Next Article in Journal
The Impact of Altimetry Corrections of Sentinel-3A Sea Surface Height in the Coastal Zone of the Northwest Atlantic
Previous Article in Journal
SAR and Optical Data Applied to Early-Season Mapping of Integrated Crop–Livestock Systems Using Deep and Machine Learning Algorithms
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Assessing Glacier Boundaries in the Ala-Archa Valley of Kyrgyzstan by Using Sentinel-1 SAR Dataset and High-Resolution UAV Imagery

1
Department of Geography, Tourism and Territorial Planning, Faculty of Geography, Tourism and Sport, University of Oradea, 410087 Oradea, Romania
2
Department of Physical Geography, Kyrgyz National University, 547-Frunze, Bishkek 720033, Kyrgyzstan
3
School of Resource and Environmental Science, Wuhan University, Wuhan 430072, China
4
Geography Department, Dimitrie Cantemir University, 540545 Targu Mures, Romania
*
Author to whom correspondence should be addressed.
Remote Sens. 2023, 15(4), 1131; https://doi.org/10.3390/rs15041131
Submission received: 29 December 2022 / Revised: 13 February 2023 / Accepted: 15 February 2023 / Published: 18 February 2023

Abstract

:
The significant retreat of glaciers in terms of climate change compels researchers to increase the frequency of studies regarding the transformations occurring in glacier boundaries. In this study, we provided glacier area delineation of Ala-Archa valley glaciers by using a Sentinel-1 SAR dataset and the InSAR Coherence technique. Since glaciers have specific patterns of movement, the low coherence method signals the presence of ice. The analysis used the pair of Sentinel-1 datasets for the summer, to ensure the lowest coherence and provide an areal estimation during the peak of ablation. The independence of the SAR images from cloud and light conditions permits us to acquire the images in a timely manner, which highly affects the results of glacier monitoring. This method has shown high potential in the mapping of debris-covered ice and the indication of dead ice. To identify and separate areas of low coherence, such as glacier lakes and unstable slopes, we used object-based mapping by using the geomorphological features of the ice. In this study, we defined a coherence value of less than 0.3 in the glacier area. Our research identified a number of 56 glaciers within the study area of 31.45 km2 and obtained highly accurate glacier maps for the glaciers with a smooth terminus. The analysis shows that automatic and manual delineation of the glaciers’ boundaries have certain limitations, but using the advantages of both scientific approaches, further studies will generate more accurate results.

1. Introduction

Studies show that the Ala-Archa valley glaciers were first systematically explored during the expedition of the Leningrad Pedagogical Institute, named after A.I. Herzen, by filed works and aerial photographs between 1960 and 1962. The result of this research was generalized in the glacier catalogue of the USSR [1]. The first monograph devoted to a detailed description of the glaciers of the Kyrgyz Ala-Too Mountain range (dynamics, morphology) was based on previous studies published in 1995 [2]. These studies mainly relied on ideas about the physical geography of the studied region. Detailed identification of the glacier area and mass changes in the Ala-Archa valley, Kyrgyz Ala-Too, northern Tien Shan Mountains since 1964 was performed by Bolch in 2015 [3]. The assessment of the glacier area and its historical evolution under climate change influences was performed in studies developed by Aizen et al. [4] and Aizen et al. [5]. The results of long-term studies on the mass balance of the Golubina glacier are summarized in the work of Azizov et al. [6]. Elevation changes to the glacier terminus were provided for the same glacier [4,6]. More recently, in 2021, the Ala-Archa valley glaciers radar zone, the equilibrium line altitude (ELA) and the accumulation–area ratio (AAR) were determined by Zholdoshbekov et al. using the Sentinel-1 SAR dataset [7].
Various investigations explored glacier ice mapping and ice divide delineation [1,2,3,4,5,6,7]. Yang et al. [8] found that the minimum amount of precipitation in the month of August is 0.52 mm/day−1. The reduced cloud cover conditions offered the opportunity of optical image survey of the Ala-Archa valley glaciers. Remote information extraction during the daytime, in the high-altitude zone, is obstructed by the increase in cloud cover due to the convection flow of air masses. Atwood et al. [9] consider that optical imagery requires good illumination, cloud-free viewing and appropriate gain settings to generate useful data for glacier mapping. In this case, it is important to implement various tools for glacier monitoring. Singh et al. [10] indicate that many methods and techniques utilize a combination of different bandwidths in the visible, NIR (Near Infrared) and SWIR (Shortwave Infrared) regions for glacier ice mapping. However, Racoviteanu et al. [11] observed that the algorithms might not work properly for dark (polluted) shaded ice sections, with debris-covered ice usually being excluded and turbid or frozen lakes being misclassified as glaciers. Paul et al. [12] indicate that debris-covered glacier detection combines the band ratio techniques presented above with the slope, because the spectral signature of debris is similar to that of surrounding moraines. Hence, the research of Atwood et al. [9] shows that the synthetic aperture radar (SAR) represents an obvious alternative to the limitations of optical satellites.
The automatic delineation of glacier area, performed with the help of the InSAR Coherence tool (Lippl et al. [10]), which we used as the basis of our methodology, covers the ablation zone. The studies of Atwood et al. [10] covered the entire glacier area, but here, description details of headwalls and nunataks are missing due to slope thresholding. In the reference studies, slope threshold has been set as 30° [10] and 26.57° (50%) [10] in order to provide glacier masking. This approach could be used for the glacier ablation zone, where profile curvature of the glacier is flat. More problematic is the determination of the boundary between glacier terminus and outwash plain [11]. In this study, we aimed to provide glacier delineation for the entire glacier area, including icefalls, corrie and headwall regions, and we used only coherence masking. The study of Shi et al. [12] provided the delineation of the glacier area by using time-series Sentinel-1 SAR images and the InSAR Coherence technique. Here, glacier area is identified by analyzing the spatiotemporal characteristics of the coherence histograms for each glacier area, such as ice, wet snow and the surrounding area, and thresholding the coherence values. For this purpose, we acquired the SAR imageries for the summer period, when glaciers show high surface velocity, respectively, coherence decreases.
In glacier mapping, we used the advantages of the SAR dataset, as independence from cloud and light conditions also conveyed the ability to detect debris-covered glaciers. The delineation of the glacier area in the debris cover was validated by using the Object-Based Classification method and high-resolution UAV images. Along with describing the technical aspects of automatic mapping of the glacier area, we aimed to assess changes in the glacier and prepare an inventory of the glacier. Using Landsat 8 optical imagery in the validation process, we excluded an area with low coherence as lakes and with the slope map lateral moraines and debris cones from the glacier area.
InSAR is a geodetic technique based on remote sensing that provides detection of displacement phenomena and instability of the land surface, such as landslides [13], land subsidence [14] and also glaciers’ movements [9,10]. Based on single-pair synthetic aperture radar interferometry (InSAR), glacier surface displacements were determined. Spatial carrier phase shifting was used to analyze the interferometric pair images considering the terrain models. Lippl et al. [10] demonstrated that InSAR Coherence represents the complex correlation between two synthetic aperture radar (SAR) images, indicating the (temporal) stability of the backscatter signal.
According to Kotlyakov [15], a glacier is a mass of ice formed by solid precipitation. The glacier moves under its own weight and generates a flowing shape. Movement represents one of the most important characteristics of the glacier. The ice, being a solid substance, has the ability to deform. Cuffey and Paterson [16] proved that the ice flux is impacted by the gravitational force, the ice substrate, the geometry of the glacier, its basin shape and snow mass accumulation. The melting water that flows at the bottom of the glacier helps it slide along the basal surface. This moving pattern of the glaciers gives us the key to determine their extent. This article is the first to consider the viscoplastic movement of the Ala-Archa valley glaciers from the point of view of mapping glaciers.
By visual observation using optical images, we found an area of low coherence in the lateral moraines and debris fans. Delineation of the glacier area was conducted using the morphological features of the glacier, implemented by Janowski et al. [17] and Abermann et al. [18]. The same methodology was implemented, along with Optical and Thermal Infrared images, in the study of Shukla et al. [19]. In the Tien Shan mountains, debris-covered glaciers were mapped by using Landsat, ASTER optical images and SRTM DEM [11]. Here, the mapping process is based on the geomorphologic features of the glacier and adjacent terrain.
As UAV aerial survey has been increasingly available for civilian purposes and its effectiveness has been repeatedly proven, it is worth examining its impact on the Ala-Archa valley’s glaciers. In 2019, specialists from the Kyrgyz Hydrometeorological Institute learned, in a joint international project, how to independently monitor a glacier using drones [20]. High-resolution images of the Turgen-Aksu glacier were obtained in order to improve glacier observations and monitoring by a group of Finnish and Kyrgyz experts from the Agency for Hydrology and Meteorology of Kyrgyzstan. Further, this article contributes to the present capabilities of UAV mapping and survey solutions which are providing new opportunities for the observation of Kyrgyzstan’s glaciers.
The main goal of this research is to delineate the glacier area of the Ala-Archa River basin from the Kyrgyz Ala-Too mountain range located in the northern Tien Shan Mountains by using Sentinel-1 synthetic aperture radar (SAR) imagery with the InSAR Coherence application (automatic delineation approach), and perform validation with the help of unmanned aerial vehicle (UAV) images (manual delineation) for the debris-covered Ak-Sai glacier. The Sentinel-1 (InSAR Coherence), high-resolution ALOS PALSAR DEM and Landsat 8 imagery were used for the entire glacier area (semi-automated delineation). The validation process of the entire glacier area is provided by comparing the areal parameters with another piece of glacier-mapping research. By this process, we obtained an idea of the trends in areal changes in glaciers.

