also be successfully observed, these are

moraines – accumulations of sediments of

unconsolidated glacial debris, regolith and rocks

(debris), sometimes called glacial till. The moraine is

distinguished by its appearance and remote sensing

methods, but it is an integral part of the glacier.

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

258

By analysing the NDVI for 1990, it can be seen

that in the area of interest, rock, sand and snow are

present to a large extent (values -0.1 - 0.1). Only

surfaces with values higher than 0.4 (low vegetation

such as mountain grasslands) are highlighted on this

display.

4.2 Glacier Mapping

Before analysing the results, only the glacier of

interest, the Great Aletsch, was separated from the

edges of other, surrounding glaciers. In this way, the

analysis focused only on the body of the Great

Aletsch glacier, which was the goal of this research.

All rasters (NDVIs and classification results using the

Random Forest method for each year of observation

1990, 2000, 2010 and 2020) were extracted using the

mask shown in Figure 6b (in QGIS using the

Extraction raster by mask layer option). This resulted

in multiple polygons (glacier area for each observed

year) that were merged into a single multipolygon.

Using Field calculator, a new attribute called Area

(km

) was added and the area of the polygon

representing the glacier for each observed year was

calculated using the expression “area($geometry)/

1000000”. For the sake of a more realistic display of

details, smoothing of the outline of the glacier was

done because the outer line of the glacier remained

square by converting the raster into a vector form.

Through the tool Processing Toolbox - QGIS

geoalgorithms - Vector Geometry Tools, Smooth

geometry was created for each layer for the observed

years 1990, 2000, 2010 and 2020.

The final data sets for analysis are represented by

4 vector layers – polygons of the mapped Great

Aletsch Glacier in selected years for each of the

methods used.

4.3 Results Analysis

The analysis involved a comparison of the mapped

glacier area obtained by individual methods, and

verifiable real field information and officially

available data. The analysis of the mapped areas of

the Great Aletsch Glacier obtained by the NDSI

calculation methods (Figure 3) and the Random

Forest classification method (Figure 4) indicates

certain changes in the glacier area in the observed

period, i.e. the retreat of the Great Aletsch Glacier in

selected years was detected and documented.

According to the results of the NDSI with a set of

rasters from: 1990, 2000, 2010 and 2020, a regular

glacier retreat is visible, which represents the melting

of glaciers in accordance with climate change. There

is very little change in the accumulation area, while

the pronounced retreat of the glacial mass is most

pronounced in the glacier tongue. This can also be

well monitored in the cuts of the accumulation area

(glacier fans). The glacier surface, especially

pronounced in the glacier tongue, was the largest in

1990 and has been continuously decreasing over the

30-year period (Figure 3). The glacier retreat is most

pronounced longitudinally and in accordance with the

orographic characteristics of the terrain (following

the glacier valley).

Figure 3. A single view of the all NDSI for the years: 1990,

2000, 2010 and 2020 (left). Enlarged parts of the glacier

surface to better see differences by year (right: 'fan' - top

right, 'tongue' - bottom right).

The results of the Random Forest classification

method (Figure 4) show that the areas in all observed

years are smaller and narrower than the glacier areas

obtained by calculating the NDSI. Also, the notches

in the accumulation area of the glacier (fan) are larger

and more pronounced, and the glacier tongue is

visibly narrower. The image details mainly show the

three observed years, while the glacier surface layer

for 1990 barely appears. Namely, the layers are

complex in time/chronology and logic dictates that

the reduction in the glacier area is followed from the

initial observed year when the glacier is largest to the

following years when melting is increased and the

area is smaller.

However, according to the results of the Random

Forest classification, the glacier area for 1990 is

smaller than its area for the following two observed

years. The irregularity may arise due to insufficient

education of the operator performing the

classification, inadequate samples for classification,

spatial resolution of the satellite image and other

technical

reasons. It should also be noted that the

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020)

259

Figure 4. A single view of the all classification results of

Random Forest method for the years: 1990, 2000, 2010 and

2020 (left). Enlarged parts of the glacier surface to better

see differences by year (right: 'fan' - top right, 'tongue' -

bottom right).

classifications for 1990, 2000 and 2010 were carried

out with Landsat 5 images with a spatial resolution of

30 m, and for 2020 with Sentinel-2 images with a

spatial resolution of 10 m. However, this fact did not

affect the calculation of the NDSI.

