Oxycline oscillations induced by internal waves in deep Lake Iseo

Lake Iseo is undergoing a dramatic deoxygenation of the hypolimnion, representing an emblematic example among the deep lakes of the pre-alpine area that are, to a different extent, undergoing reduced deep-water mixing. In the anoxic deep waters, the release and accumulation of reduced substances and phosphorus from the sediments are a major concern. Because the hydrodynamics of this lake was shown to be dominated by internal waves, in this study we investigated, for the first time, the role of these oscillatory motions on the vertical fluctuations of the oxycline, currently situated at a depth of approximately 95 m, where a permanent chemocline inhibits deep mixing via convection. Temperature and dissolved oxygen data measured at moored stations show large and periodic oscillations of the oxycline, with an amplitude of up to 20 m and periods ranging from 1 to 4 days. Deep motions characterized by larger amplitudes at lower frequencies are favored by the excitation of second vertical modes in strongly thermally stratified periods and of first vertical modes in weakly thermally stratified periods, when the deep chemical gradient can support baroclinicity regardless. These basin-scale internal waves cause a fluctuation in the oxygen concentration between 0 and 3 mg L−1 in the water layer between 85 and 105 m in depth, changing the redox condition at the sediment surface. This forcing, involving approximately 3 % of the lake’s sediment area, can have major implications for the biogeochemical processes at the sediment–water interface and for the internal matter cycle.


32
The physical processes occurring at the sediment-water interface of lakes are crucially controlling the 33 fluxes of chemical compounds across this boundary (Imboden and Wuest, 1995), implying severe 34 implications for water quality. In stratified lakes, the boundary-layer turbulence is primarily caused 35 by wind-driven internal wave motions (Imberger, 1998). Consequentially, the periodicity of these Alpnach, a 34 m deep lake known to feature pronounced seiching. 48 In thermally stratified lakes, a further driver for flux unsteadiness is the periodic occurrence of cyclic 49 convective turbulence in the sediment area located in the water layer characterized by thermocline 50 fluctuations. Here the sediments are exposed to pronounced temperature oscillations during internal 51 seiches. In particular, during the upslope current, cold deep water flows over the warmer sediments. 52 The resulting intermittent instability drives free convection, accelerating the fluxes at the sediment-53 water interface by more than one order of magnitude, as experimentally observed by Kirillin et al. 54 these layers has been shown to support higher vertical baroclinicity (Salvadé et al., 1988;Roget et al. 65 2017), whose amplitude is typically larger than that of the thermocline. Accordingly, we hypothesize  where the oxycline is located, in order to provide an estimation of the spatial and temporal extent of 85 oxygen fluctuations at the surface of deep sediments. As deep sediments are generally known to be 86 potentially redox-sensitive phosphorus (P) sinks, we discuss our results in light of the expected P 87 fluxes from the contiguous sediments. The importance of this research is motivated by the observation 88 that Lake Iseo is currently undergoing a change in mixing pattern and P recycling, so that a deeper 89 understanding of the sediment P release dynamics is crucial to forecast the possible future trajectories   Lake Iseo, made in 1967, the lake was described as a monomictic and oligotrophic lake, featuring a 97 fully oxygenated water column and P concentrations of a few µg L -1 . Starting in the 1980s, the 98 accumulation of solutes from biomass processing, in combination with climatic factors, has gradually 99 inhibited deep water renewal. With decreased mixing depth, deep water quality deteriorated, 100 including increases in P concentrations and persistence of anoxic conditions (Garibaldi et al., 1999). conditions, to prevent a deeper convective mixing. In the anoxic monimolimnion below, the amount 106 of P does not show any signs of decrease (currently with a space averaged concentration of ~111 µg 107 TP L -1 ), and a recent field campaign has shown that the P stock is supplied to a comparable extent (~ 108 15 tons of P year -1 ) by both sedimentation from the layers above and by the fluxes from the sediment.  Figure 1. We followed the meteorological and mechanical forcing at the lake surface in high 114 temporal resolution (60 s) by means of two on-shore stations measuring wind speed and direction, air 115 temperature, air humidity and short wave radiation (SS-1 and SS-2). Furthermore, a floating station 116 (LS-N) measured wind speed and direction and net long-wave radiation. LS-N is further equipped 117 with eleven submerged loggers that measure the temperature (± 0.01°C accuracy, 60 s interval), well 118 describing the vertical movements of the thermal gradient (see Fig. 2). In October 2017, we added 119 three additional temperature loggers at 55, 75 and 113 m depth, to better describe the temperature 120 fluctuations below the thermocline thanks to their higher accuracy (± 0.002°C).

121
To capture the vertical fluctuations of the oxycline, we installed four submersible instruments 122 (miniDOT, Precision Measurement Engineering, Vista, Ca, USA) at LS-S between 85 and 105 m, 123 measuring dissolved oxygen (DO) for nine consecutive months at a 1 min -1 sampling frequency (see 124   Table 1). These loggers rely on a fluorescence-based oxygen measurement with an accuracy of ± 5% 125 of the measured value (mg L -1 ). The LS-S logger chain was installed at a 105 m deep location in the 126 southern basin. Two additional loggers were installed in the northern station (LS-N) at 85 and 95 m.

127
As shown in Table 1 focus on the data analysis from July 2017 to February 2018, which fully captures the evolution of the 130 oxygen content during the transition from a strongly to a weakly stratified period.

131
On 21-22 July 2017, we also conducted a field campaign aimed at investigating the oxygen profiles  In this study, we used two numerical models to better highlight different dynamic aspects of the 140 measured internal oscillations. This required to identify the temporal evolution of the periodicity and 141 the spatial structure of the natural modes in Lake Iseo. In the case of Lake Iseo, the bathymetry was discretized with a 160 x 160-m horizontal 147 grid, while the averaged vertical density structure was schematized with a 4 to 2 layers structure on a 148 monthly basis, as detailed in Table 2. As typical for the sub-alpine lakes, a pronounced 3-layers     and only a salinity stratification is present.

296
With regard to the spatial structure of these modes, Table 2 summarizes the 3D results in terms of 297 maximum interface displacements at different lake locations in four representative periods of the year.

298
In the following, we will mostly focus on the V1H1 and V2H1 oscillations of the thermocline (ξ2) and  The analysis of the measured data previously shown suggests the presence of both a V1H1 and a 334 V2H1 response. The obtained numerical results allowed us to clarify the nature of these oscillations 335 and extend the spatial information provided by local measurements.

336
The field data showed that the thermocline oscillates regularly over a 1 day period from July to 337 November (Fig. 3b). At that time, the natural period of V1H1 is daily, too, confirming that the upper 338 water layers are dominated by this type of motion. Conversely, for the deeper oscillations we observed 339 a different type of response: a shorter daily oscillation in Oct-Nov (e.g. Fig. 4), and a longer 2-4 days 340 oscillation from August to September, and from December to January (e.g. Fig. 5, 6). In the first case, 341 (Oct-Nov), the daily period of the measured oscillations fits the natural period of V1H1. This is 342 consistent with the spatial structure of this motion (see e.g. Fig 4 and related comments), characterized 343 by a counter-phase response at the two lakes ends (H1) and a similar amplitude at the different depths 344 (V1). In the strongly stratified period (Aug-Sept), we occasionally observed a decoupled internal 345 wave response at the different depths. By comparing the periodicity of the measured oscillations and 346 that of the natural modes (see Fig. 3), the thermocline appears to be dominated by a V1H1 motion (~ 347 1 day period), while the oxycline by a V2H1 motion (~ 2-3 days period). Again, this is consistent 348 with the observed spatial structure of these motions (see e.g. Fig. 5 and related comments). It is of 349 major interest to observe that in correspondence of the summer stratification at LS-S, the deeper 350 amplitude ξ4 of the V2H1 mode results 5.7 times larger than ξ2 . This is likely to explain why V2H1 351 mode is largely dominant deeper in the waters, while it is sheltered by the V1H1 oscillations around 352 the thermocline. Finally, in the third period (Dec-Jan) we observed the superimposition of ~2 days 353 and ~4 days large oscillations at the depth of the oxycline. According to the periodicities of the natural 354 modes, they correspond to a V1H1 and V2H1 mode (see Fig. 3e). With respect to the previous case 355 (Aug. 17), the evidence of a large amplitude V1H1 mode at this depth is consistent with the increased 356 displacements reported in Table 3 in correspondence of a weaker density stratification.   The resulting basin-specific areas subjected to alternating redox conditions are shown in Figure 10b. 452 We estimated that overall 1.9 km 2 of bottom sediments of Lake Iseo, 3% of the whole area, are on the P binding potential of sediments that are regularly in oxic but now temporarily exposed to anoxic 478 waters are not found to differ from those in permanently oxic conditions (Aller, 1994). In contrast to 479 that, the role of sediments as sink and source of P in regularly anoxic environments is known to be 480 controlled by other diagenetic processes including P supply, microbial mineralization, aluminium and 481 sulphide availability (Hupfer and Lewandowski, 2008). However, the collective susceptibility of 482 these processes to excursion in oxygen availability is less well understood.    Table 1. Summary of the oxygen data measured in Lake Iseo at a sampling frequency of 1 min -1 .    Table 3. Maximum vertical displacement ξ of the layer interfaces with respect to their equilibrium 599 level for the first three vertical modes in Lake Iseo at four different locations. These locations, whose 600 depth is z, are shown in Fig.1. The interfaces displacements were simulated with AEM3D by forcing 601 with a spatially uniform sinusoidal wind, with a maximum speed of 5 m/s and a period equal to the 602 natural one predicted by the modal model (see grey shading in Table 2). ξi indicates the upper interface 603 of each i th layer, whose depth is reported in Tab Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2018-542 Manuscript under review for journal Hydrol. Earth Syst. Sci. Discussion started: 23 October 2018 c Author(s) 2018. CC BY 4.0 License.