The recent launch of Sentinel-2 satellites with their 10 m spatial resolution sensors and high revisit times has opened up new possibilities in cryospheric research, especially with respect to short term monitoring of glacier changes and the production of glacier inventories. Italy has a long history in this field, with the first inventory dating back to the 1950s, and a recent one produced from aerial orthophotos in 2015, including data acquired between 2005 and 2011. To detect recent glacier changes, we created a new glacier inventory from 15 Sentinel-2 scenes plus 2 Landsat 8 OLI scenes, all acquired between 2015 and 2017. Glacier classification was based on a semi-automatic approach, i.e. the band ratio of a red (panchromatic for Landsat 8) and shortwave infrared channel, with an additional threshold on the blue band to avoid misclassification of ice under shadows. Glacier outlines extracted from such an approach were then manually edited using false colour composites to correct debris covered glacier areas and other misclassifications, e.g. seasonal snow. Thus, we were able to describe 881 glaciers with an area above 0.1 km² using three years of data, compared to a higher number of years required in previous inventories, providing a useful tool for the assessment of glacier changes and modelling purposes. The largest glacier in the inventory is Adamello/Mandrone in the Adamello range with its 14.36 km², followed by the debris covered Miage glacier in the Mont Blanc group, while only small glacierets exist in the Maritime Alps and Canin/Montasio ranges at the western and eastern ends of the Italian Alps. Forni Glacier in the Ortles-Cevedale Group, once the second largest glacier in Italy, is now split in two separate bodies. During the observation period, Italian glaciers covered a total of 349 km², which implies a 5 % glacier recession in comparison with the former Italian glacier inventory. We found however no significant change in the number of glaciers, owing to the opposite effects of glacier extinction and fragmentation. We then specifically investigated changes related to individual glaciers, mountain ranges within the Italian Alps and Italian administrative regions, to provide a detailed comparison against the former inventory. The continuous acquisition of Sentinel-2 data will ensure the inventory is constantly updated and glacier recession rates can be determined in the years to come.

We present and discuss results of ground-based measurements and remote assessments of glacier albedo in the Ak Shyirak massif (Inner Tian Shan). Tendencies of decreasing albedo are typical for many glacierized mountain regions and lead to accelerating glacier downwasting. In the Ak-Shyirak massif the situation is complicated by technogenic dust pollution, whose contribution to albedo changes was unknown. Ground-based albedo measurements were carried out across 11 glaciers during summer and autumn in 2015-2018 using Sp Lite 2 Kipp&Zonen albedometer. Using ground albedo data from 2 glaciers and satellite imagery obtained on the same days, we developed models for ground correction of albedo values obtained from satellite data. A time series of images from Landsat and Sentinel 2 satellites for 1991-2018 were analysed to assess glacier albedo in the Ak Shyirak massif in different seasonal weather conditions. We compiled a series of albedo maps for the study area. In contrast to some other mountain regions we do not observe a decrease of glacier albedo in the Ak Shyirak massif since the 1990s. Glacier albedo in the Ak Shyirak massif is driven by natural environmental factors, mostly by elevation and preceding weather conditions. Technogenic impact (mechanical disturbance of glaciers within the Kumtor mine) is detected only locally, while technogenic dust on the background of natural dust is not traceable on glaciers with satellite imagery across the Ak-Shyirak massif.

The recent increase of medium resolution satellite programs that consistently senses the Earth surface has brought great potential for glaciology. Satellites like the Sentinel-1&2 tandems, Landsat 7&8 duo and Rapideye constellation sense mountain ranges almost without missing a day. Consequently, high resolution time-series can be constructed, for snow-cover or crack propagation, or other single scene information. Properties derived from combinations of acquisitions, such as, glacier velocity is more challenging. Velocity estimation is based upon image matching, which can be hindered by coherence loss or cloud cover. Consequently, generated velocity data have gaps and have scattered coverage at different time intervals.

This steady data stream has resulted in several initiatives to use image matching techniques in batch processing mode. Resulting in glacier velocity services being set-up (e.g.: cryoportal, GoLIVE, Geodetic data portal), but these initiatives are satellite specific. If such products are merged it is possible to increase temporal and spatial coverage and increase the reliability of the velocity estimates. We developed such a workflow that merges glacier velocity products of different time spans and type. This combined dataset covers several mountain ranges and has a regular time step at a monthly pace. This consistent time-series make it possible to observe glacier dynamics that are of short nature. As the satellite recordings already span several years, the dynamics of tidewater glaciers or (partial) surges can be observed and analyzed. This study therefor shows the great potential of bringing together different satellite systems.

Keywords: Remote Sensing Big Data, Sentinel-1, Google Earth Engine, Software, Global glacier surface velocity monitoring service

Abstract

Glaciers are a natural global resource and one of the principal climate change indicator at global and local scale, being influenced by temperature and snow precipitation changes. Among the parameters used for glacier monitoring, the glaciers surface velocity is an important element, since it influences the events connected to glaciers changes (mass balance, hydro balance, glaciers stability, landscape erosion). The surface glacier velocity can be measured using remote sensing techniques with repetition of optical satellite imagery and SAR data, using speckle and/or feature tracking methods, that allow potentially to perform a continuous monitoring over wide areas. Optical satellite data presents some drawbacks over SAR images: those images are dependent from illumination and generally affected by weather conditions, and so they are unable to capture a clear view through cloud coverage, such as the glacial regions are often.

Concerning optical data, the NASA’s Global Land Ice Velocity Extraction project (GoLIVE) today is the unique project providing near-real-time global land ice velocity dataset from Landsat 8 OLI panchromatic images (band 8, 15 m pixel resolution) at global scale. The final aim of this project is to implement a near real-time archive of these data at the National Snow and Ice Data Center (NSIDC) (Fahnestock et al., 2015).

Thanks to Copernicus program, Sentinel-1 SAR imagery is available under a free access policy with high temporal (6-days revisit time) and spatial resolution (up to 5 m). Due to the huge amount of data available, it is necessary to change the processing way from the standard procedure bring-data-to-users to the opposite one bring-users-to-data, moving towards the Big Data paradigm. The leading idea of this work is to continuously retrieve glaciers surface velocity using Sentinel-1 SAR amplitude data and exploiting the potentialities of the Google Earth Engine (GEE) platform. GEE archive includes more than thirty years of historical imagery and scientific datasets (including Sentinels imagery), daily updated and expanded. Users can directly benefit the GEE facilities directly uploading their own algorithms on this dedicated infrastructure, removing the required time for data transfer and allowing the development of innovative applications.

The algorithm for glacier surface velocity monitoring designed at the Geodesy and Geomatics Division of the University of Rome “La Sapienza” is based on off-set tracking applied to SAR images, using the amplitude information; the developed software has been integrated in the GEE platform, and it can automatically process large stacks of SAR imagery to retrieve glacier surface velocity field time series. Several thousands of Sentinel-1 images were up to now processed, with regards to relevant mountain glaciers (i.e. Baltoro (Karakoram), San Rafael and San Quintin (Chile), Aletsch (Switzerland)), and some of the obtained results were also successfully validated with respect to already available and renown laptop software dedicated to glacier surface velocity retrieval (i.e. ESA SNAP, CIAS).

The aim of this work was to carry out a comparison test over some huge glaciers (San Quintin, San Rafael, Baltoro, etc.) between Sentinel-1 SAR and GoLIVE optical velocities in order to further verify the accuracy of our methodology. GoLIVE project provides velocity fields generated using a correlation algorithm from pairs of Landsat 8 panchromatic images. This dataset includes many scenes with significant cloud cover, resulting in the low availability of data that made difficult a complete comparison. This underlines the great advantages of SAR applications with respect to optical data for glaciers remote sensing, SAR imagery can fill in the lack of data of optical products in all these unfortunate cases by providing near real-time monitoring capabilities across the globe and a better understanding of complex phenomenon.

Overall these results clearly highlight the unprecedented potential of the Big Data analysis to automatically monitor glacier surface velocity at global scale, coming from the exploitation of the synergy between GEE and Sentinel-1 imagery, and pave the way to a global glacier surface velocity monitoring service.

