In many parts of the world, glaciers are the main source of water supply for hundreds of millions of people, a source of electricity generation or an element responsible for the local structure of flora and fauna. It is recognized that the dynamics of the mass balance of glaciers are an early response to the currently observed climate change, associated with long-term industrial and post-industrial activities. Due to the location of glaciers and difficult direct access to them, research expeditions are exposed to high costs associated with ensuring safety and logistics, as well as supplying them with equipment necessary to conduct research. The example of the COVID-19 pandemic drew attention to additional, previously unforeseen difficulties with the organization of research expeditions. In relation to the above, remote methods of obtaining data on glaciers, including widely used remote sensing methods, are becoming extremely important. The polar areas that accumulate most of the ice on the Earth are observed with selected satellites much more often than areas located in moderate and low latitudes. This is due to the overlapping of successive acquisition paths towards the poles. The shortened time interval between acquisitions creates potentially high possibilities of using this data.
Landsat and Sentinel satellites are optical sensors. They acquire images passively, which means that they record solar radiation reflected and re-emitted by the Earth’s surface. One of the greatest limitation in the use of this type of sensors is the cloud cover, which obscures the area and prevents further analysis. In the most cases, working with images that contain cloud cover is impossible or very difficult. Cloud cover also limits the possibilities of correct georeferencing of images, which may be an additional factor reducing their suitability for use. The main aim of the work is to evaluate the usefulness of Landsat 8 images in monitoring the frontal zone of the Aavatsmarkbreen by analyzing the cloud cover on the imagery covering its area.
The Aavatsmarkbreen is a tidewater glacier located in the Kaffiøyra region, in the north-west part of Spitsbergen (Svalbard). On the basis of modern research capabilities it has an area of 73 km2 and terminates in the Hornbæk Bay with about 4 km wide and 40 m high ice cliff (Lankauf 2002, Sobota 2021). Since the end of the Little Ice Age, area of the frontal part of Aavatsmarkbreen decreased by about 72%. The Area Of Interest extent was presented on the Figure 4. It covers the terminus zone of the Aavatsmarkbreen and was defined using a 300 m buffer (i.e. 10 Landsat 8 OLI pixels) to the west, north and south of the maximum extent of tongue observed in 2013-2020 period (late surge phase, 9th July 2015) and 1000 m to the east (up-glacier direction) of minimum extent in the same period (1st October 2020). AOI covers area of 12.45 km2 , i.e. 13,838 of 30 m Landsat 8 OLI pixels.
Dataset used in the current work consists of all of the Landsat 8 scenes available for download at USGS service that meet following criteria: (A) were acquired from the beginning of the mission (2013) to the end of 2020, and (B) cover the entire area of interest (AOI). Such specified dataset consists of 868 images. Detailed characteristic of it, concerning time and geometry (WRS-2 path and row) of acquisition is given in table 1 and table 2. AOI visibility on each image was calculated using Quality Assessment Band (QA) which constitute an integral part of the Landsat 8 dataset. QA consists of several pixels with values containing information about their content and thus also possible cloudiness. Pixels which correspond to a high concentration of ice or snow and “clean terrain” were used. For each image, a reclassification was performed, in which pixels with the values 2720, 2724, 2728, 2732 appropriate for the “clean terrain” attribute and 3744, 3748, 3752, 3756 corresponding to a high concentration of snow and ice with a low probability of cloudiness were considered as visible terrain. Details of pixels values was presented in table 4. The ratio of the number of such pixels to all pixels overlapping the AOI was then calculated. For each image, the percentage of visible terrain was obtained. These values are grouped into AOI visibility classes.
Of all the satellite scenes available for download, only 176 (approx. 20,0%) contained fully visible terrain, and therefore suitable for further use. As many as 59.1% of the images were covered with clouds in over 95%. The largest amount of satellite data was recorded in 2018-2020, which was the result of the introduction of the ascending scenes of the Landsat 8 program. Similarly, the distinction of the total visibility class largely contributed to the years 2018-2020, but the annual percentage value of the share of the Fully visible class in relation to the total number of imagery in the analysed years did not show major deviations. In the monthly division, the largest number of images was recorded in the summer, while the value of the share of completely visible scenes at that time was the lowest (it did not exceed 15% of all scenes). The dominant class were scenes completely covered with clouds, whose variability in individual months ranged from just over 42% to more than 70%. In the annual distribution of useful imagery, the most stable situation occurred in 2018-2020 (Fig. 8). Fully useful scenes covered mainly the spring period (March, April and May), which determines the cyclone activity in Svalbard (Fig. 7). At almost equal intervals, scenes of the Fully visible class were recorded. Based on the basic knowledge of Svalbard’s climate, the results obtained were considered reliable, but the usefulness status of satellite imagery was found to be unsatisfactory. A valuable addition to the presented study was the comparison of the results of the visibility of the area on the analysed images to the course of meteorological conditions with the weather station located in Ny-Ålesund, located about 30 km from the AOI. The share of all imagery was compared with the daily and monthly cloudiness recorded at the Station.
In conclusion, the work focused on the analysis of satellite images, for which cloudiness is the main factor limiting the potential possibilities of their use in the described glaciological studies. It is also worth remembering that one area was included in the work – the frontal zone of the Aavatsmarkbreen. The study omitted glaciers from other areas of the Earth where climatic conditions could have influenced different results.
Marcin Nowak [firstname.lastname@example.org], Uniwersytet Mikołaja Kopernika w Toruniu, Centrum Badań Polarnych
Kamil Czarnecki [email@example.com], Uniwersytet Mikołaja Kopernika w Toruniu, Centrum Badań Polarnych
APA: Nowak, M., & Czarnecki, K. (2023). Analiza zachmurzenia na zobrazowaniach Landsat 8 w latach 2013‑2020 jako ocena stopnia ich przydatności w monitoringu arktycznych lodowców. Przegląd Geograficzny, 95(2), 127-147. https://doi.org/10.7163/PrzG.2023.2.1
MLA: Nowak, Marcin, and Czarnecki, Kamil. "Analiza zachmurzenia na zobrazowaniach Landsat 8 w latach 2013‑2020 jako ocena stopnia ich przydatności w monitoringu arktycznych lodowców". Przegląd Geograficzny, vol. 95, no. 2, 2023, pp. 127-147. https://doi.org/10.7163/PrzG.2023.2.1
Chicago: Nowak, Marcin, and Czarnecki, Kamil. "Analiza zachmurzenia na zobrazowaniach Landsat 8 w latach 2013‑2020 jako ocena stopnia ich przydatności w monitoringu arktycznych lodowców". Przegląd Geograficzny 95, no. 2 (2023): 127-147. https://doi.org/10.7163/PrzG.2023.2.1
Harvard: Nowak, M., & Czarnecki, K. 2023. "Analiza zachmurzenia na zobrazowaniach Landsat 8 w latach 2013‑2020 jako ocena stopnia ich przydatności w monitoringu arktycznych lodowców". Przegląd Geograficzny, vol. 95, no. 2, pp. 127-147. https://doi.org/10.7163/PrzG.2023.2.1