2. Study Site

Our research is focused on the glaciers of the Ala-Archa valley. The Ala-Archa valley is situated on the northern slope of the Kyrgyz Ala-Too mountain range. The Kyrgyz Ala-Too range is the largest orographic unit in northern Tien Shan, with an extension of 454 km. The highest point is Batysh Alamudun peak, 4895 m.a.s.l., mean elevation 3700 m.a.s.l. [21] The range mainly consists of Proterozoic and Paleozoic magmatic and sedimentary rocks. Along the Kyrgyz range, there is an exposed Caledonian intrusion; however, there is also a Hercynian intrusion [22]. Kyrgyz Ala-Too, viewed from the north, is a steeply rising and highly eroded range with alpine glacial landforms in high altitude.
The water released on the northern slopes drains into the Chu River basin, which is the main artery for the (semi-) arid lowlands in Kyrgyzstan and Kazakhstan, including Bishkek, the capital of Kyrgyzstan [3]. Due to a high share of the population (30.86% of Kyrgyzstan for 2020) [23] and a large area of arable agricultural land (33.9% of Kyrgyzstan for 2022) [24], the water intake in Chui oblast, including Bishkek city (10.1% from the territory of Kyrgyzstan), is, accordingly, the highest in Kyrgyzstan (34.13% of Kyrgyzstan’s water intake for 2021) [25].
According to Kattel [22], the climatic conditions of these mountain systems are extreme continental. The research area belongs to northern Kyrgyzstan climatic region, which is characterized by a moderately warm and sufficiently humid climate, with a maximum of precipitation in spring and early summer. The distribution of precipitation on the northern slope of the Kyrgyz Ala-Too range is determined primarily by the orientation of the mountain slopes [22]. The general circulation of the atmosphere over Kyrgyzstan is mainly determined by the westerlies. The amount of precipitation registered in the research is determined by the outer location of the Kyrgyz Ala-Too range in the Tien Shan Mountain system, representing a significant orographic obstacle in the way of the humid air masses brought by the north-western circulation. The vertical gradient of the annual precipitation in the northern slope of the Kyrgyz Ala-Too range varies from 515 to 750 mm between in elevations 1100 and 3200 m.a.s.l [26]. The Ala-Archa glaciers receive an average of 700 mm annual precipitation, mainly during the spring to summer months (48% from April to June) [27]. The elevation of the Kyrgyz Ala-Too range (Batysh Alamudun peak, 4895 m) has sufficient height for the glaciers’ formation. In the high-altitude zone, where precipitation has exceeded evaporation, glaciers and snowfields occupy a considerable number of glacial mountain cirques and valleys. Overall, in the Ala-Archa valley, there are 56 glaciers covering a total area of 31.45 km2 (13% of glacier area in the entire Kyrgyz Ala-Too mountain range) between elevations of 3261 and 4720 m. A number of 33 glaciers with an area of less than 0.1 km2 have an area of 1.6 km2. The largest glaciers in the entire Kyrgyz Ala-Too range and the Golubina glacier, known as the reference glacier, are located in the same valley. The main mass of the existing glaciers is compactly located on the slopes of the axial ridge of the Alamudun spur. In addition, a small group of glaciers is located on the slopes of the Sokuluk spur on the upper reaches of the Adygene River [2].
The validation process of automatic delineation of the Ala-Archa glaciers by using the Sentinel-1 dataset and InSAR Coherence technique was performed in the ablation zone of the Ak-Sai glacier by using UAV orthogonal imagery acquired on site. Ak-Sai (N42°30′36″ E74°32′42″) is a spectacular valley-type glacier located in the eastern part of the Ala-Archa valley. The area of the glacier is 4.36 km2 with a length of 6.5 km. The glacier’s elevation ranges from 3260 m.a.s.l. to 4565 m.a.s.l. The highest part of the Ak-Sai glacier is located in the cirque, near the Korona peak. Compared with other large glaciers in the Ala-Archa valley, the Ak-Sai glacier shows a minimum elevation of the snout line. This situation can be explained with the presence of a debris cover, which was mentioned below. As with most debris covers in the Tien Shan mountains, the surface moraine is formed by the debris avalanche from framing rocks and cliffs. Debris thickness increases down-glacier with sub-debris ice melt suppressed over most of the lower tongue [28]. In this area, the highest and most prominent mountain peaks of the Kyrgyz Ala-Too range are located. The Ak-Sai glacier, with its icefall and debris-covered tongue, is the most representative glacier for the automatic and manual delineation of all Ala-Archa valley glaciers. The Ak-Sai glacier was formed at the confluence of two ice flows with the ablation zone gradually narrowing in the northern direction, being covered by a continuous system of cracks and icefalls (Figure 1). Due to the protection of the massive rock formations of Boks peak (4240 m) from direct sunlight, the ablation zone is pressed towards the left side at the bottom of the icefall. The ablation zone, below the icefall, is covered by supraglacial moraines.

3. Data and Methods

3.1. Data

The coherence estimation and the subsequent glacier outlines of the Ala-Archa valley were computed by using C-band Sentinel-1 SAR Single-Look Complex (SLC) dataset with the Interferometric Wide Swath mode (Table 1). The utilized dataset was characterized by 14 m spatial resolution, 250 km swath (3 sub-swats) and VV polarization. During data collection for our research area, the ascending orbit was chosen due to less overlay shadow compared to the descending orbit. The time of acquisition for the Sentinel-1 SAR image (Table 1) meets several requirements: the warm season, when glacier velocity is high, accordingly providing low coherence and end of ablation term, which shows accurate glacier extension maps. It is typically difficult to obtain continuous precise and clear patterns of the velocities of each glacier because of the influence of various factors, such as topographic relief, surface characteristics, image quality and resolution [29]. However, the study of Guan et al. [30] shows a high surface velocity of the glacier during the warm period of time. In the study of Shi et al. [12], the low coherence between June and September in the Qinghai–Tibetan Plateau is shown.
For the purpose of validating the debris-covered Ak-Sai glacier, 852 images were used, which were obtained using a DJI’s Mavic 2 Pro quadcopter with a 1-inch CMOS sensor. The spatial resolution is 20 cm. The UAV images were acquired on 22 August 2021, within the time frame of receiving the master and slave Sentinel-1 images to assess the InSAR Coherence. The survey covered the ablation zone of the Ak-Sai glacier with an adjacent area of approximately 1.53 km2. In order to increase the accuracy of photogrammetric processing, the image overlay was preserved by more than 80%. The average altitude was maintained at approximately 160 m in the ablation zone and 200 m in the icefall area. During the data processing period, in order to clarify glacial geomorphological features, on 23 February 2022, a second UAV survey was conducted in the same area.
During the fieldwork, a number of eight ground control points (GCP) were also established to determine the geographical reference. We captured a large number of aerial images in order to better identify the geomorphological features of the debris cover.
Accuracy assessment and glacier hypsometry parameters (including glacier inventory) for the entire glacier area were obtained by using ALOS PALSAR DEM with spatial resolution of 12.5 m [31]. We implemented the latest generated DEM for our research area. The acquisition date of implemented ALOS PALSAR DEM is 20 October 2010. For identification of the low-coherence area of glaciers, we used pan-sharpened Landsat 8 imagery.