Still, these irregularities did not affect the spatial

patterns and the glacier retreat was still pronounced

during the observed years in accordance with the

orographic characteristics of the terrain. The retreat

can also be monitored on the cuts of the accumulation

area (fan). By comparing the results of the two

mapping methods, it is possible to see that the

classification according to the Random Forest method

shows much more pronounced glacier moraines with

debris material, i.e. the classification does not

systematize these parts as a glacier class but as a non-

glacial surface, which further glacier retreat in the

observed years. The above reduces the accuracy of

mapping glacier surfaces because the moraine is an

integral part of the glacier, but consists of

accumulations of sediments of unconsolidated glacial

debris, regolith and rocks (debris). Several of the

previously mentioned reasons reduce the accuracy of

the classification method for correctly mapping the

Great Aletsch Glacier and lead to the conclusion that

the Random Forest method is quite subjective and

subject to greater limitations in correctly mapping

glaciers than the NDSI calculation method. The area

of the Great Aletsch Glacier for 2010, according to

the Random Forest classification method, is 76,583

km2, which is a deviation of -1.98% compared to the

official GLAMOS data for the same year (78,131

, Table 2).

Table 2: Reference systems of input data in research.

Year Area (km2)

NDSI

Area (km2)

Random

Forest

Area

(km2)

GLAMOS

1976 86,628

1990 83.155 73.157

2000 83.583 75.623

2010 81.766 76.583 78,131

2016 78,488

2020 78.801 70.459

A comparison of the implemented glacier

mapping methods showed that, in all observed years,

the glacier area mapped using the NDSI calculation

was significantly larger than that mapped using the

Random Forest classification method (Figure 5, Table

2). The largest difference in glacier areas was also

observed for 1990, in accordance with the

aforementioned irregularity for 1990.

Figure 5. Comparison of the areas of the Great Aletsch Glacier obtained by the NDSI methods (blue tones) and Random

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

260

Table 2 lists the areas of the Great Aletsch Glacier

(in km

) calculated using the NDSI and Random

Forest classification methods, as well as the areas of

the glaciers from the GLAMOS database for 1976,

2010 and 2016 in order to make the results

comparable since reference in situ information could

not be collected.

Comparison with actual data from the GLAMOS

database for approximately the same period shows

that, in general, the NDSI index method provides

more reliable and consistent data. In all observed

years, the glacier area is larger through the NDSI

index compared to the actual data, although the

reference mapping boundary was selected as the

NDSI value ≥ 0.1 based on viewing satellite images

Timelapse Google Earth Engine. The area of the

Great Aletsch Glacier for 2010 using the NDSI index

method is 81,766 km2, which is a deviation of 4.65%

compared to the official GLAMOS data for the same

year (78,131 km

The survey did not include the collection of

reference field data; therefore, the reference boundary

of the mapping was determined arbitrarily and a value

of NDSI ≥ 0.1 was selected based on the review of

Google Earth Engine satellite images. The

assumption is that reference data from the field would

contribute to raising the reference limit around NDSI

≥ 0.4, and thus the mapped area of the glacier would

be smaller and more consistent with GLAMOS

official data.

The analysis determined that the mapping of the

Great Aletsch Glacier was more reliable and of higher

quality based on the calculation of NDSI with a

reference value of NDSI ≥ 0.1. Therefore, these

results were linked to the DEM Copernicus GLO-30

digital relief model (30m) to produce a new vector

layer and marked hypsometric elevations showing the

lowest recorded points of the mapped glacier in the

observed years. In 1990, the lowest point of the Great

Aletsch Glacier was at approximately 1637 m above

sea level, in 2000 at approximately 1670 m, in 2010

at 1735 m above sea level, and in 2020 the retreat of

the glacier mass reached 1798 m above sea level

(Figure 6).

