2019
One-dimensional hydrodynamic models are nowadays widely recognized as key
tools for lake studies. They offer the possibility to analyze processes at
high frequency, here referring to hourly timescales, to investigate
scenarios and test hypotheses. Yet, simulation outputs are mainly used by
the modellers themselves and often not easily reachable for the outside
community. We have developed an open-access web-based platform for
visualization and promotion of easy access to lake model output data updated
in near-real time (
Aquatic research is particularly oriented towards providing relevant tools and expertise for practitioners. Understanding and monitoring inland waters is often based on in situ observations. Today, the physical and biogeochemical properties of many lakes are monitored using monthly to bi-monthly vertical discrete profiles. Yet, part of the dynamics is not captured at this temporal scale (Kiefer et al., 2015). An emerging alternative approach consists in deploying long-term moorings with sensors and loggers at different depths of the water column. However, this approach is seldom used for country-level monitoring, although it is promoted by research initiatives such as GLEON (Hamilton et al., 2015) or NETLAKE (Jennings et al., 2017).
It is common to parameterize aquatic physical processes with mechanistic models and ultimately use them to understand aquatic systems through scenario investigation or projection of trends in, for example, a climate setting. In the last decades, many lake models have been developed. They often successfully reproduce the thermal structure of natural lakes (Bruce et al., 2018). Today's most widely referenced one-dimensional (1-D) models include (in alphabetical order) DYRESM (Antenucci and Imerito, 2000), FLake (Mironov, 2005), General Lake Model (GLM; Hipsey et al., 2014), GOTM (Burchard et al., 1999), LAKE (Stepanenko et al., 2016), Minlake (Riley and Stefan, 1988), MyLake (Saloranta and Andersen, 2007), and Simstrat (Goudsmit et al., 2002). The results from these models are mainly used by the modellers themselves and often not easily accessible for the outside community.
The performance of lake models is determined by the physical representativeness of the algorithms and by the quality of the input data. The latter include (i) lake morphology, (ii) atmospheric forcing, (iii) hydrological cycle (e.g., inflow, outflow, and/or water level fluctuations), and (iv) light absorption. In situ observations, such as temperature profiles, are required for calibration of model parameters. To support this approach, it is important to promote and facilitate the sharing of existing datasets of observations among scientists and practitioners. Conversely, scientists and practitioners should benefit from the model output, which is often ready to use, high frequency, and up to date. Yet, model output data should not only be seen as a tool for temporal interpolation of measurements. Models also provide data of hard-to-measure quantities which are helpful for specific analyses (e.g., the heat content change to assess impact of climate change or the vertical diffusivity to estimate vertical turbulent transport). Models finally support the interpretation of biogeochemical processes which often depend on the thermal stratification, mixing, and temperature. In a global context of open science, collaboration between the different actors and reuse of field and model output data should be fostered. Such win–win collaboration serves the interests of lake modellers, researchers, field scientists, lake managers, lake users, and the public in general.
In this work, we present a new automated web-based platform to visualize and
distribute the near-real-time (weakly) output of the one-dimensional
hydrodynamic lake model Simstrat through an user-friendly web interface. The
current version includes 54 Swiss lakes covering a wide range of
characteristics from very small volume such as Inkwilersee (
We use the 1-D lake model Simstrat v2.1 to model 54 Swiss lakes or reservoirs
(see Appendix A for details of modeled lakes) in an automated way. Simstrat
was first introduced by Goudsmit et al. (2002) and has been
successfully applied to a number of lakes
(Gaudard et al.,
2017; Perroud et al., 2009; Råman Vinnå et al., 2018; Schwefel et
al., 2016; Thiery et al., 2014). Recently, large parts of the code were
refactored using the object-oriented Fortran 2003 standard. This version of
Simstrat provides a clear, modular code structure. The source code of
Simstrat v2.1 is available via GitHub at
In addition to the improvements already described by Schmid and Köster (2016), Simstrat v2.1 includes (i) the possibility to use gravity-driven inflow and a wind drag coefficient varying with wind speed – both described by Gaudard et al. (2017) – and (ii) an ice and snow module. The ice and snow module employed in the model is based on the work of Leppäranta (2014, 2010) and Saloranta and Andersen (2007), and is further described in Appendix B.