3.2. Methodology

In this research, the coherence image was obtained using a common interferometric flowchart, computed with the European Space Agency Sentinel Application Platform (ESA SNAP) toolbox. This process consists of pre-image (IW Image Splitting, Orbital Auxiliary, Coregistration steps) and image processing (Interferogram Generation, Phase Removal, Phase Filtration, Phase Unwrapping and Terrain Correction steps) (Figure 2). Nearest neighbor algorithm was utilized in DEM and image resampling methods.
SAR products require additional orbit state vector (OSV) information to improve their spatial location accuracy. In SNAP, the corresponding processing node is called Apply-Orbit-File, which automatically downloads the OSV file and updates the scene’s metadata [32]. After Orbit auxiliary process, the research area was subsetted by using TOPSAR Split tool. During image splitting, it is important to choose the swath and sub-swath of area of interest and provide the same actual size for master and slave images.
In the introductory part, we emphasized the ability of the glacier to move. Ice, regardless of its solid state, has the ability to deform due to its own mass and slope. Internal deformation dominantly occurs through ice creep. Glacier flow through ice creep results from movement within or between individual ice crystals, with ice behaving as a nonlinear viscous material [33]. Significant motion of the glacier surface between acquisitions can reduce the coherence estimate [9].
Coherence is a product of the DInSAR methodology and a measure of the quality of the produced interferogram [34]. According to Tzouvaras et al. [35], it describes the similarity of the reflected radar signal between two images. Closson et al. [36] state that In SAR coherence is used to describe systems that preserve the phase of the received signal. Atwood et al. [9] claim that the complex coherence (ɣ) between two complex SAR images u₁ and u₂ is defined as:
ɣ = E { u 1 u 2 * } E { | u 1 | 2 } E { | u 2 | 2 }
where E {.} is the expectation value. According to Woodhouse [37], E is shown as the sum of N elements, where N is the number of samples over which the coherence is being estimated, either in time or space (for instance, over a collection of pixels). A coherence value of 1 signifies maximum stability, while 0 shows complete temporal de-correlation of the surface between the two acquisition dates [10].
Interferometric coherence was derived by using co-registration process and Sentinel-1 Single-Look Complex (SLC) product. After co-registration, the complex interferogram as the pointwise complex multiplication of corresponding pixels in the co-registered InSAR pairs was calculated [38].
In the coherence image bursts, forms of horizontal strips appeared between the sub-swaths. The next step of image processing, TOPSAR Deburst, involves removing bursts and combining sub-swaths.
The post-processing stage contains several steps, such as setting the coherence threshold, majority filtering, overlay and shadow masking, detecting and removing another area with the low coherence and accuracy assessment (Figure 3). In order to distinguish the glacier area from other low coherence zones (debris of lateral moraine), coherence threshold masking was applied. The threshold values were proposed by Lippl et al. [10] for the coherence value less than 0.2. In this research, when we used the 0.2 coherence value of thresholding and obtained highly noised glacier zones, we were forced to provide more kernel filtering process steps to close the gaps between the patches in binary mask image. This step would lead to distortion of the glacier outlines. By increasing the coherence thresholding up to 0.3, we obtained comparatively continuous glacier extension zones by preserving the glacier area shapes as in the 0.2 coherence value threshold (Figure 3).
In an intermediate step, the morphological filters (inflation and dilation) were used to fill the small holes in the existing mask and remove small patches in the mask that originated from the speckle pattern in the SAR data [9]. For this, we defined the kernel size referring to the number of pixels in both directions in a two-dimensional space [10]. Small patches were deleted by using boundary clean and majority filter tools. The kernel majority filtering tool changes the small patches of the binary mask image based on the major neighboring pixels. In our research, we used options half for replacement threshold and four for the numbers for use.
After majority filtering process, we provided the overlay and shadow masking. Due to the mountainous terrain, theslant range of backscattering rays and the inclined-facing angle of the SAR images will generate overlay and shadow phenomena. The overlay indicates radiometric distortions for terrain with slopes facing towards the sensor and opposite slopes falling into shadow. Overlay and shadow map were extracted during the terrain correction process.
After the post-image procedure, along with the glacier area, we detected low coherence in several areas. With the exception of the glaciers which have moving features, low coherence is expected in vegetation areas, water bodies [39] and landslides [35]. Additionally, coherence estimation is used for detection of unstable slopes such as landslides and areas of sediment transportation [40]. Vectorization and accuracy assessment steps contain dissection of low-coherence areas from the glacier body. Glacier lakes and low coherence zones in debris cones were detected by pan-sharpened Landsat 8 image and masked out. To remove the debris fans and lateral moraines in adjacent territory of glaciers and show continuous areas of low coherence with the glaciers, we leaned on the morphological features of the glaciers. Adjacent areas with the glacier showing low coherence were eliminated by detecting glacier margin by using the shade map which we created by using DEM. In the figure of the glacier in transverse section, which is given in the studies of Shukla et al. [41] and Shukla et al. [19], the glacier area is separated from the lateral (periglacial) moraine by a concave furrow, and the moraines themselves are represented by convex lines. Those glacier geomorphologic features are identifiable by using shaded map of high-resolution ALOS PALSAR DEM. In order to clarify the features of the objects, we used shadow maps with azimuths of 180° and 315°. In our research, glaciers with an area ≤0.5 km2 appear with a smooth surface and a more gentle slope than adjacent territory. In the glacial lakes of the Adygine, Uchitel and Maliy Alaarchinskiy glaciers, coherence was measured below 0.2. Further, glacier lakes were separated from glaciers manually by using pan-sharpened Landsat 8 [42,43] optical imagery.
After the majority filtering, we extracted the outline of the glacier area, in a validation process provided by using pan-sharpened Landsat 8 imagery. Pan-sharpening is one of the branches of data fusion that has recently received increasing attention in the remote sensing community [44,45]. Remote sensing data fusion, as one of the most commonly used techniques, aims to integrate the information acquired with different spatial and spectral resolutions from sensors mounted on satellites, aircraft and ground platforms to produce fused data that contain more detailed information than each of the sources [46]. We enhanced spatial resolution of Landsat 8 imagery by pan-sharpening technique that assumes fusion of multispectral (30 m) and panchromatic (15 m) images.
Landsat 8 imagery was acquired for 10th of August 2021. In a row with the availability of data, several conditions should be taken into account when choosing data for the study of the boundaries of the glacial zone: They should correspond to the end of the ablation season, when the elevation of the snow line is maximal, and the snow cover does not interfere with delineation of the boundaries of glaciers; it is desirable that images have low cloud cover; and images should not be taken after snowfalls [47]. It was not possible to obtain a cloud-free Landsat 8 imagery on the exact same time-scope with the InSAR Coherence survey. In the Tien Shan mountains, the peak of ablation is observed in August and the first half of September, and we took images with acquisition in this time range.
For the small glaciers, we used manual mapping provided by creating polygons of the shapefile in intended area of the glacier. To ensure accuracy in mapping, especially around areas of shade, each individual image was viewed using multiple band combinations of Red-Green-Blue as 5-4-3, 4-3-2 and 3-2-1 [48,49]. We would like to emphasize that band combination is a more efficient tool to identify glacier area in the upper zone (close to the headwall), where, due to the greater slope and the northern aspect, shade is quite frequent. High-resolution imagery makes it possible to map small glaciers and clearly detect formation of new glacial units formed as a result of the dissection of a larger glacier. The glaciers with an area ≤0.1 km2 were distinguished from perennial snow by comparing time-series Landsat 7 and Landsat 8 images (≤10 years), which were acquired during the peak of ablation. The small glaciers located around Ak-Sai and Adygine glaciers were differentiated from perennial snow by the presence of bared ice and multiple stratifications of ice during the field work.
The vectorization process was performed by using ArcGIS 10.8 software (ESRI). During the glacier area calculation, we used WGS 1984 datum.

