Data Archeology – Reconstructing the Long-Term Temperature Record for Lake Erken

posted in: Lake model blog | 0

Lake Erken is one site where PROGNOS models are being tested and developed. The Erken Laboratory run by Uppsala University has been a site for limnological research since 1946 (Pettersson 2012), but in early years the research efforts were linked to individual research projects and PhD studies, making data availability sporadic over time. Since 1988 meteorological and water temperature data have been collected automatically, but even under this period there are gaps in data related to instrument failure, and other logistical problems.

We now know that global warming is having an ongoing impact on Sweden’s climate, and recent studies (OReilly 2015) suggest that lakes are also being affected. However, these changes can be subtle and since they occur over a long time period may not be easily noticed, especially since lake water temperatures can vary naturally from year to year.  To truly understand the nature of lake warming a consistent long-term record of lake water temperature is needed.  Simone Moras, a master student at Uppsala University, recently used the GOTM model developed for Lake Erken by the PROGNOS project to create a long-term lake water temperature record.  Since the meteorological data needed as input to GOTM, are more readily available than the measured temperature data, we collected meteorological data starting from 1961 and used measured water temperature data to extensively calibrate the GOTM temperature simulations.

Examples of the paper records of form the early days of the Erken Laboratory. On the left are strip charts of solar radiation measurement. On the right or measurements of air temperature


Meteorological data were available digitally from the automated monitoring stations at Lake Erken and several nearby meteorological stations run by the Swedish Meteorological and Hydrological Institute (SMHI). However, before the mid 1980’s data were not digitally available, but could be retrieved from paper records stored at the Erken Laboratory and at SMHI (Fig. 1). Considerable effort was required to digitize these data and to develop methods to consistently merge the data from all sources, but in the end we produced a long-term record of hourly meteorological input to the GOTM model that allowed us to produce a 51 year record of hourly water temperature profiles, that closely match the measured data when it is available (Fig.  2).

Comparison of simulated and measured water temperature at lake surface (~0.5 m) and at the 10 m depth. Simulated is the solid line, measured is the red points. Data from the 1960’s are from paper strip chart recordings.


Analysis of the new temperature data shows that climate change has already had significant effects on Lake Erken’s temperature and thermal structure. Water temperatures are now warmer than during the early years of the study.  This is most evident for the epilimnetic temperatures which are now 1.4 degrees C warmer and have shown a statistically significant rate of increase since 1961.  Examining these data by season (Fig. 3), shows that the strongest trends in increasing water temperature began in 1991, although spring season epilimnetic temperatures show a significant increase beginning in 1961.  From 1991 epilimnetic temperatures have increased at a rate of (0.7, 0.4, and 0.9 °C/decade), while hypolimnetic temperatures have increased, at a rate of (0.6, 0.8 and 0.9 °C/decade) for the spring, summer and fall seasons.  Even more striking are changes in the timing of thermal stratification (Fig. 4).  The lake is now stratified 40 days longer than in the early 1960’s, as a consequence of both an earlier onset of stratification (17 days earlier) and a later loss of stratification (25 days later). There are also significant decreases in the average depth of the thermocline and increases in water column stability during the stratified period.

Fig 3 Mean seasonal water temperature for epilimnion and , hypolimnion of Lake Erken. The trend line is shown only when the trend was significant. The dashed lines divide the two sub-intervals in which the Mann-Kendal analysis was performed (1961-1990 and 1991-2016)


This study shows that significant changes in the water temperature and thermal structure of Swedish lakes, such as Erken, have occurred and continue to occur. Climate change in not a problem of the future, but one that has been occurring for at least 4 decades, as evidenced from changes in Erken water temperature.  These changes in turn will influence lake biogeochemistry and ecology.  Further work in PROGNOS will couple biogeochemical models to GOTM so that such effects can be investigated.

Changes in the duration of thermal stratification in Lake Erken over the period 1961-2016. Slope of the tr.end line is 7.894 days/decade, (p<0.01)

By using readily available meteorological data to simulate lake water temperature with GOTM, and using more sporadic measurements of water temperature to calibrate the model, we were able to produce a long-term continuous record of water temperature that reviled climate change effects that had previously gone unnoticed. We believe that this methodology could be profitably used to evaluate the effects of climate change on other lakes and provide a more detailed assessment of the changes that are now underway.

Petterson, K. 2012. Limnological Studies in Lake Erken Sweden. In L. Bengtsson, R.W. Herschy, and R. W. Fairbridge, editors. Encyclopedia of Lakes and Reservoirs, Springer.

O’Reilly, C. M., et al. 2015. Rapid and highly variable warming of lake surface waters around the globe. Geophysical Research Letters 42:10,773-710,781.