Dynamic processes in ice can be difficult to understand when observations are limited to fieldwork opportunities or cloud-free summer months. The recent availability of high-resolution satellite Synthetic Aperture Radar (SAR) data at an unprecedented repeat observation rate provides significantly improved opportunities for investigating time-varying processes in polar regions. Here, using time-series of data at approximately weekly intervals over five years from the TerraSAR-X and Sentinel-1 satellites, we explore calving and surging at several glaciers in Svalbard. We derive velocity maps using speckle and feature tracking and manually digitise evolving glacier ice-fronts to determine advance/retreat rates and calving styles.

We analyse data from at least five surge-type glaciers (Tunabreen, Morsnevbreen, Aavatsmarkbreen, Monacobreen and Negribreen) in the active surge phase and one consistently fast-flowing glacier (Kronebreen). Each glacier is typically captured in more than 150 TerraSAR-X images (2013-2015) and an equal number from Sentinel-1 (2015-2018) yielding hundreds of image pairs suitable for feature tracking. We apply standard techniques to derive velocity maps at spatial sampling rates of 40m for TerraSAR-X and 100m for Sentinel-1. Overlap between these two missions demonstrates that the reduction in spatial resolution between TerraSAR-X (~2m) and Sentinel-1 (~10m) does not hamper the retrieval of valuable ice dynamic information.

As well as providing a means to rapidly detect several new glacier surges over the observation period, the improved spatial and temporal detail in these observations allows significantly better understanding of processes. Glacier retreat rates are shown to vary substantially from year to year and to not be much influenced by ice flow rates. We use these data to explore the enthalpy budget as an explanation of surge-type glacier activity, and to further investigate controls on calving rates. Velocity maps and time-series animations reveal a variety of behaviours previously hidden from view and allow new insights into glacier dynamic behaviour.

The Pamir Mountains are one of the regions where many surge-type glaciers exist. However, so far this region has been studied less intense than the neighbouring Karakoram where a complete inventory of surge-type glaciers exists (e.g. Bhambri et al. 2017). In particular, time-series of satellite data revealed unprecedented details about the temporal evolution of glacier surges and helped identifying them. The related studies revealed a high variability of surge types (e.g. regarding surge duration and frequency, advance rates, flow velocities) and provided new insights on the governing processes. However, the reasons for a glacier of being surge-type or not are still not well understood and complete baseline data are missing for the Pamir.

In this study we present a new inventory based on earlier studies (Kotlyakov et al. 2008, LV et al. 2018) and a systematic analysis of (a) Landsat satellite images acquired over the 1972 to 2018 period and (b) Corona / Hexagon reconnaissance satellite images from the 1961 to 1980 period. Additionally, we used (c) a 1968 DEM derived from the Corona stereo images, the SRTM DEM from 2000 and the ALOS AW3D30 DEM from around 2009 to obtain elevation change patterns. Fine spatial details of glacier changes were revealed by analysing high-speed animations of image time series for all glaciers. Very high-resolution images as available in Google Earth and maps from bing.com were utilised to confirm observations and typical surface structures for surging glaciers. We only included glaciers with an observed active surge and digitised their minimum and maximum extent to obtain length changes. We also noted the timing of their surges (start, end), terminus advance rates and the type of surge (e.g. pulse-like, slow advance, tributary surge, debris load, blocking release).

In total, we identified at least 115 surge-type glaciers of which about 10 are uncertain, as their characteristics are also compliant with a normal advance. Most of them are situated in the central and eastern Pamir but a few are also found in the southern part. Compared to neighbouring regions such as the Karakoram where surge durations take place over many years (often >10), surges in the Pamir are often short-lived (2-3 years) with high advance rates and a pulsating behaviour. All surging glaciers are valley glaciers, some of them are very small (0.3 km²), and interacting tributaries are common, i.e. the surge of one glacier is blocking the flow of another glacier (e.g. Bivachny). A small percentage surges internally (i.e. without an advancing terminus) and a few glaciers surged more than two times during the study period (e.g. Medvezhy). There is no consistency in the timing of surges, i.e. they occur unregularly over the study region. Some glaciers are about twice as long after a surge than before. Despite considerable uncertainties in the DEMs (e.g. due to artefacts), the typical surge pattern is well visible in the elevation change grids for many of the glaciers, which also helped identifying them.