3.3. Glacier Mapping by Using Orthogonal UAV Image

The automatic delineation of the debris-covered Ak-Sai glacier was compared with the polygons that were created by manual delineation using orthogonal UAV imagery. In our study, 852 JPEG images were obtained using DJI’s Mavic 2 Pro quadcopter with a 1-inch CMOS sensor on 22 August 2021. The survey covered the ablation zone of the Ak-Sai glacier with an adjacent area of approximately 1.53 km2 (Figure 4).
In order to increase the accuracy of photogrammetric processing, the image overlay was preserved by more than 80%. The average altitude was maintained at approximately 160 m in the ablation zone and 200 m in the icefall area. The difference in flight altitude from the ground is explained by the landform characteristics of the glacier surface. On a relatively flat terrain of the ablation zone, 160 m from the ground was enough to obtain high-quality images; in the second flight, due to an icefall, we were forced to increase the altitude from the ground to 200 m. In term of the data processing period, in order to clarify glacial geomorphological features, on 23 February 2022, a second UAV survey was conducted in the same area.
During the fieldwork, 8 ground control points (GCP) were established by using differential GPS to determine the geographical reference. We captured a large number of aerial images in order to better identify the geomorphological features of the debris cover. The obtained images were processed using the professional AgisoftMetashape software (Agisoft LLC, St.Petersburg, Russia). In this software, photogrammetric processing includes point detection, pair selection, point matching and camera location determination processes. As a result of photo alignment, sparse point cloud model can be generated by matching corresponding points on the overlapped images. Then, based on the assumed camera positions, a dense point cloud was generated. After formation of the dense cloud, in the mesh-generation step, a 3-D visualization of the surface model was generated. This level involves filling in the gaps between the cloud dots. In the areas of the dense cloud where the points were relatively sparse, the surface could be created by connection of each group of adjacent points into triangular face, which are easily connected into a mesh. The next texture generation was needed to display the image. Furthermore, the debris-covered glacier was mapped based on the orthogonal imagery. Glacier maps created by using automatic (Sentinel-1) and manual (orthogonal UAV images) approaches were compared with each other, described and analyzed for each specific area.
The presence of dead ice, supraglacial and lateral moraines in the ablation zone of the Ak-Sai glaciers constituted a difficult task for manual glacier mapping, as glacier debris cover gives the same spectral signature as the surrounding area. Therefore, when mapping the ablation zone of the Ak-Sai glacier, we used Object-Based Classification method and an orthogonal mosaic image from a UAV. Three-dimensional visualization created during meshing process and the field ground photographs helped us to analyze the geomorphological features to identify the glacier body. As Fischer et al. [50] observed, the use of high-resolution source data for glacier mapping implies a high level of visible detail of glaciers’ structures. These characteristic features of the boundary (geomorphologic feature) of Ak-Sai glacier allowed us to conduct glacier area mapping by using high-resolution UAV images: the upper part of the north-eastern side of the ablation zone is bordered by lateral moraine. The glacier snout line is identified by convexity of the glacier terminus and adjacent end moraine. As the massive rocks of Boks peak obstruct the sunlight, the glacier is pressed to south-west side, and surface tilts in a north-east direction. As we move away from Boks peak, the melting of glaciers is much more intense. In the lowest part of north-eastern site of glacier, we observed significant reduction in glacier thickness. This led to the formation of dead ices. Further degradation of the glacier can turn this area into an outwash plain, which was observed in several patches. This area is clearly defined using InSAR Coherence methods. According to Kotlyakov’s Glossary of Glacier Terminology [15], dead ice is identified as a remnant of a part of the glacier that has stopped moving (i.e., stagnant ices), located in the terminus of the glacier and that has no clear border with the main glacier body. Due to high coherence, demonstrating the absence of movement, we classified this area as dead ice and excluded it from the Ak-Sai glacier area (Figure 4).
Patches of bared ice have been identified in the lateral moraines. As far as they do not form a single body with the main glacier, the authors decided to classify them as a dead ice. According to Kotlyakov’s Glossary of Glacier Terminology [15], dead ice is the remnants of glacier that has stopped moving. Mainly located in the terminus of glacier and often does not have a clear border with the main glacier area. In our study, attempts were made to divide the glacier body from the glacier area by estimating the coherence. Our study allowed to differentiate between glacier and dead ice.
In the validation process of the automatic delineation of the glacier outline, the methodology proposed by Lippl et al. [10] was used. In this research, the accuracy assessment was provided by the creation of four polygons: true positive and false negative value within the reference glacier outline and false positive and true negative for beyond the reference glacier outline. A true positive refers to the glacier area in the measured and referenced datasets, a false negative for the glacier area only in the referenced dataset, false positive for the glacier area in measured dataset and true negative for the absence of glacier area in both the measured and referenced datasets (Figure 5).
We used the surrounding area of the glacier falling on the UAV orthogonal imagery as a buffer zone, instead of a radius approach. The false-negative zone is imposed on the icefall and speckle patterns, which were preserved after majority filtering due to large area. The same glacier area class includes the thin ice zone on the north-east side of the glacier ablation zone. Under conditions of a negative glacier mass balance, the ablation zone is retreating by ablative agents, mainly due to surface melting. We observed that by losing ice thickness, the ends of the glacier tongue can turn into clusters of scattered blocks of dead ice. The main feature of dead ice (stagnation when motion ceased) [15] explains the increased InSAR coherence in this part of the glacier.
As Atwood et al. [9] indicate, in mountainous terrain, both glaciers and surrounding topography may offer low coherence interferometry. The false-positive zone is represented by debris cover with high probabilities of rock falls. Here, in comparison with images from UAVs and field photographs, the false-positive area corresponds to debris fans and lateral moraines. During the imagery post-processing, small patches with lower coherence were removed after the comparison with other data sources (Figure 5). One of the small areas with low coherence was overlapped by a hanging glacier around Boks peak (4240 m).

4. Results and Discussion

4.1. Glaciers Hypsometry and Morphology

The results of glacier delineation are shown in Table 2. Our research identified a number of 56 glaciers with a total area of 31.45 km2 in the Ala-Archa valley. The mean area of glaciers is 0.49 km2. A number of five glaciers are larger than 2 km2, occupying 62.04% of the total area of glaciers. The largest glacier in the Kyrgyz Ala-Too range, Golubina, has an area of 5.138 km2, Ak-Sai has an area of 4.357 km2, Tuyuk has an area of 4.127 km2, Top-Karagai has an area of 3.052 km2 and Adygine has an area of 2.846 km2. According to the area classes, the number of glaciers is dominated by small glaciers. Glaciers with an area of less than 0.5 km2 share 82% of the total number and cover 11.7% of the area.
The hypsometry of glacier areas is provided by masking the obtained glacier area with ALOS PALSAR (12.5 m spatial resolution) DEM. In Figure 6A, we present the hypsometry of the glacier areas. The snout line of the debris-covered Ak-Sai glacier descends to 3261 m.a.s.l. The glacier snout line elevation is well correlated with the glacier size classes. The snout line altitude of the glaciers with an area larger than 1.0 km2 is 3523 m.a.s.l., for those between 0.5 and 1.0 km2 it is 3580 m.a.s.l. and 3787 m.a.s.l. for those with 0.1–0.5 km2 glacier area. The mean altitude of glaciers’ area is 3990 m.a.s.l. Most of the glacier area is located between 3800 and 3900 m.a.s.l., sharing 21.68% of total glacier area, 17.15% between 3900 and 4000 m.a.s.l. and 16.42% between 3700 and 3800 m.a.s.l.
The characteristics of the glacier aspects are defined by the distribution of glacier areas along the northern slope (Figure 6B) due to the lower insolation of the northern side and windward position relative to the main moisture-bearing air masses. It is worth noting that a more objective comparison of the slope distribution of glacier area can be performed by including glaciers of the southern macro-slope of the Kyrgyz Ala-Too range. The glaciers of the northern macro-slope are much larger than those on the southern slope. This distribution of glaciers depends on the morphological features of the Kyrgyz Ala-Too range. The main axis of the ridge is located further south from the highest peaks. Therefore, the northern spurs of the Kyrgyz Ala-Too range have significant glacier areas. Due to these circumstances, the large valley glaciers are confined to the slopes of the northern spurs [2]. The north-western slope has a larger glacier area than the north-east side. Such a situation can be explained by the concentration of the largest glaciers on the Alamudun spur, which has general expansion to the north-western side. According to the analysis of the distribution of glacier area in each aspect, 46.01% of glaciers face north, 45.95% to the north-west, 5.02% to the north-east, 1.51% to the south, 0.45% to the south-west, 0.39% to the west, 0.31% to the east and 0.33% to the south-east.
Due to alpine types of landforms with highly dissected relief, the Ala-Archa valley has a significant variety of glacier types. We observed that most of the glaciers belong to the hanging and corrie types; there are many hanging glaciers and various transitional forms (Table 3). Mostly, large glaciers belong to the valley type (63.6% of the total area). Despite the large number (65.6%), corrie and hanging glaciers occupy less area (6% of the total area). Transitional forms, such as corrie–valley and hanging–valley glaciers (12% of the total number) have a significant area (6.8%). Glaciers of the depression formed as a result of the collapse of valley glaciers are widespread. Usually, these glaciers occupy the upper reaches of the valleys [1]. Glaciers with id no. 251-3, 251-4, 254-1 and 254-2 might be included in the couloir glacier.

