The timing of freezing and thawing of freshwater lakes across the Northern Hemisphere is locally variable but show overall consistent long-term trends towards shorter ice coverage. A shorter duration of ice cover is characterized by delayed freeze-up and earlier break-up dates. The recent rate of these changes for lakes located within the Northern Hemisphere even exceeded the long-term averaged trend. These changes will have significant impact on aquatic biology and ecology but also socio-economic consequences as valuable recreational resources and transportation on ice roads and by ship are effected.
The link between lake ice phenology and air temperature is well established. Lake ice phenology responds locally and regionally to long-term trends in air temperature driven by large-scale climate forcing and in turn effects regional climate and weather events. Therefore, lake ice is a well-established proxy for climate variability and change. Lake ice phenology has been highlighted since the early 1990s as a key variable for cryospheric observations and is nowadays defined as one of the Essential Climate Variables (ECVs) by the Global Climate Observing System (GCOS). Global observation depend upon in situ and satellite observations, but only few products are available.
Different sensors covering divers time periods and spatial coverage demonstrated the potential of remote sensing for this field of research and monitoring purposes. This study utilized the unique potential of the Advanced Very High Resolution Radiometer (AVHRR) for climate studies. AVHRR offers each day global coverage from the early 1980s expectedly until 2020. Data from the full-resolution archive of the Remote Sensing Research Group of the University of Bern (RSGB) are used to derive lake ice phenology.
The pilot study presented here, focuses on lakes in the Baltic region for which extended observational data is available to validate the results. An automated extraction technique, developed by Latifovic & Pouliot (2007), is applied and adapted to the pre-processed data set. The method utilizes the fact that the reflectance increases with the formation of ice in the visible spectral range. An improvement upon this method includes information of the thermal infrared. In contrast to other studies using thermal infrared the thresholds are derived from the data itself, making it unnecessary to define arbitrary or lake specific thresholds. The validation of both methods shows overall higher accuracy for the new approach. Using the new approach, two time series are derived and discussed in terms of changing ice cover duration and climate conditions. According to findings in literature, in situ time series of the lakes revealed no trends in lake ice phenology, despite large warming trends in air and water temperature of both lakes.
Contact: Helga Weber
Advisor/Supervisor: Michael Riffler, Stefan Wunderle
