the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Feb 2025
Modeling and observing the Lake Albano dynamics
Abstract. Lake Albano is a monomictic volcanic crater lake in Central Italy with CO2-rich waters presenting CO2 concentration varying over time. Depending on the period of the year, the lake is characterized by strong stratification or rather overturning events. In the warm season, the heating of the surface water results in a highly stratified vertical density profile, while in the cold season, the surface water cooling leads to a potential vertical instability of the water column. In this case, a partial/deep overturning of the lake water column may occur with the degassing in the atmosphere of the CO2 which was accumulated as dissolved species in the deep water layers following seismically induced gas recharge, months to years before. Such a process has been periodically observed in Lake Albano in the past and could pose a potential hazard to the surrounding environment and population. A 3D numerical model is implemented to investigate the lake dynamics and the occurrence of overturning events. The model is validated and calibrated using both historical observations and measurements acquired during this study. These include temperature and salinity profiles from the deepest central portion of the lake, surface water temperature time series recorded by sensors installed on the lake shores, mounted on remotely operated vehicles, and on low-cost, innovative, self-powered drifting buoys. The latter have also been used to assess the modeled surface circulation of the lake.
- Preprint
(6316 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-286', Anonymous Referee #1, 28 Apr 2025
- AC1: 'Reply on RC2', Anita Grezio, 18 Aug 2025
-
RC2: 'Comment on egusphere-2025-286', Anonymous Referee #2, 08 May 2025
In this study, a three-dimensional model is implemented to investigate the thermal dynamics and mixing in Lake Albano, a crater lake with high accumulation of dissolved CO2 and thus a certain potential for a limnic gas eruption.
The purpose of the present study is not sufficiently clearly defined. The text states “to estimate the potential gas hazard of Lake Albano, numerical modeling of the lake water dynamics is crucial for understanding its current and future behavior and stability.” and “In this study, we investigate the characteristics of lake stratification and overturning events at Lake Albano through the results of 3D numerical model simulations, ...”. However, these are rather general statements, and there are no specific research questions or hypotheses mentioned.
Predicting mixing in deep warm “monomictic” lakes is typically a challenging task for lake models. I write “monomictic” in quotation marks, because in many of these lakes, the seasonal mixing depth is variable depending on the meteorological conditions in winter, and only reaches the full depth of the lake every few years in cold winters. The lakes are thus often not truly monomictic. This occasional complete mixing results in a typical slow increase in hyoplimnion temperature during years with incomplete mixing and a faster decline in temperature and a sudden increase in oxygen concentrations during complete mixing events. This is the case, for example, for many of the Northern Italian deep lakes (Rogora et al., 2018, 10.1007/s10750-018-3623-y), and based on previous data, it is probably also true for Lake Albano (Ellwood et al., 2009, doi:10.3274/JL09-68-2-12). Whether or not a complete overturn occurs in a specific year may depend on relatively minor differences in atmospheric forcing, and mixing can also be inhibited by additional chemical density stratification due to salinity (typically resulting from organic matter mineralisation and/or calcite dissolution) and in the case of CO2-rich lakes such as Lake Albano also due to the density contribution of CO2.
That said, I am not convinced that the model presented here can be used to reliably predict mixing in Lake Albano for the following reasons:
- Most importantly, in my opinion, it does not make sense to prescribe a constant vertical turbulent diffusivity in a lake model if the goal of the model is to project vertical mixing. The turbulent diffusivity typically varies by several orders of magnitude both vertically depending on forcing and stratification (with highest diffusivity in the surface layer, low diffusivity in the metalimnion and intermediate diffusivity in the hypolimnion) and seasonally. A correct representation of these dynamics in vertical diffusivity is required to reliably predict vertical mixing in a lake. As far as I know, the SHYFEM model does have the option to calculate vertical diffusivity using a range of turbulence closure approaches. Why was none of these options used?
- The model performance in predicting surface temperature is rather bad, with RMSE > 3°C overall and > 2°C in winter, and has a clear bias (Figure 7). A temperature difference of a few tenths of a degrees can determine whether the lake does or does not mix completely in winter. I think significant additional work would be required to better calibrate the heat flux parameterizations in the model to achieve better agreement with observed surface temperatures. Potentially, also the ERA5 forcing data is not representative for the lake. Specifically, in this hilly region the elevation of the ERA5 grid cell could differ quite a bit from the lake’s elevation and thus result in a bias in air temperature, and local wind speed could significantly differ from the ERA5 grid (as also implied by the analysis of observed currents). It might be helpful to compare with observations from the local meteo station in Viterbo, just a few km south of the lake at similar elevation (https://oscar.wmo.int/surface/index.html#/search/station/stationReportDetails/0-20000-0-16216).
- There is no evaluation whether the model is actually able to correctly reproduce previous observed mixing events.
Furthermore, I disagree with the assessment that salinity is not important for density stratification and mixing in Lake Albano. The annual mean profile presented in Figure 3 shows a salinity difference of 0.1 g/kg between the surface and the bottom of the lake. This approximately corresponds to a density difference of 0.08 g/kg (see, eg., Boehrer and Schultze, 2008, doi: 10.1029/2006RG000210). This is equal to the density difference between water at 9 °C and water at 8 °C. Thus, if the deep water of Lake Albano has a temperature of 9°C, the surface water would have to cool down to 8°C to reach the density of the deep water and allow mixing. The analysis of mixing depth purely based on temperature differences in section 3.4 is therefore in my opinion not valid. This also means that a model would likely need to adequately reproduce the sources of salinity in Lake Albano to correctly predict mixing in years beyond the first year where salinity may by prescribed by the initial conditions.