4.2. Updating the Inventory of Glaciers

Based on the obtained glacier area, an inventory of glaciers was composed (Table 4). Glacier numeration is provided according to the Soviet glacier inventory [1]. In the past, larger glaciers have been fractured into several smaller ones. The newly appeared glacier’s numeration is provided by adding additional numbers (for example, the numeration of Adygine glacier is 235; the number 235-1 is given to the glacier separated from the Adygine glacier) and these secondary numbers ascend with increasing distance from the main glacier area (Figure 7). Small glaciers (less than 0.08 km2) are not represented in the reference glacier inventory [1]. These glaciers are numerated using numerical fractions. For example, glaciers located between 256 and 257 are numbered as 256/1, 256/2 and so on. The newly performed glacier inventory contains glacier name (as a reference, we used the glacier inventory, 1973 edition [1] and 1: 50,000 toposheets), centroid coordinates (WGS 1984 datum used), river catchment (1: 50,000 toposheets), glacier type (glacier inventory, 1973 edition), mean aspect (glacier inventory, 1973 edition, and aspect map obtained by ALOS PALSAR DEM) and altitudes of lowest and highest points (ALOS PALSAR DEM). The length of the glacier along the central line was estimated using the approaches proposed by López et al. [51]. According to this reference, the center line of glaciers should be provided by using the following criteria: (i) glacier length is represented by a line which corresponds to the longest distance followed by the glacier; (ii) the length was measured from the lowest to the highest point of the glacier; (iii) the origin is the central position of the glacier tongue; and (iv) the length follows surface flow trajectories as they are identifiable on the satellite image. In addition, attention should be paid to the following points [52]: (v) cross elevation contours perpendicularly; (vi) flows strictly downhill; (vii) does not cut corners; (viii) is in the center of the glacier below z m e d ; (ix) ends at the lowest glacier point. In our research, area criteria number 5 and 6 were implemented according to mean contour lines (for each 100 m), as the contour lines with the least interval show complex lines in rough surface, declining from the middle glacier. For the Tuyuk glacier, where the highest point is situated in the middle region, the length of the glacier is given for the longest extension and between the highest and lowest glacier points.
The glacial area has a significant decreasing tendency in the Ala-Archa valley (between 1963 and 2003, −15.22%, and 1981 and 2003, −10.61% [4]) and adjacent area (between 1970 and 2000 in Ili and Kungoy Ala-Too, 12%, Teskei Ala-Too, 8% [53]). Compared with the results of Bolch [3], between 2010 and 2021, the glacier area decreased from 33.3 to 31.45 km2 (−5.56%). In previous glacier studies along the area, the glacier number and area classes were indicated. Obviously, due to glacier degradation and consequently by separation them into several parts, the number is increasing. Due to glacier area size thresholding and majority filtering, it is a problematic task to identify small glaciers using InSAR Coherence and Sentinel-1 SAR dataset.
Compared to previous research [3], decreases in glacier area took place through glacier retrievals in the snout line, especially in the Top-Karagai, Adygine and Adyginetor glaciers. The glacier no. 257 in the Murat River basin and the adjacent area of glaciers no. 233, 229-1 and 256-4 have completely disappeared. Glacier no. 241-1 is divided into two parts.
The analysis finds that during the period 1964 and 2021, the Ak-Sai glacier surface decreased from 4.8 to 4.357 km2 (Table 5). Including glacier degradation, the considerable change in glacier area is explained by the dissection processes (in our inventory, identified with id no. 255-1 and 255-2).
In the table are presented area changes of the large glaciers since 1964. A comparable analysis for other large glaciers such as glaciers no. 243, 244 and 245 was not provided, since in the schematic map of Bolch [3] for 1999 and Aizen et al. [4] for 2003, they were represented as an integrated glacier body and, accordingly, areal changes are not specified.
We feel that the large number of glaciers and their structural changes can be explained correlating the glacier retreat under climate change conditions with the implementation of different assessment approaches and tools during the research. In the reference studies, the degradation of glaciers is explained due to climate changes [4] and, accordingly, balance mass losses [3]. Mass balance estimation and consequent studies as glacier length estimations of the reference Golubina glacier [6] confirm this point of view. One of the most important parameters of glacier dynamics, as ELA, shows a slightly increasing tendency. The ELA varied from 3848 m.a.s.l. in 1997 [54] to 3868 m.a.s.l. (mean ELA between 2015 and 2019) [7].

5. Conclusions

Our study revealed that automatic and manual delineation of glacier boundaries remains interdependent, especially for delineating the entire area of glaciers and their various typologies. This study aimed to assess the boundaries of Kyrgyzstan’s Ala-Archa glaciers with more perspective tools such as InSAR Coherence by using Sentinel-1 SAR imagery and providing validation with high-resolution UAV imagery. Cloud-free conditions led to acquired imagery for a specific time, for glacier survey at the end of the ablation and with a minimum amount of snow cover, which met the requirements of the study.
Even with all the advantages of cloud cover penetration technology, we faced considerable difficulties determining the glacier boundaries in the accumulation zone by slope and coherence inception. There are several limitations in this research related to the low-coherence icy debris fans or to the glacier areas situated below the bergschrund which have a higher positive slope and do not fall into intended slope classification (<30°).
The article sheds light on the Ala-Archa valley lateral moraines situated around the glacier’s snout lines that provide low coherence but can be identified as a glacier zone if they reach a given threshold. The study confirms that InSAR Coherence represents the finest tool for distinguishing glacier and dead ice. Among the separated plots of dead ice, we identified a dead ice zone which was combined with the main glacier body.
These results highlight the importance of the automatic identification of glacier areas using InSAR coherence that can be implemented for medium to large glaciers. The estimation of the glacier-covered area using the InSAR Coherence process can work for medium and large glaciers, but not for small ones, because without checking by using high-resolution images, the small glaciers would be identified as speckle noise. Therefore, almost all the small areas with low coherence within the assumed glacier areas were erased during the majority filtering. The glaciers with an area less than and around 0.01 km2 are difficult to identify due to the geomorphologic features of small glaciers. In alpine types of landforms, they are situated in the greater slopes and corries. From the view of the InSAR Coherence study, glacier boundaries can be lost due to low coherence of adjacent area simply due to not being detected due to overlay and shadow phenomena. Glaciers equal to or smaller than 0.01 km2 can be removed during the kernel size classification, as a coherence sign with a small area might be identified as a speckle noise. In our research, the minimum size of the glacier reached 0.005 km2. The delineation of small glaciers becomes possible by identifying and separating the glacier area from the continuous low-coherence area by using high-resolution DEM and object-based image classification.
Implementation of the InSAR Coherence technique in alpine glacial landforms has shown that it can be used to identify the glacier area in the glacier terminus which is free from lateral moraines. For example, we took good values about the snout line position of Golubina, Adygine, Manas, Tuyuk, Toktogul and for most of the glaciers with an area between 0.2 and 0.5 km2. The annual update of the same study provides an alternative method for glacier dynamics observation. In automatic glacier mapping, it is necessary to consider the types of glacier and local geomorphologic characteristics, especially the presence of lateral moraines and debris cones. InSAR Coherence lets us identify glacier areas for debris-covered glaciers (Ak-Sai and Top-Karagai).
Further research should take a closer examination on the dead-ice sections identified as a glacier area on the north-eastern side of the ablation zone using high-resolution UAV imagery. Continuing on-site UAV measurements are indispensable for the delineation of the glaciers’ boundaries and surface changes.
Overall delineation of small glacier areas is a vital topic for glacier mapping. In most of the studies, authors used a minimum glacier size threshold of 0.01 km2 for coarse to medium image resolutions [49]. Our results show that in situ measurements are improving data accuracy in the determination of the Ala-Archa valley glaciers’ boundaries.

Author Contributions

Conceptualization, E.Z., M.V. and M.D.; methodology, E.Z. and Q.D.; software, E.Z. and Q.D.; validation, E.Z., M.V. and M.D.; investigation, E.Z.; resources, M.D.; writing—original draft preparation, E.Z.; writing—review and editing, M.V. and Q.D.; supervision, M.V. and M.D.; All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by DCU-IR-BE-073776 project.

