Lake Tarfala, Northern Sweden - Remote Sensing of Ice Phenology Using Sentinel-1 Backscatter and Coherence Time Series

<p>Lake Tarfala is a small (~0.5 km<sup>2</sup>) glacier-proximal lake in the Kebnekaise Mountains in Northern Sweden, located at an altitude of 1162 meters above sea level, and close to Tarfala Research Station run by Stockholm University. Only very limited direct monitoring of lake ice phenology using ground observations is available so far, and, long polar nights and often persistent cloud cover at such altitude limit the use of optical remote sensing. However, active microwave radar signals illuminate the target and penetrate through the cloud cover allowing to monitor the lake independent of weather or time of day. In this study, we opt for the Level-1 GRD (Ground Range Detected) and SLC (Single Look Complex) products from the twin Sentinel-1 satellites which provide a coverage of Lake Tarfala at a very high spatial and temporal resolution. We aim to make use of a total of 60 scenes (June 2020 - May 2021) to create the backscatter and coherence time series. Further, we aim to associate the variation in intensity seen in the backscatter time series to the backscattering potential of the medium. It has been shown [1] that an increase in intensity is observed when transitioning from ice-free waters to the initial freeze-up (ice-on) stage. Around ice-on, the intensity would, however, be comparatively low as the ice cover would be very thin and not yet fully developed. The availability of in-situ high-resolution time-lapse imagery and air temperature data from a pilot project carried out during the fall of 2020 [2] will be exploited to assist in the detection of the initial ice formation and freeze-up. Over the course of winter, ice will continue to thicken and a subsequent increase in backscatter intensity is expected until it reaches a saturation point where it stabilises, until the onset of melt in the subsequent spring/summer, when finally, the detection of ice-off (water free of ice) can be characterised by low backscatter values. Furthermore, loss of interferometric coherence upon the onset of melt will aid the backscatter time series when it fails to show a clear signal. We expect to track and provide a complete timeline of the different ice-phenology stages, namely the onset of freezing and the date of complete ice-on, the ice-thickening, the onset of surface melt and the date of complete ice-off. We expect that this study will provide a basis for Arctic lake ice monitoring for various applications such as management of winter water resources, understanding the seasonal and inter-annual land-atmosphere greenhouse gases and energy flux exchanges and biological productivity.</p><p>References:</p><p>1. Morris, K., Jeffries, M.O., Weeks, W.F. Ice processes and growth history on Arctic and sub-Arctic lakes using ERS-1 SAR data. Polar Rec. 1995, 31, 115-128.</p><p><br>2. Weckström, J., Korhola, A. Kirchner, N., Virtanen, T., Schenk, F., Granebeck, A., Prakash, A. “Lake Thermal and Mixing Dynamics under Changing Climate” and “Towards a multi-approach detection and classification of ice phenology at Lake Tarfala”. Pilot projects funded by Arctic Avenue (a spearhead research project between the University of Helsinki and Stockholm University).</p>