5 CONCLUSIONS

The analysis demonstrated that glacier mapping using

the automated calculation of the NDSI index yielded

more accurate and reliable results than the Random

Forest classification method. While both methods

showed some deviations from the GLAMOS

reference data—NDSI slightly overestimating glacier

areas and Random Forest slightly underestimating

them—the

NDSI method produced spatial patterns

Figure 6. The retreat (melting) of the Great Aletsch Glacier

mapped using the NDSI method, with the lowest recorded

elevations.

that were more consistent with the orographic

characteristics of the terrain. This method allowed for

better visualization of glacier retreat, particularly

along the tongue of the glacier and in the

accumulation area, with fewer mapping errors

observed on the surface of the Great Aletsch Glacier.

According to relevant literature, the precise

determination of a reference NDSI threshold can

significantly enhance the accuracy of glacier surface

mapping. In this study, the reference threshold was

set to NDSI ≥ 0.1, informed by literature and

validated using Google Earth Engine Timelapse

imagery. For 2010, the NDSI method estimated the

area of the Great Aletsch Glacier at 81.766 km²,

deviating by 4.65% from the official GLAMOS data

(78.131 km²). Conversely, the Random Forest

method showed a smaller deviation of -1.98% for the

same year. However, visual analysis revealed that the

Random Forest method produced more significant

mapping errors, while the NDSI approach provided

more coherent and reliable results.

The findings suggest that combining these

methods (data fusion) has the potential to further

improve the accuracy of glacier mapping and

monitoring. The NDSI index, in particular, proved to

be a simple yet effective methodology for glacier

mapping, especially when the reference threshold is

verified using reliable in situ data. This study

highlights the utility of using free satellite data and

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020)

261

open-source software for monitoring changes in the

surface area of large glaciers, aligning with the

primary goal of demonstrating the feasibility of

accessible and cost-effective approaches to glacier

monitoring.

REFERENCES

Ashraf, A., Rustam, M., Khan, S.I., Adnan, M., Naz, R,

2016: Chapter 4 Remote Sensing of the Glacial

Environment Influenced by Climate Change,

Environmental Applications of Remote Sensing, edited

by Maged Marghany, 99-129.

Bolch, T., Menounos, B., & Wheate, R. (2010). Landsat-

based inventory of glaciers in western Canada, 1985–

2005. Remote Sensing of Environment, 114(1), 127–

137. doi:10.1016/j.rse.2009.08.015

Burns, P., Nolin, A., 2014: Using atmospherically-

corrected Landsat imagery to measure glacier area

change in the Cordillera Blanca, Peru from 1987 to

2010, Remote Sensing of environment, 140: 165-178.

Chu, X., Yao, X., Duan, H., Chen, C., Li, J., and Pang, W.

2022: Glacier extraction based on high-spatial-

resolution remote-sensing images using a deep-learning

approach with attention mechanism, The Cryosphere,

16, 4273–4289, https://doi.org/10.5194/tc-16-4273-

2022.

D’Agata, C., Bocchiola, D., Soncini, A., Maragno, D.,

Smiraglia, C., & Diolaiuti, G. A. (2018). Recent area

and volume loss of Alpine glaciers in the Adda River of

Italy and their contribution to hydropower production.

Cold Regions Science and Technology, 148, 172–184.

doi:10.1016/j.coldregions.2017.12.010.

Deems, J. S., Painter, T. H., & Finnegan, D. C. (2013).

Lidar measurement of snow depth: a review. Journal of

Glaciology, 59 (215), 467–479. doi:10.3189/

2013jog12j154.

DEM GLO-30 Public - European Space Agency, Sinergise

(2021). Copernicus Global Digital Elevation Model.

Distributed by OpenTopography. Accessed: 2023-05-

Dozier, J., 1989: Spectral signature of alpine snow cover

from the Landsat Thematic Mapper. Remote Sensing of

Environment, 28, 9–22.

Frank, P., Winsvold, S.H, Kääb, A., Nagler, T., Schwaizer,

G, 2016, Glacier remote sensing using Sentinel-2. Part

II: Mapping glacier extents and surface facies, and

comparison to Landsat 8, Remote Sensing, 8(7):575

GLAMOS - Glacier Monitoring in Switzerland,

https://www.glamos.ch/en/#/B36-26, (access on

20/01/2025).

GLAMOS 1880-2021, The Swiss Glaciers 1880-22020/21,

Glaciological Reports No 1-142, Yearbooks of the

Cryospheric Commission of the Swiss Academy of

Sciences (SCNAT), published since 1964 by VAW /

ETH Zurich, doi:10.18752/glrep_series

GLAMOS (2022). Swiss Glacier Length Change, release

2022, Glacier Monitoring Switzerland, doi:10.18750/

lengthchange.2022.r2022. (access on 20/01/2025).

Glacier Inventory (1976): Müller, F., Caflisch, T. &

Müller, G. 1976, Firn und Eis der Schweizer Alpen

(Gletscherinventar). Publ. Nr. 57/57a. Geographisches

Institut, ETH Zürich, 2 Vols. & Maisch, M., Wipf, A.,

Denneler, B., Battaglia, J. & Benz, C. 2000, Die

Gletscher der Schweizer Alpen: Gletscherhochstand

1850, Aktuelle Vergletscherung, Gletscherschwund-

Szenarien. (Schlussbericht NFP 31). 2. Auflage. vdf

Hochschulverlag an der ETH Zürich, 373 pp. & Paul,

F. 2004, The new Swiss glacier inventory 2000 –

application of remote sensing and GIS. PhD Thesis,

Department of Geography, University of Zurich,

Schriftenreihe Physische Geographie, 52, 210 pp.

Glacier Inventory (2010): Fischer, M., Huss, M., Barboux,

C. & Hoelzle, M. 2014, The new Swiss Glacier

Inventory SGI2010: relevance of using high-resolution

source data in areas dominated by very small glaciers.

Arctic, Antarctic, and Alpine Research, 46, 933–945.

Glacier Inventory (2016) - Linsbauer, A., Huss, M., Hodel,

E., Bauder, A., Fischer, M., Weidmann, Y., Bärtschi, H.

& Schmassmann, E. 2021, The new Swiss Glacier

Inventory SGI2016: From a topographical to a

glaciological dataset. Frontiers in Earth Science, 22,

doi:10.3389/feart.2021.704189.

Hansch, R. Handbook of Random Forests, Theory and

Applications for Remote Sensing, Series in Computer

Vision: Volume 7, World Scientific Publishing Co Pte

Ltd (30 Nov. 2025)

Hock R., Truffer, M. (2024) Glacier hazards: Will they

change in the future? Arctic, Antarctic, and Alpine

Research, Volume 56, 2024 - Issue 1

Piermattei, L., Zemp, M., Sommer, C., Brun, F., Braun, M.

H., Andreassen, L. M., Belart, J. M. C., Berthier, E.,

Bhattacharya, A., Boehm Vock, L., Bolch, T., Dehecq,

A., Dussaillant, I., Falaschi, D., Florentine, C.,

Floricioiu, D., Ginzler, C., Guillet, G., Hugonnet, R.,

Huss, M., Kääb, A., King, O., Klug, C., Knuth, F.,

Krieger, L., La Frenierre, J., McNabb, R., McNeil, C.,

Prinz, R., Sass, L., Seehaus, T., Shean, D., Treichler,

D., Wendt, A., and Yang, R. (2024) Observing glacier

elevation changes from spaceborne optical and radar

sensors – an inter-comparison experiment using

ASTER and TanDEM-X data, The Cryosphere, 18,

3195–3230, https://doi.org/10.5194/tc-18-3195-2024,

2024.

Keshri, A. K., Shukla, A. and Gupta, R. P. (2009): ASTER

ratio indices for supraglacial terrain mapping',

International Journal of Remote Sensing,30:2,519 —

524

Li, S., Leinss, S., Hajnsek, I. (). Cross-Correlation Stacking

for Robust Offset Tracking Using SAR Image Time-

Series, IEEE Journal of Selected Topics in Applied

Earth Observations and Remote Sensing, Volume: 14,

15 p.

Masek, J.G., Vermote, E.F., Saleous, N., Wolfe, R., Hall,

F.G., Huemmrich, F., Gao, F., Kutler, J., and Lim, T.K.

(2006). A Landsat surface reflectance data set for North

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

262

America, 1990-100, IEEE Geoscience and Remote

Sensing Letters. 3:68-72.

MeteoSwis, 2025, Federal Office of Meteorology and

Climatology, The climate of Switzerland,

https://www.meteoswiss.admin.ch/climate/the-

climate-of-switzerland.html, (access on 20/01/2025).

Meyer J, Deems JS, Bormann KJ, Shean DE and Skiles SM

(2022) Mapping snow depth and volume at the alpine

watershed scale from aerial imagery using Structure

from Motion. Front. Earth Sci. 10:989792. doi:

10.3389/feart.2022.989792

Nie, W., Du, Q., Zhang, X., Wang, K., Liu, Y., Wang, Y.,

Gou, P., Luo, Q., & Zhou, T. (2024). Research on

Glacier Changes and Their Influencing Factors in the

Yigong Zangbo River Basin of the Tibetan Plateau,

China, Based on ICESat-2 Data. Water, 16(18), 2617.

https://doi.org/10.3390/w16182617

Pandey, P., Venkataraman, G., 2012: Climate change effect

on glacier behavior: A case study from the Himalayas,

Earthzine: Climate Consensus.

Pandey, P., Ramanathan, A., Venkataraman, G., 2016:

Chapter 5 - Remote sensing of mountain glaciers and

related hazards, Environmental Applications of Remote

Sensing, edited by Maged Marghany, 131-162

Paul, F., Winsvold, S., Kääb, A., Nagler, T., & Schwaizer,

G. (2016). Glacier Remote Sensing Using Sentinel-2.

Part II: Mapping Glacier Extents and Surface Facies,

and Comparison to Landsat 8. Remote Sensing, 8(7),

575. doi:10.3390/rs8070575

Riggs, G. A., Hall, D. K., and Salomonson, V. V., 2006.

MODIS Snow Products User Guide. http://modis-

snow-ice.gsfc.nasa.gov/sugkc2.html (access on

20/01/2025).

Romanov, P., and Gutman, G., 2000. Automated

monitoring of snow cover over North America with

multispectral satellite data. Journal of Applied

Meteorology, 39, 1866–1880.

Rosenthal, W. and Dozier, J., 1996. Automated mapping of

montane snow cover at subpixel resolution from the

Landsat Thematic Mapper. Water Resources Research,

32(1), 115–130.

Rouse, J.W., Haas, R.H., Schell, J.A. and Deering, D.W.

(1974) Monitoring Vegetation Systems in the Great

Plains with ERTS. Third ERTS-1 Symposium NASA,

NASA SP-351, Washington DC, 309-317.

Shafique, M., Faiz, B., Bacha, A. S., & Ullah, S. (2018).

Evaluating glacier dynamics using temporal remote

sensing images: a case study of Hunza Valley, northern

Pakistan. Environmental Earth Sciences, 77(5).

doi:10.1007/s12665-018-7365-y

UNESCO (2025)- "Swiss Alps Jungfrau-Aletsch".

unesco.org. Paris, France: UNESCO World Heritage

Centre 1992-2016, UN. 13 December 2001.

https://whc.unesco.org/en/list/1037/ (access on

20/01/2025).

Wang, Q.; Yang, Y.; Hu, J.; Zhang, J.; Li, Z.; Wang, Y.

(2025). Temporal and Spatial Assessment of Glacier

Elevation Change in the Kangri Karpo Region Using

ASTER Data from 2000 to 2024. Atmosphere, 16, 110.

https://doi.org/10.3390/atmos16010110

Wen M., Wang, T., (2024) Review of SAR imaging

geodesy for glacier velocity monitoring, Geodesy and

Geodynamics, https://doi.org/10.1016/

j.geog.2024.08.004

Remote Sensing-Based Temporal Analysis of Aletsch Glacier Retreat (1990–2020)

263