Data Availability Statement

Data available from the authors upon request.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Maksimov, E.V.; Bajaeva, V.Y. Catalog of Glaciers of the USSR; Gidrometeoizdat: Leningrad, Russia, 1973; Volume 14, pp. 1–77. [Google Scholar]
  2. Maksimov, E.V.; Osmonov, A.O. Features of Modern Glaciations and Dynamics of Glaciers of the Kyrgyz Ala-Too; Ilim: Bishkek, Kyrgyzstan, 1995; pp. 41–42. [Google Scholar]
  3. Bolch, T. Glacier area and mass changes since 1964 in the Ala Archa Valley, Kyrgyz Ala-Too, northern Tien Shan. Led iSneg 2015, 129, 28–39. [Google Scholar] [CrossRef] [Green Version]
  4. Aizen, V.B.; Kuzmichenok, V.A.; Surazakov, A.B.; Aizen, E.M. Glacier Changes in the Central and Northern Tien Shan during the Last 140 Years Based on Surface and Remote-Sensing Data. Ann. Glaciol. 2006, 43, 202–213. [Google Scholar] [CrossRef] [Green Version]
  5. Aizen, V.B.; Aizen, E.M.; Kuzmichonok, V.A. Glaciers and hydrological changes in the Tien Shan: Simulation and prediction. Environ. Res. Lett. 2007, 2, 45019. [Google Scholar] [CrossRef]
  6. Azisov, E.; Hoelzle, M.; Vorogushyn, S.; Saks, T.; Usubaliev, R.; Esenamanuulu, M.; Barandun, M. Reconstructed Centennial Mass Balance Change for Golubin Glacier, Northern Tien Shan. Atmosphere 2022, 13, 954. [Google Scholar] [CrossRef]
  7. Zholdoshbekov, E.; Garg, V.; Thakur, P.K.; Duishonakunov, M.; Voda, M. Delineation of Radar Zones of Glaciers in The Ala-Archa Valley of Kyrgyz Republic. Geogr. Tech. 2021, 16, 100–110. [Google Scholar] [CrossRef]
  8. Yang, H.; Xu, G.Y.; Mao, H.X.; Wang, Y. Spatiotemporal Variation in Precipitation and Water Vapor Transport Over Central Asia in Winter and Summer Under Global Warming. Front. Earth Sci. 2020, 8, 297. [Google Scholar] [CrossRef]
  9. Atwood, D.K.; Meyer, F.J.; Arendt, A.A. Using L-band SAR coherence to delineate glacier extent. Can. J. Remote Sens. 2010, 36, 186–195. [Google Scholar] [CrossRef]
  10. Lippl, S.; Vijay, S.; Braun, M. Automatic delineation of debris-covered glaciers using InSAR coherence derived from X-, C- and L-band radar data: A case study of Yazgyl Glacier. J. Glaciol. 2018, 64, 811–821. [Google Scholar] [CrossRef] [Green Version]
  11. Bolch, T.; Kamp, U. Glacier Mapping in High Mountains Using DEMs, Landsat and ASTER Data. 2005. Available online: https://www.semanticscholar.org/paper/Glacier-mapping-in-high-mountains-using-DEMs%2C-and-Bolch-Kamp/123fcb9070bb27cce3b899bd53cbe787931bf25a (accessed on 28 January 2023).
  12. Shi, Y.; Liu, G.; Wang, X.; Liu, Q.; Zhang, R.; Jia, H. Assessing the Glacier Boundaries in the Qinghai-Tibetan Plateau of China by Multi-Temporal Coherence Estimation with Sentinel-1A InSAR. Remote Sens. 2019, 11, 392. [Google Scholar] [CrossRef] [Green Version]
  13. Moretto, S.; Bozzano, F.; Mazzanti, P. The Role of Satellite InSAR for Landslide Forecasting: Limitations and Openings. Remote Sens. 2021, 13, 3735. [Google Scholar] [CrossRef]
  14. Zhang, Y.; Liu, Y.; Jin, M.; Jing, Y.; Liu, Y.; Liu, Y.; Sun, W.; Wei, J.; Chen, Y. Monitoring Land Subsidence in Wuhan City (China) using the SBAS-InSAR Method with Radarsat-2 Imagery Data. Sensors 2019, 19, 743. [Google Scholar] [CrossRef] [Green Version]
  15. Kotlyakov, V.M.; Alekseev, V.R.; Volkov, N.V. Glyatsiologicheskiy Slovar; Kotlyakov, V.M., Ed.; Gidrometeoizdat: Leningrad, Russia, 1984; pp. 202–203. [Google Scholar]
  16. Cuffey, K.M.; Paterson, W.S. The Physics of Glaciers; Academic Press: Cambridge, MA, USA, 2010; pp. 286–307. [Google Scholar]
  17. Janowski, L.; Tylmann, K.; Trzcinska, K.; Rudowski, S.; Tegowski, J. Exploration of Glacial Landforms by Object-Based Image Analysis and Spectral Parameters of Digital Elevation Model. IEEE Trans. Geosci. Remote Sens. 2022, 60, 4502817. [Google Scholar] [CrossRef]
  18. Abermann, J.; Fischer, A.; Lambrecht, A.; Geist, T. On the potential of very high-resolution repeat DEMs in glacial and periglacial environments. Cryosphere 2010, 4, 53–65. [Google Scholar] [CrossRef] [Green Version]
  19. Shukla, A.; Arora, M.K.; Gupta., R.P. Synergistic approach for mapping debris-covered glaciers using optical–thermal remote sensing data with inputs from geomorphometric parameters. Remote Sens. Environ. 2010, 114, 1378–1387. [Google Scholar] [CrossRef]
  20. Finish Meteorological Institute. Available online: https://en.ilmatieteenlaitos.fi/press-release/1096339490URL (accessed on 24 October 2021).
  21. Umyrzakov, A.K. Atlas of the Kirgiz Soviet Socialist Republic; Natural Conditions and Resources; Publishing House GUGK: Moscow, Russia, 1987; Volume 1, p. 36. [Google Scholar]
  22. Kattel, D.B.; Mohanty, A.; Daiyrov, M.; Wang, W.; Mishra, M.; Kulenbekov, Z.; Dawadi, B. Evaluation of glacial lakes and catastrophic floods on the northern slopes of the Kyrgyz Range. Mount. Res. Dev. 2020, 40, 37–47. [Google Scholar] [CrossRef]
  23. National Statistic Committee of the Kyrgyz Republic. Demographic Yearbook of the Kyrgyz Republic. 2020. Available online: http://www.stat.kg/ru/publications/demograficheskij-ezhegodnik-kyrgyzskoj-respubliki/ (accessed on 21 January 2023).
  24. National Statistic Committee of the Kyrgyz Republic. Final Report About the Size of Cultivated Area Under the Hrvest of 2022. 2022. Available online: http://www.stat.kg/ru/publications/statisticheskij-byulleten-zaklyuchitelnyj-otchet-o-razmerah-ploshadi-poseva-selskohozyajstvennyh-kultur-v-razreze-oblastej-i-rajonov-kyrgyzskoj-respubliki-po-sostoyaniyu-na-1-iyunya-2013g/ (accessed on 21 January 2023).
  25. National Statistic Committee of the Kyrgyz Republic. The Environment in the Kyrgyz Republic: Statistical Compilation 2017–2021. 2022. Available online: http://www.stat.kg/media/publicationarchive/c210d76d-91e9-4e8e-a597-e49217759846.pdf (accessed on 21 January 2023).
  26. Dyurgerov, M.B.; Liu, C.; Xie, Z. Oledenenie Tian Shania (Tien Shan Glaciation); VINITI: Moscow, Russia, 1995; p. 57. [Google Scholar]
  27. Aizen, V. Glaciation and Its Evolution at Northern Tien Shan’s Peripheral Ranges. Ph.D. Thesis, Russian Academy of Science, Moscow, Russia, 1988. [Google Scholar]
  28. Fyffe, C.L.; Brock, B.W.; Kirkbride, M.P.; Mair, D.W.F.; Arnold, N.S.; Smiraglia, C.; Diolaiuti, G.; Diotri, F. Do debris-covered glaciers demonstrate distinctive hydrological behaviour compared to clean glaciers? J. Hydrol. 2019, 570, 584–597. [Google Scholar] [CrossRef]
  29. Ruiz, L.; Berthier, E.; Masiokas, M.; Pitte, P.; Villalba, R. First Surface Velocity Maps for Glaciers of Monte Tronador, North Patagonian Andes, Derived from Sequential Pléiades Satellite Images. J. Glaciol. 2015, 61, 908–922. [Google Scholar] [CrossRef] [Green Version]
  30. Guan, W.; Cao, B.; Pan, B.; Chen, R.; Shi, M.; Li, K.; Zhao, X.; Sun, X. Updated surge-type glacier inventory in the West Kunlun Mountains, Tibetan Plateau, and implications for glacier change. J. Geophys. Res. Earth Surf. 2022, 127, e2021JF006369. [Google Scholar] [CrossRef]
  31. NASA’s Alaska Satellite Facility (ASF) DAAC. Available online: https://search.asf.alaska.edu/ (accessed on 15 February 2022).
  32. The pyroSAR Developers. Available online: https://pyrosar.readthedocs.io/en/latest/general/OSV.html (accessed on 1 February 2023).
  33. Chandler, B.M.P.; Evans, D.J.A. Glacial Processes and Sediments. In Encyclopedia of Geology, 2nd ed.; Alderton, D., Elias, S., Eds.; Academic Press: Oxford, UK, 2021; pp. 830–856. [Google Scholar]
  34. Rocca, F.; Prati, C.; Guarnieri, A.M.; Ferretti, A. Sar Interferometry and its Applications. Surv. Geophys. 2000, 21, 159–176. [Google Scholar] [CrossRef]