Finally, it would be useful to add some discussion on the model choice. There are several widely tested 1D lake models available to simulate vertical mixing in lakes (eg GLM, Simstrat), and the simple morphology of Lake Albano with a single deep basin seems to be an ideal case for applying such a model. I do not really see anything in the present manuscript that would justify applying a computationally far more expensive 3D model instead, which is less tested in the context of the present study to simulate vertical mixing in lakes. The manuscript does include some discussion of the spatial dynamics of mixing, which typically starts near the shore due to faster local cooling in shallower parts of the lake. However, this is a known and well-studied effect (eg. Bouffard et al., 2025, doi: 10.1017/flo.2024.31), and there is no discussion in the manuscript why this spatial variability should be relevant for investigating vertical mixing in the context of potentially dangerous CO2 emissions.
Some details:
- The title of the manuscript is rather unspecific and could refer to any type of dynamics in the lake.
- I think in equation (20), there should be division, not multiplication with A0. The numbers in the figure look reasonable, though, so I assume this was calculated correctly.
Citation: https://doi.org/10.5194/egusphere-2025-286-RC2 - AC1: 'Reply on RC2', Anita Grezio, 18 Aug 2025
- AC1: 'Reply on RC2', Anita Grezio, 18 Aug 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 431 | 74 | 17 | 522 | 21 | 33 |
- HTML: 431
- PDF: 74
- XML: 17
- Total: 522
- BibTeX: 21
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
review for https://doi.org/10.5194/egusphere-2025-286
"Modeling and observing the 4 Lake Albano dynamics"
by Anita Grezio et al. (13 co-authors)
In their manuscript, the authors investigate the overturn and
hence the possible sudden release of dissolved CO2 to the atmosphere.
They present measured data and data from a 3D numerical model;
The investigation covers 4 years. Those differ in the depth of
the deep mixing.
The authors present a big package of work. Doing both, field
measurements and running a numerical model requires a lot of
effort. The general understanding of processes is reported correctly.
The tools that are chosen are sort of suited for the purpose, though
not necessarily the ideal approach. The biggest shortcoming lies
in the fact that the goal of the investigation is not fully supported
by the results and I would even - looking at the results - interpret
things differently. (There is no good display of the mixing depth;
Howmoller is questionable; Schmidt number not suited and inconsistent
_ see below)
There are shortcomings in the Figures:
Firstly, Figures (4), (5), 6, 7, 10, 11, 12, 13, 14, (15)
contain text, axes values, legends that are not readable (those in
brackets "hardly readable"): If this text contains information for the
reader, larger fonts are needed. If there is no information, remove the
text.
Secondly, the captions are not acceptable. Figures must be complete
in themselves: looking at the graphics and reading the caption should
suffice to understand what is displayed.
Thirdly, the figures 10, 11, 12, 13: I understand there is lots of
work going into producing such displays: Hence I can relate to the fact
that the authors want to show their efforts: I agree, one or two of
those panels make sense, but in general there is no longitudinal
information: showing profiles would make more sense, and would indicate
temperature gradients better than the colour coding.
Fourth, Figure 16: something is wrong! each year should start with a
value similar to the end of the previous year: the end of 2021 does not
fit with the beginning of 2022.
The equations (no numbers) on page 20 are not understandable and
probably formally not correct.
The major conclusion of the manuscript claims that deep circulation
happenend in 2020 and 2022, but not in 2021 nor in 2023.
While I can relate to the observation of variability, looking at the
displays in Fig 10, I would say the lake has not mixed deeper than 95 m
in 2020. In 2022 (Figure 12) I can see that an overturn may have happened.
I agree: no indication of this in 2021 nor 2023.
The Schmidt number is not really suited for an indication of mixing depth.
The Howmoller approach delivers other results than the optical
impression of the displays. Do the authors state clearly, whether they
think Howmoller is a useful approach?
Usually mixing depths can be detected (1) in measurements: by oxygen
profiles, better even CO2 profiles as the release of this gas was the
major intention of this study (2) in numerical simulations: mixing depth
can be detected by inserting particles in the deep water.
In conclusion, I find this contribution is not ready for publication:
while some of the shortcomings can be removed easily, a few things
remain, which are hard to correct.
smaller points:
line 42: not only volcanic: gas pressure of concern also through geochemical
processes (Sanchez-Espana: Guadiana pit lake) or pollution (Horn et al.:
Vollert pit).
line 141: add the charges to the ions.
Figure 3: 0.1 g/kg of salinity difference is NOT homogeneous; This has even
noticeable effect on density.
line 153 why ERA 5 , was there no better source for weather data?
Figure 6: coast ? -> shore line
Figure 7: This comparison of temperature. I see a deviation between
1 and 5°C from the SE observation, also in winter: Why are the authors
convinced that the model is good enough to simulate deep mixing?
line 246: different -> separate
Figure 10, 11, 12 , 13: where is north or south: write clearly the
months over the columns and dates in front of rows and write in the
caption what is displayed.
Figure 15 has a bad colourbar for the temperature.
line 343: what is the density approach that is used?