A Python script was developed to (i) retrieve the newest forcing data directly from data providers and integrate them into the existing datasets, (ii) process the input data and prepare the full model and calibration setups, (iii) run the calibration of the model for the chosen model parameters, (iv) provide output results, and (v) update the Simstrat online data platform to display these results. The script is controlled by an input file written in JavaScript Object Notation (JSON) format, which specifies the lakes to be modeled together with their physical properties (depth, volume, bathymetry, etc.) and identifies the meteorological and hydrological stations to be used for model forcing. The overall workflow is illustrated in Fig. 1.
General workflow diagram. Model input
Input data sources used for the model.
Model parameters. The geothermal heat flux is based on existing geothermal data for Switzerland:
The asterisk (
Table 1 summarizes the type and sources of the data fed to Simstrat. For
meteorological forcing, homogenized hourly air temperature, wind speed and
direction, solar radiation, and relative humidity from the Federal Office of
Meteorology and Climatology (MeteoSwiss, Switzerland) weather stations are used. For
each lake, the closest weather stations are used. Air temperature is
corrected for the small altitude difference (see Appendix A) between the
lake and the meteorological station, assuming an adiabatic lapse rate of
The timeframe of the model is determined by the availability of the
meteorological data (air temperature, solar radiation, humidity, wind,
precipitation). Initial conditions for temperature and salinity are set
using conductivity–temperature–depth (CTD) profiles or using the temperature
information from the closest lake. We apply different data patching methods
to remove data gaps from the forcing depending on the length of the data
gap. For small data gaps with duration not exceeding 1 d, the dataset is
linearly interpolated. In total,
Model parameters are set to default values, and four of them are calibrated
(see Table 2). The parameters
Illustration of the interactive map displayed on the home page of
the online platform:
The online platform (accessible at mean lake temperature: heat content: Schmidt stability: timing of summer stratification: we use a threshold based on the Schmidt stability
to determine the beginning and end of summer stratification. The lake is assumed to be stratified
for timing of ice cover: we use the existence of ice to determine beginning and end of ice covered period.
Performance of the model for the different lakes, as shown by the
root mean square error (RMSE) and the correlation coefficient. Six lakes
(with symbol
From these results, we create static and interactive plots. The latter are
created using the Plotly Python library (see history (e.g., contour plot of the whole temperature time series, line plot of the whole time series of Schmidt stability); current situation (e.g., latest temperature profile); statistics (e.g., average monthly temperature profiles, long-term trends).
All output and processed data are directly available from the online
platform.
Analysis of model output allows to compare the response of the different systems to specific events or to long-term changes. The Simstrat model web interface provides regional long-term high-frequency data updated in near-real time as output. This represents a novel way to monitor, analyze, and visualize processes in aquatic systems and, most importantly, grant the entire community direct access to the findings. The coupling between Simstrat and PEST provides an effective way to calibrate model parameters. The uncertainty quantification finally allows an appropriate informed use of the output data. Yet more advanced methods for both parameter estimation and uncertainty quantification such as Bayesian inference (Gelman et al., 2013) should be applied to Simstrat.
Out of the 46 calibrated lakes, the post-calibration root mean square error
(RMSE) is
We illustrate the potential of high-frequency lake model data with two examples: first by briefly showing the long-term changes caused by climate change in Lake Brienz (Sect. 3.1), and secondly by investigating the differential response of lakes across Switzerland to episodic forcing (short-term extremes; Sect. 3.2).
Evolution of several indicators for Lake Brienz over the period
1981–2018; all linear regression have
Over the period 1981–2015, yearly averaged simulated surface temperatures
in Lake Brienz increased with a significant (
Comparison of timing of stratification and ice cover for the considered lakes. The colored areas represent the mean periods of summer stratification (red) and ice cover (blue); the vertical lines represent the last year (here 2017). The transparency for the ice cover indicates the freezing frequency: full transparency means that ice was never modeled, while no transparency means that ice was modeled every winter. Lakes are ordered from left (low elevation) to right (high elevation). The time period of data used is indicated in Appendix A.
The temperature increase was significantly smaller in the hypolimnion, with
a minimum trend at the lake bottom of 0.16
The vertically heterogeneous warming modeled in Lake Brienz is consistent
with previous observations showing that the difference in warming between
the surface and the bottom increases the strength and duration of the
stratified period (Zhong et al., 2016; Wahl and
Peeters, 2014). We simulate an earlier onset of the stratification in spring
of
Such analyses can be extended to all modeled lakes. An intercomparison of the temporal extent of summer stratification and winter ice cover period is illustrated in Fig. 5. An altitude-dependent decrease of the duration of summer stratification is observed, along with a stronger corresponding increase in the duration of the inverse winter stratification from 1200 m a.s.l. This is possibly linked to an altitude dependency of climate-driven warming in Swiss lakes, first reported by Livingstone et al. (2005), which may be caused by a delay in meltwater runoff (Sadro et al., 2018). Here, this process is not directly resolved but incorporated through the calibration procedure spanning all seasons.
In conclusion, the online platform provides all the data to estimate the past warming rate of lakes and evaluate how the different external processes contribute to their heat budgets. The change in the thermal structure depends mostly on the change in atmospheric forcing, yet other factors such as the changes in discharge and temperature from the tributaries and the light absorption into the lake should also be taken into account. We specifically show that the warming rate of the lake surface temperature significantly differs from that of depth-averaged temperature, thereby highlighting the benefit of using either in situ observations resolving the thermal structure over the water column or hydrodynamic model output for assessing climate change impacts on lake thermal structure.
A major drawback of traditional lake monitoring programs in Switzerland is the coarse temporal resolution, with measurements often performed on a monthly basis. This resolution only allows to detect long-term trends when measurements are conducted over an extended period typically longer than 30 years. However, traditional monitoring programs cannot resolve the impact of short-term events and their consequences for the ecosystem. This is a strength of high-frequency (hourly timescale) lake modeling, which allows for simulation and comparison of the effects associated with rapid and often severe events such as storms. Based on high-frequency observations, Woolway et al. (2018) showed the effects of a major storm on Lake Windermere. They observed a decrease in the strength of the stratification, a deepening of the thermocline and the onset of internal waves oscillations ultimately upwelling oxygen-depleted cold water into the downstream river. Furthermore, Perga et al. (2018) illustrated how storms could be just as important as gradual long-term trends for changes in light penetration and thermal structure in an alpine lake.
Here, we demonstrate how high-frequency model output can be used to study the
influence of specific events on the thermal dynamics of lakes. As an
example, we focus on 28 June 2018, when Switzerland experienced
a strong but by no means exceptional storm with northeasterly winds mainly
affecting the northwestern part of the country – the mean wind speed
during that day is shown spatially in Fig. 6a. The evolution of the
stratification strength, illustrated here by the Schmidt stability, is given
in Fig. 6b for one of the most affected lakes, Lake Neuchâtel
(
So far, climate-driven warming has been recognized to cause an overall increase in lake stratification strength and duration, and a gradual warming of the different layers (Schwefel et al., 2016; Zhong et al., 2016; Wahl and Peeters, 2014). Air temperature trend was the most studied forcing parameter. Yet, the dynamics of extreme events (such as heat waves, drought spells, storms), including their changes in strength and distribution, has been comparatively overlooked. Scenario exploration, climate change studies, or historical forcing reanalysis should be integrated in such web-based hydrodynamic platforms to assess their roles in modifying the lake thermal structures and heat storage.
The workflow presented in this paper allows openly sharing
high-frequency, up-to-date and permanently available lake model results for
multiple users and purposes. We demonstrated the benefit of the platform
through two simple case studies. First, we showed that the high-frequency
modeled temperature data allow a complete assessment of the effect of
climate change on the thermal structure of a lake. We specifically show the
need to evaluate changes in all atmospheric forcing, in the watershed or
throughflow heat energy, and in light penetration to accurately assess the
evolution of the lake thermal structure. Then, we showed that the high-frequency
modeled data can be used to investigate special events such as
wind storms; there, in situ measurements under current temporal resolution
are failing. More generally, these results are well suited for the following
applications and target groups:
For the public, the platform serves as an informative website enabling easy
access to broad quantities of regional scientific results, with the
intention of raising interest about lake ecosystem dynamics. For lake managers, the platform makes relevant information available, such
as (i) near-real-time temperature and stratification conditions of the
lakes and (ii) simple statistical analyses such as monthly temperature profiles
and long-term temperature trends. For researchers, this work can facilitate (i) scenario modeling of any of
the lakes, as the basic model setup is ready to use, (ii) improvement of the
lake model with addition of previously unresolved processes (e.g.,
resuspension with changed light properties), (iii) access to variables that
were previously not or irregularly available (e.g., vertical diffusivity,
heat content, stratification, and heat fluxes), and (iv) specific comparative
analyses, whereby a given question can be investigated simultaneously over
many lakes (e.g., the impact of climate change or a regional storm).
By promoting a cross exchange of expertise through openly sharing of in situ and model data at high frequency, this open-access data platform is a new path forward for scientists and practitioners.
The workflow was developed for Swiss lakes but can be easily extended to
other geographical area or at global scale by using other meteorological
input data. Simstrat and the Python workflow are available on
This table summarizes the main properties of the 54 lakes we model in this work. The full dataset is available as a JSON file. The superscript “a” after the lake name indicates that this lake was not calibrated due to the lack of observational data. MeteoSwiss is the (Swiss) Federal Office of Meteorology and Climatology. FOEN is the (Swiss) Federal Office for the Environment. The superscript “b” indicates lakes where Secchi disk depths are available. For lakes with clearly defined multiple basins such as Lake Lucerne, Lake Zurich, Lake Constance and Lake Lugano, each basin is considered as a separated lake connected to the other basins by inflows/outflows.
Continued.
Continued.
Continued.
Continued.
Meteorological stations.
The ice and snow module employed is based on the work of
Leppäranta (2014, 2010) and Saloranta and Andersen (2007), and includes the following physical processes:
air-temperature-dependent formation and growth of black ice, including the insulating effect of a snow cover; snow layer build-up, including the compression effect due to the weight of fresh snow; buoyancy-driven formation of white ice; shortwave irradiance reflection and penetration into the underlying water column; and melting of snow, white and black ice due to both the direct heat flux
through the atmospheric interface and the absorption of shortwave
irradiance.
Three layers are used to represent black ice, white ice, and snow. An instant supply of water through cracks in the black ice is assumed to occur in order to form white ice. The water stored in ice and snow is neither withdrawn during ice formation nor added during melting to the water balance. Furthermore, the effect of liquid water pools on top of or between the layers is neglected.
The ice module is activated as the water temperature in the topmost grid
cell
If ice cover is present and if the atmospheric temperature
There,
If the snow mass
In this model, we assume continuous supply of water through cracks in the black ice to form white ice. The formation of white ice takes place instantaneously each time step and we do not consider the influence of pools under the snow for melting or shortwave irradiance penetration.
If ice cover is present and if
There,
Calculating
As
After obtaining
Equation (B24) is only applied to
To test the ice module, Simstrat was calibrated in Sihlsee with PEST using
monthly resolved vertical temperature profiles (2006 to 2008, RMSE 1.2
Ice model performance in Sihlsee (2012 to 2018) showing modeled white ice (orange), black ice (green), and total ice cover (white and black ice combined, in blue) against measurements (black).
The algorithm below is based on the equations from the lake HeatFluxAnalyzer
(see declination of the Sun (rad): cosine of the solar zenith angle (–): air mass thickness coefficient (–): dew point temperature ( precipitable water vapor (cm): attenuation coefficient for water vapor (–): attenuation coefficient for aerosols (–): attenuation coefficient for Rayleigh scattering and permanent gases (–):
effective solar constant (W m clear-sky solar radiation (W m
The new version of Simstrat was developed by FB, AG, and LRV. The workflow was developed by AG. The ice model was developed by LVR. The concept of the workflow was defined by DB. All authors contributed to the validation of the model and interpretation of the results. AG and DB wrote the manuscript with contributions from FB, LVR, and MS.
The authors declare that they have no conflict of interest.
We thank Davide Vanzo for helping with the docker and the scripts, and
Michael Pantic for helping restructuring version 2.1 of Simstrat. We
finally thank Marie-Elodie Perga for her comments on a preliminary version
of the paper. The full list of acknowledgements regarding in situ
observations can be found here:
This paper was edited by Min-Hui Lo and reviewed by three anonymous referees.