  35. Tzouvaras, M.; Danezis, C.; Hadjimitsis, D.G. Small Scale Landslide Detection Using Sentinel-1 Interferometric SAR Coherence. Remote Sens. 2020, 12, 1560. [Google Scholar] [CrossRef]
  36. Closson, D.; Milisavlievic, N. InSAR Coherence and Intensity Changes Detection. In Mine Action–The Research Experience of the Royal Military Academy of Belgium; InTech: London, UK, 2017; Volume 23, pp. 156–176. [Google Scholar] [CrossRef] [Green Version]
  37. Woodhouse, I.H. Introduction to Microwave Remote Sensing; CRC Press: Boca Raton, FL, USA, 2006; pp. 36–37. [Google Scholar]
  38. Touzi, R.; Lopez, A.; Bruniquel, J.; Vachon, P.W. Coherence estimation for SAR imagery. IEEE Trans. Geosci. Remote. Sens. 1999, 37, 135–149. [Google Scholar] [CrossRef] [Green Version]
  39. Wang, Z.; Wang, Y.; Wang, B.; Xiang, M.; Wang, R.; Xu, W.; Song, C. Multi-Frequency Interferometric Coherence Characteristics Analysis of Typical Objects for Coherent Change Detection. Remote Sens. 2022, 14, 1689. [Google Scholar] [CrossRef]
  40. Friedl, B.; Hölbling, D. Using SAR Interferograms and Coherence Images for Object-Based Delineation of Unstable Slopes. In Proceedings of the FRINGE 2015 Workshop: Advances in the Science and Applications of SAR Interferometry and Sentinel-1InSAR Workshop, Frascati, Italy, 23–27 March 2015; European Space Agency: Frascati, Italy, 2015. [Google Scholar] [CrossRef] [Green Version]
  41. Shukla, A.; Gupta, R.P.; Arora, M.K. Delineation of debris-covered glacier boundaries using optical and thermal remote sensing data. Remote Sens. Lett. 2010, 1, 11–17. [Google Scholar] [CrossRef]
  42. USGS EarthExplorer. Available online: https://earthexplorer.usgs.gov/ (accessed on 20 February 2022).
  43. Voda, M.; Montes, Y.S. Descending mountain routes future: The North Yungas and Fagaras Geosystem’s comparative study. Geogr. Tech. 2018, 13, 152–166. [Google Scholar] [CrossRef]
  44. Rahaman, K.R.; Hassan, Q.K.; Ahmed, M.R. Pan-Sharpening of Landsat-8 Images and Its Application in Calculating Vegetation Greenness and Canopy Water Contents. ISPRS Int. J. Geo-Inf. 2017, 6, 168. [Google Scholar] [CrossRef] [Green Version]
  45. Gupta, S.K.; Negru, R.; Voda, M. The Indian Himalaya’s unique attributes: Hemkund Sahib and The Valley of Flowers. Geogr. Tech. 2018, 13, 62–73. [Google Scholar] [CrossRef]
  46. Zhang, J. Multi-source remote sensing data fusion: Status and trends. Int. J. Image Data Fusion 2010, 1, 5–24. [Google Scholar] [CrossRef] [Green Version]
  47. Ganyushkin, D.; Chistyakov, K.; Derkach, E.; Bantcev, D.; Kunaeva, E.; Terekhov, A.; Rasputina, V. Glacier Recession in the Altai Mountains after the LIA Maximum. Remote Sens. 2022, 14, 1508. [Google Scholar] [CrossRef]
  48. Andreassen, L.M.; Nordli, Ø.; Rasmussen, A.; Melvold, K.; Nordli, Ø. Langfjordjøkelen, a Rapidly Shrinking Glacier in Northern Norway. J. Glaciol. 2012, 58, 581–593. [Google Scholar] [CrossRef] [Green Version]
  49. Leigh, J.R.; Stokes, C.R.; Carr, R.J.; Evans, I.S.; Andreassen, L.M.; Evans, D.J.A. Identifying and Mapping Very Small (<0.5 km2) Mountain Glaciers on Coarse to High-Resolution Imagery. J. Glaciol. 2019, 65, 873–888. [Google Scholar] [CrossRef] [Green Version]
  50. Fischer, M.; Huss, M.; Barboux, C.; Hoelzle, M. The New Swiss Glacier Inventory SGI2010: Relevance of Using High-Resolution Source Data in Areas Dominated by Very Small Glaciers. Arct. Antarct. Alp. Res. 2014, 46, 933–945. [Google Scholar] [CrossRef] [Green Version]
  51. López, P.; Chevallier, P.; Favier, V.; Pouyaud, B.; Ordenes, F.; Oerlemans, J. A regional view of fluctuations in glacier length in southern South America. Glob. Planet. Change 2010, 71, 85–108. [Google Scholar] [CrossRef]
  52. Machguth, H.; Huss, M. The length of the world’s glaciers–a new approach for the global calculation of center lines. Cryosphere 2014, 8, 1741–1755. [Google Scholar] [CrossRef] [Green Version]
  53. Narama, C.; Kääb, A.; Duishonakunov, M.; Abdrakhmatov, K. Spatial variability of recent glacier area changes in the Tien Shan Mountains, Central Asia, using Corona (~1970), Landsat (~2000), and ALOS (~2007) satellite data. Glob. Planet. Change 2010, 71, 42–54. [Google Scholar] [CrossRef]
  54. Aizen, V.; Aizen, E. Hydrological cycles on the north and south peripheries of mountain glacial basins of central Asia. Hydrol. Process. 1997, 11, 451–469. [Google Scholar] [CrossRef]
Figure 1. (A): The location of Ak-Sai glacier in the Kyrgyz Ala-Too range and the extent of the UAV survey area (red line); (B): geographic location of research area; (C): the map of Ak-Sai glacier; (D): field photos of Ak-Sai glacier, taken in February 2022. The physical and background shade maps are composed from ALOS PALSAR DEM and FCC images of Ak-Sai glacier from Landsat 8, acquired on 10 August 2021.
Figure 1. (A): The location of Ak-Sai glacier in the Kyrgyz Ala-Too range and the extent of the UAV survey area (red line); (B): geographic location of research area; (C): the map of Ak-Sai glacier; (D): field photos of Ak-Sai glacier, taken in February 2022. The physical and background shade maps are composed from ALOS PALSAR DEM and FCC images of Ak-Sai glacier from Landsat 8, acquired on 10 August 2021.
Remotesensing 15 01131 g001
Figure 2. The input methodological flowchart of: (A): the automatic delineation of glacier area by using Sentinel-1 SAR dataset and InSAR Coherence tool; (B): the creation of orthogonal images by using high resolution UAV images and (C): manual glacier feature detection using pan-sharpened Landsat 8 imagery and ALOS PALSAR DEM.
Figure 2. The input methodological flowchart of: (A): the automatic delineation of glacier area by using Sentinel-1 SAR dataset and InSAR Coherence tool; (B): the creation of orthogonal images by using high resolution UAV images and (C): manual glacier feature detection using pan-sharpened Landsat 8 imagery and ALOS PALSAR DEM.
Remotesensing 15 01131 g002
Figure 3. Processing chain for obtaining glacier contours using InSAR Coherence technique, where in the picture (A) terrain-corrected InSAR Coherence image is shown, (B): overlay and shadow map of Sentinel-1 SAR image in research area, (C): coherence thresholding map, (D): kernel size classification, (E): overlay and shadow masking, (F): vectorization and accuracy assessment.
Figure 3. Processing chain for obtaining glacier contours using InSAR Coherence technique, where in the picture (A) terrain-corrected InSAR Coherence image is shown, (B): overlay and shadow map of Sentinel-1 SAR image in research area, (C): coherence thresholding map, (D): kernel size classification, (E): overlay and shadow masking, (F): vectorization and accuracy assessment.
Remotesensing 15 01131 g003
Figure 4. The result of the manual mapping of the glacier area using orthogonal mosaic imagery from UAV highlighting the main geomorphologic features that were used to determine the glacier boundary.
Figure 4. The result of the manual mapping of the glacier area using orthogonal mosaic imagery from UAV highlighting the main geomorphologic features that were used to determine the glacier boundary.
Remotesensing 15 01131 g004
Figure 5. Comparison the results of automatic delineation of Ak-Sai glacier with reference.
Figure 5. Comparison the results of automatic delineation of Ak-Sai glacier with reference.
Remotesensing 15 01131 g005
Figure 6. Glaciers’ hypsometry for the Ala-Archa valley (A); the aspect distribution of the glacier area and number (mean aspect of each glacier) (B).
Figure 6. Glaciers’ hypsometry for the Ala-Archa valley (A); the aspect distribution of the glacier area and number (mean aspect of each glacier) (B).
