Supraglacial meltwater lakes
Changes of the glacier surface during the year
Climate and its variability are highly relevant to the Greenland Ice sheet and its outlet glaciers. Warmer air temperatures lead, e.g., to increased melt rates and thus larger amounts of melt water on the ice surface. The meltwater accumulates in topographically induced sinks and form supraglacial lakes. The darker lake surface lowers the albedo of the ice, which further enhances melting.
The meltwater can drain down to the glacier bed. There it reduces the friction between the moving ice and the solid sediment floor below. The lakes therefore serve as a link between atmospheric processes on the one hand and glacier flow dynamics and freshwater input on the other.
In subproject 7, we investigate the meltwater lakes and their dynamics - with high temporal and spatial resolution - using multi-spectral and radar-interferometric remote sensing.
What was the main question you aimed to answer during the first project phase?
The main goal of subproject 7 was the fully automatic detection of supra-glacial lakes in optical satellite images – especially their extent in a chronological sequence over multiple years on the 79°N Glacier and Zachariæ Isstrøm. Furthermore, we aimed to determine of the thawing period and annual lake extent.
Which methods did you use to answer your research question?
We used data from the new Sentinel-2 Copernicus mission (start 2015) to detect all visible water on the glacier surface using the ratio of the backscatter in the blue and red spectral range. The resulting maps were corrected for topographic shadows and clouds, and clipped to topographically induced sinks, to automatically extract maps of meltwater lakes. In order to be able to capture the development of the lakes over time, this method was applied to all satellite images that have been acquired since the start of the mission.
What were your main results?
We found 880 lakes in total upstream the grounding lines of 79°N Glacier and Zachariæ Isstrøm, which we tracked individually in 479 time steps over the course of four years. This enabled us to generate a time series with a temporal resolution of 1.5 days and a spatial resolution of 10 meters, which is currently the densest time series on supraglacial lakes worldwide. The results were recently published under open access license: Hochreuther et al. (2021).
What are your goals for GROCE-2?
- In GROCE-2, we will process and analyze large time series of optical satellite imagery in regard to supraglacial meltwater lakes and meltwater channels. We will continue applying the existing algorithms used in GROCE-1 and develop new techniques using artificial intelligence (deep learning) in order to increase the accuracy and to speed-up the processing time.
- We aim to extent the existing time series with upcoming satellite data. Furthermore, we aim for a complete coverage of Greenland. In this context, we want to implement also an artificial intelligence based cloud detection procedure in order to directly discard unsuitable scenes of satellite images.
- Another goal of the sub-project is to map the temporal variability of supraglacial lake extent over the NEGIS) and to determine their depth. We would like to carry out field measurements and combine our multispectral measurements with data from ICESat-2 satellite missions. By doing so, we will provide first assessments of hydrological balances as input to other GROCE sub-projects and also provide evidence where the meltwater enters the ice.
We work closely with TP8 (with regard to the meltwater budgets) and with TP6 (with regard to field work).