Remotesensing 15 01131 g006
Figure 7. Location of glaciers in Ala-Archa valley. As the hill shade in the background, we used ALOS PALSAR DEM.
Figure 7. Location of glaciers in Ala-Archa valley. As the hill shade in the background, we used ALOS PALSAR DEM.
Remotesensing 15 01131 g007
Table 1. List of C-Band Sentinel-1 datasets used for glacier coherence estimation.
Table 1. List of C-Band Sentinel-1 datasets used for glacier coherence estimation.
MissionProduct TypeBeam ModeOrbitOrbit TypePolarizationAcquisition Date
S1ASLCIW39299AscendingVV19.08.2021
S1ASLCIW39474AscendingVV31.08.2021
Note: “S1A” represents Sentinel-1, “A”—mission, “SLC”—Single-Look Complex product type, “IW”—Interferometric Wide Swath beam mode, “VV”—dual vertical polarization.
Table 2. Glacier area with respect to the size classes.
Table 2. Glacier area with respect to the size classes.
Size Class
(km2)
NumberTotal AreaMean Area (km2)
(n)(%)(km2)(%)
<0.13358.91.063.40.03
0.1–0.258.90.662.10.13
0.2–0.5610.71.976.30.33
0.5–1.023.61.223.90.61
1.0–2.058.97.0222.31.40
2.0–5.047.114.3845.73.40
>5.011.85.1416.35.14
Total5610031.451000.49
Table 3. Morphological types of glaciers with respect to the size classes.
Table 3. Morphological types of glaciers with respect to the size classes.
Glacier TypesGlacier NumberGlacier Area
Count%km2%
Hanging2137.50.521.65
Corrie1526.81.243.94
Valley814.319.9263.33
Corrie–valley47.11.133.59
Slope glacier35.60.571.81
Hanging–valley23.60.963.05
Kettle glacier23.64.0812.97
Asymmetrical–valley11.83.059.70
All5610031.45100
Table 4. Database table of glacier inventory of Ala-Archa valley. Glacier length * between lowest and highest points, ** by longest glacier extension.
Table 4. Database table of glacier inventory of Ala-Archa valley. Glacier length * between lowest and highest points, ** by longest glacier extension.
IDGlacier Numeration (ID Code)Glacier NameCentroid Coordinates (WGS 1984 Datum)River CatchmentGlacier TypeMean AspectArea, km2Length, mAltitude of Lowest Point, m.a.s.l.Altitude of Highest Point, m.a.s.l.
1228/1 42°34′22.8″N 74°24′57.6″EKashkasuuHangingN0.00711739304000
2229Adyginetor42°32′02.4″N 74°24′57.6″EAdyginetorCorrie–valleyNE0.25770338004340
3229-1 42°32′16.8″N 74°24′46.8″EAdyginetorHangingNE0.01418539954090
4229-2 42°32′34.8″N 74°24′50.4″EAdyginetorHangingN0.00615439003950
5230 42°31′55.2″N 74°25′19.2″EAdyginetorHangingN0.01013039304015
6231 42°31′51.6″N 74°25′51.6″EAdyginetorCorrieN0.07835037654010
7232 42°31′22.8″N 74°25′04.8″EAdygineCorrie–valleyN0.499123037704260
8233 42°30′46.8″N 74°25′04.8″EAdygineCorrieNE0.01919039204020
9234 42°30′21.6″N 74°25′26.4″EAdygineCorrieN0.28990036754035
10235Adygene42°29′56.4″N 74°25′58.8″EAdygineKettle glacierN2.846282036004150
11235-1 42°30′10.8″N 74°26′49.2″EAdygineHangingNW0.01729037503950
12235/1 42°29′24.0″N 74°26′24.0″EAla-ArchaCorrieNE0.00816737703890
13235/2 42°29′24.0″N 74°26′02.4″EAla-ArchaCorrieSE0.01230038903975
14236 42°28′30.0″N 74°25′44.4″EAla-ArchaCorrieN0.13358036803820
15237 42°28′15.6″N 74°25′26.4″EAla-ArchaCorrieN0.14868037754000
16238 42°28′15.6″N 74°25′58.8″EAla-ArchaHangingN0.04112537303815
17239 42°27′39.6″N 74°25′58.8″EAla-ArchaHanging–valleyNW0.701155035003945
18240 42°27′21.6″N 74°25′48.0″EAla-ArchaCorrieE0.09351037303915
19241Mal AA Zap42°26′42.0″N 74°25′33.6″EAla-ArchaKettle glacierNE1.231160035903900
20241-1 42°27′07.2″N 74°25′22.8″EAla-ArchaSlope glacierNE0.01522037103755
21241-2 42°27′18.0″N 74°25′19.2″EAla-ArchaCorrieSE0.06530037803860
22242Mal AA Vost42°26′13.2″N 74°26′13.2″EAla-ArchaValleyN0.516194035104040
23243Bol AA Zap42°25′48.0″N 74°27′07.2″EAla-ArchaValleyN1.413213034953950
24244Toktogul42°25′55.2″N 74°28′01.2″EAla-ArchaValleyNW1.200257034704115
25245Manas42°26′31.2″N 74°28′40.8″EAla-ArchaValleyNW1.377288034904250
26246 42°26′56.4″N 74°28′19.2″EAla-ArchaCorrieW0.08757039404050
27246-1 42°26′49.2″N 74°28′04.8″EAla-ArchaCorrieW0.02624038003940
28247 42°27′10.8″N 74°28′12.0″EAla-ArchaCorrieNW0.08064036404010
29247-1 42°27′07.2″N 74°28′30.0″EAla-ArchaHangingNW0.01918039804160
30248 42°27′21.6″N 74°28′40.8″EAla-ArchaCorrie–valleyNW0.20690035654000
31249 42°27′36.0″N 74°29′06.0″EAla-ArchaCorrieN0.11868037204010
32250Golubina42°27′10.8″N 74°29′52.8″EJindi-SuuValleyN5.138466033404420
33250-1 42°27′36.0″N 74°30′18.0″EJindi-SuuCorrie–valleyNW0.16386036204060
34251Tuyuk42°27′36.0″N 74°31′48.0″ETuyuk-SuuValleyNW4.1272810 * (3510 **)33904180
35251-1 42°28′26.4″N 74°32′27.6″ETuyuk-SuuSlope glacierS0.490124040204440
36251-2 42°27′43.2″N 74°32′42.0″ETuyuk-SuuSlope glacierSW0.06031041104230
37251-3 42°28′12.0″N 74°30′14.4″ETuyuk-SuuHangingNE0.01635038004100
38251-4 42°28′12.0″N 74°30′10.8″ETuyuk-SuuHangingN0.01129038104040
39252 42°28′12.0″N 74°30′37.2″ETuyuk-SuuCorrieN0.0353036453835
40253Top-Karagai42°29′06.0″N 74°32′52.8″ETop-KaragaiAsymmetrical–valleyN3.052471037004500
41253-1 42°30′07.2″N 74°31′37.2″ETop-KaragaiCorrieSW0.05235040354300
42253-2 42°30′14.4″N 74°31′51.6″ETop-KaragaiHangingNW0.11956042454530
43254Teketor42°30′57.6″N 74°31′08.4″ETeketorHanging–valleyN0.250109036904410
44254-1 42°31′26.4″N 74°31′19.2″ETeketorHangingN0.00613040904200
45255Ak-Sai42°30′36.0″N 74°32′42.0″EAk-SaiValleyNW4.357651532604565
46255-1 42°31′40.8″N 74°31′26.4″EAk-SaiHangingNE0.02231536603935
47255-2 42°31′22.8″N 74°31′26.4″EAk-SaiHangingNE0.03226041204230
48256Uchitel’42°31′12.0″N 74°33′25.2″EAk-SaiValleyNW1.796456536004720
49256-1 42°31′48.0″N 74°33′10.8″EAk-SaiHangingNW0.01115042704380
50256-2 42°31′48.0″N 74°33′18.0″EAk-SaiHangingNW0.00612043404420
51256-3 42°32′02.4″N 74°33′07.2″EAk-SaiHanging NW0.01523041604350
52256/1 42°32′16.8″N 74°33′00.0″EAk-SaiHangingSW0.01123742704410
53256/2 42°32′27.6″N 74°32′45.6″EAk-SaiHangingNW0.00511340904161
54256/3 42°33′10.8″N 74°32′52.8″ESharkyratmaHangingNW0.08030439404164
55256/4Murat42°33′36.0″N 74°32′56.4″EMurat-SaiHangingNW0.02720041204196
56258Kashka-Suu42°34′25.0″N 74°32′49.8″EKashka-SuuHangingN0.04030838054005
Table 5. Glacier area changes for the period 1964–2021, previous glacier area assessment based on the research of Bolch [3] and Aizen [4].
Table 5. Glacier area changes for the period 1964–2021, previous glacier area assessment based on the research of Bolch [3] and Aizen [4].
GlacierGlacier Area, km2
196419992021
Uchitel2.22.01.796
Adygine3.53.02.846
Top-Karagai3.73.33.052
Tuyuk5.25.04.127
Ak-Sai4.84.54.357
Golubina5.65.425.138
All39.234.531.45
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zholdoshbekov, E.; Duishonakunov, M.; Du, Q.; Voda, M. Assessing Glacier Boundaries in the Ala-Archa Valley of Kyrgyzstan by Using Sentinel-1 SAR Dataset and High-Resolution UAV Imagery. Remote Sens. 2023, 15, 1131. https://doi.org/10.3390/rs15041131

AMA Style

Zholdoshbekov E, Duishonakunov M, Du Q, Voda M. Assessing Glacier Boundaries in the Ala-Archa Valley of Kyrgyzstan by Using Sentinel-1 SAR Dataset and High-Resolution UAV Imagery. Remote Sensing. 2023; 15(4):1131. https://doi.org/10.3390/rs15041131

Chicago/Turabian Style

Zholdoshbekov, Emilbek, Murataly Duishonakunov, Qingyun Du, and Mihai Voda. 2023. "Assessing Glacier Boundaries in the Ala-Archa Valley of Kyrgyzstan by Using Sentinel-1 SAR Dataset and High-Resolution UAV Imagery" Remote Sensing 15, no. 4: 1131. https://doi.org/10.3390/rs15041131

APA Style

Zholdoshbekov, E., Duishonakunov, M., Du, Q., & Voda, M. (2023). Assessing Glacier Boundaries in the Ala-Archa Valley of Kyrgyzstan by Using Sentinel-1 SAR Dataset and High-Resolution UAV Imagery. Remote Sensing, 15(4), 1131. https://doi.org/10.3390/rs15041131

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop