Department Surface Waters - Research and Management

Lake Kivu

A fascinating ecosystem and a source of energy

Lake Kivu is situated in East Central Africa and lies on the border of its riparian countries, the Democratic Republic of the Congo and Rwanda. The Lake's surface covers 2385 km2 and its volume amounts to 550 km3 with a maximum depth of 485 m. Its surrounding area offers home to about 2 million people.

Lake Kivu is an amazing and unique aquatic system in several respects. Not only is its water column strongly density stratified, the lake water also contains an estimated amount of 62 km3 of methane and 300 km3 of carbon dioxide. The extraordinarily high concentrations of these gases pose a severe risk to the riparian population in the event of uncontrolled degassing. Such a catastrophic event seems very unlikely under the current conditions, but cannot be completely ruled out in case of an extraordinary event such as a volcanic eruption within the lake. However, the dissolved methane is not only a danger but also a valuable resource.

The involvement of Eawag in research on Lake Kivu has been initiated by the eruption of the volcano Nyiragongo in 2002. Since then, we have conducted several research and consulting projects on Lake Kivu, which are summarized below. The research carried out concerns the unique ecosystem as well as the methane extraction.

The brochure Methane Extraction from Lake Kivu – Scientific background gives a general overview of the important processes that need to be considered when using the methane resources from Lake Kivu. The scientific knowledge on the physics, geochemistry and biology of Lake Kivu has been summarized in the book Lake Kivu - Limnology and biogeochemistry of a tropical great lake.

Results from research projects

Double diffusion

The accumulation of huge amounts of methane and carbon dioxide in Lake Kivu is only possible because density increases strongly with depth below ~60 m. Thus the seasonal convective mixing which happens in most other lakes is inhibited. Upward transport occurs through slow upwelling (< 1 m per year) caused by sub-aquatic inflows, and in part by “double-diffusion”. Investigating upward transport is important for a general understanding of Lake Kivu and in particular for a sustainable extraction of the methane.

Double-diffusion occurs because both temperature and salinity (in Lake Kivu also dissolved gases) are increasing with depth. In regions where the stabilizing gradient of salinity is two to eight times stronger than the destabilizing gradient of temperature, mixed layers (m-scale) form, which are separated by stable interfaces (dm-scale). Investigating the vertical fluxes of temperature, salts and dissolved gases through such double-diffusive staircases is the goal of this project.

Three field campaigns in 2010 (see movie for impressions), 2011 and 2015 have so far been conducted and the main results are:
 

  1. We showed that the most common parameterization for vertical fluxes through double-diffusive systems works well for the step sizes observed in Lake Kivu, but underestimates the heat flux in systems with larger steps (like in parts of the Arctic Ocean) by up to a factor of four.
     
  2. We conducted direct numerical simulations of double-diffusive interfaces under Lake Kivu conditions and showed that the interface structure is well reproduced by the simulations and that the heat flux through double-diffusive interfaces is close to molecular.
     
  3. The results above rely on measuring large gradients within the temperature and salinity interfaces. Even though microstructure sensors are extremely fast (ms) they are not fast enough to accurately resolve double-diffusive interfaces. We therefore developed an in-situ method to improve the current knowledge of the sensor responses.

Publications

The role of double diffusion for the heat and salt balance in Lake Kivu

Double diffusion in lakes and oceans can transform vertical gradients into staircases of convectively mixed layers separated by thin stable interfaces. Lake Kivu is an outstanding double‐diffusive natural laboratory with > 300 such steps over the permanently stratified deep basin. Here, we use 315 microstructure profiles (225 measured in Rwanda and 90 in the DRC) to shed light on the heat and salt balances of Lake Kivu. Comparing profiles from 2011 and 2015 reveals warming of 8.6 mK yr−1 below 80 m depth and negligible changes in salinity. The double‐diffusive layering is coherent over horizontal distances of 20–30 km and remained unchanged between 2011 and 2015, indicating little variability. The mean estimated dissipation within mixed layers is 1.5 × 10−10 W kg−1. If unshaped Batchelor microstructure spectra are interpreted as nonturbulent, the rescaled dissipation of 0.44 × 10−10 W kg−1 corresponds to a vertical heat flux of 0.10 W m−2, which agrees with the molecular heat flux through the adjacent stable interfaces. Using estimates of upwelling, temporal changes of temperature and salt, and vertical double‐diffusive fluxes, we established heat and salt balances, which require lateral heat and salt inputs. For salt, lateral input of freshwater at the main gradients balances upwelling. For temperature, however, the divergence of the vertical double‐diffusive fluxes can only be balanced by horizontal inputs supplying cool water above and warm water below the main gradients. This suggests that lateral inputs of water at various depths are the main drivers for this unique double‐diffusive phenomenon in Lake Kivu.
Sommer, T.; Schmid, M.; Wüest, A. (2019) The role of double diffusion for the heat and salt balance in Lake Kivu, Limnology and Oceanography, 64(2), 650-660, doi:10.1002/lno.11066, Institutional Repository

Double diffusion in saline Powell Lake, British Columbia

Powell Lake contains a deep layer of relic seawater separated from the ocean since the last ice age. Permanently stratified and geothermally heated from below, this deep layer is an isolated geophysical domain suitable for studying double-diffusive convection. High-resolution CTD and microstructure measurements show several double-diffusive staircases (Rρ = 1.6 to 6) in the deep water, separated vertically by smooth high-gradient regions with much larger density ratios. The lowest staircase contains steps that are laterally coherent on the basin scale and have a well-defined vertical structure. On average, temperature steps in this staircase are 4 mK, salinity steps are 2 mg kg−1, and mixed layer heights are 70 cm. The CTD is capable of measuring bulk characteristics of the staircase in both temperature and salinity. Microstructure measurements are limited to temperature alone, but resolve the maximum temperature gradients in the center of selected laminar interfaces. Two different algorithms for characterizing the staircase are compared. Consistent estimates of the steady-state heat flux (27 mW m−2) are obtained from measurements above and below the staircase, as well as from microstructure measurements in the center of smooth interfaces. Estimates obtained from bulk interface gradients underestimate the steady-state flux by nearly a factor of 2. The mean flux calculated using a standard 4/3 flux law parameterization agrees well with the independent estimates, but inconsistencies between the parameterization and the observations remain. These inconsistencies are examined by comparing the underlying scaling relationship to the measurements.
Scheifele, B.; Pawlowicz, R.; Sommer, T.; Wüest, A. (2014) Double diffusion in saline Powell Lake, British Columbia, Journal of Physical Oceanography, 44(11), 2893-2908, doi:10.1175/JPO-D-14-0070.1, Institutional Repository

Double-diffusive interfaces in Lake Kivu reproduced by direct numerical simulations

Double diffusion transforms uniform background gradients of temperature and salinity into “staircases” of homogeneous mixed layers that are separated by high-gradient interfaces. Direct numerical simulations (DNS) and microstructure measurements are two independent methods of estimating double-diffusive fluxes. By performing DNS under similar conditions as found in our measurements in Lake Kivu, we are able to compare results from both methods for the first time. We find that (i) the DNS reproduces the measured interface thicknesses of in situ microstructure profiles, (ii) molecular heat fluxes through interfaces capture the total vertical heat fluxes for density ratios larger than three, and (iii) the commonly used heat flux parameterization underestimates the total fluxes by a factor of 1.3 to 2.2.
Sommer, T.; Carpenter, J. R.; Wüest, A. (2014) Double-diffusive interfaces in Lake Kivu reproduced by direct numerical simulations, Geophysical Research Letters, 41(14), 5114-5121, doi:10.1002/2014GL060716, Institutional Repository

Interface structure and flux laws in a natural double-diffusive layering

The diffusive regime of double-diffusive convection generates staircases consisting of thin high-gradient interfaces sandwiched between convectively mixed layers. Simultaneous microstructure measurements of both temperature and conductivity from the staircases in Lake Kivu are used to test flux laws and theoretical models for double diffusion. Density ratios in Lake Kivu are between one and ten and mixed layer thicknesses on average 0.7 m. The larger interface thickness of temperature (average 9 cm) compared to dissolved substances (6 cm) confirms the boundary-layer structure of the interface. Our observations suggest that the boundary-layer break-off cannot be characterized by a single critical boundary-layer Rayleigh number, but occurs within a range of O(102) to O(104). Heat flux parameterizations which assume that the Nusselt number follows a power law increase with the Rayleigh number Ra are tested for their exponent η. In contrast to the standard estimate η = 1/3, we found η = 0.20 ± 0.03 for density ratios between two and six. Therefore, we suggest a correction of heat flux estimates which are based on η = 1/3. The magnitude of the correction depends on Ra in the system of interest. For Lake Kivu (average heat flux 0.10 W m−2) with Ra = O(108), corrections are marginal. In the Arctic Ocean with Ra = O(108) to O(1012), however, heat fluxes can be overestimated by a factor of four.
Sommer, T.; Carpenter, J. R.; Schmid, M.; Lueck, R. G.; Schurter, M.; Wüest, A. (2013) Interface structure and flux laws in a natural double-diffusive layering, Journal of Geophysical Research C: Oceans, 118(11), 6092-6106, doi:10.1002/2013JC009166, Institutional Repository

Revisiting microstructure sensor responses with implications for double-diffusive fluxes

Thin high-gradient interfaces that occur naturally within double-diffusive staircases are used to estimate the response characteristics of temperature and conductivity microstructure sensors. The knowledge of these responses is essential for resolving small-scale turbulence in natural water bodies and for determining double-diffusive fluxes of heat and salt. Here, the authors derive microstructure sensor responses from observed differences in the statistical distributions of interface thicknesses at various profiling speeds in Lake Kivu (central Africa). In contrast to the standard approach for determining sensor responses, this method is independent of any knowledge of the true in situ temperature and salinity structure. Assuming double-pole frequency response functions, the time constants for the Sea-Bird Electronics SBE-7 conductivity sensor and the Rockland Scientific International FP07 thermistor are estimated to be 2.2 and 10 ms, respectively. In contrast to previous assumptions, the frequency response for the SBE-7 is found to be substantial and dominates the wavenumber response for profiling speeds larger than 0.19 m s−1.
Sommer, T.; Carpenter, J. R.; Schmid, M.; Lueck, R. G.; Wüest, A. (2013) Revisiting microstructure sensor responses with implications for double-diffusive fluxes, Journal of Atmospheric and Oceanic Technology, 30(8), 1907-1923, doi:10.1175/JTECH-D-12-00272.1, Institutional Repository

Simulations of a double-diffusive interface in the diffusive convection regime

Three-dimensional direct numerical simulations are performed that give us an in-depth account of the evolution and structure of the double-diffusive interface. We examine the diffusive convection regime, which, in the oceanographically relevant case, consists of relatively cold fresh water above warm salty water. A 'double-boundary-layer' structure is found in all of the simulations, in which the temperature (T) interface has a greater thickness than the salinity (S) interface. Therefore, thin gravitationally unstable boundary layers are maintained at the edges of the diffusive interface. The TS-interface thickness ratio is found to scale with the diffusivity ratio in a consistent manner once the shear across the boundary layers is accounted for. The turbulence present in the mixed layers is not able to penetrate the stable stratification of the interface core, and the TS-fluxes through the core are given by their molecular diffusion values. Interface growth in time is found to be determined by molecular diffusion of the S-interface, in agreement with a previous theory. The stability of the boundary layers is also considered, where we find boundary layer Rayleigh numbers that are an order of magnitude lower than previously assumed.
Carpenter, J. R.; Sommer, T.; Wüest, A. (2012) Simulations of a double-diffusive interface in the diffusive convection regime, Journal of Fluid Mechanics, 711, 411-436, doi:10.1017/jfm.2012.399, Institutional Repository

Stability of a double-diffusive interface in the diffusive convection regime

In this paper, the authors explore the conditions under which a double-diffusive interface may become unstable. Focus is placed on the case of a cold, freshwater layer above a warm, salty layer [i.e., the diffusive convection (DC) regime]. The “diffusive interface” between these layers will develop gravitationally unstable boundary layers due to the more rapid diffusion of heat (the destabilizing component) relative to salt. Previous studies have assumed that a purely convective-type instability of these boundary layers is what drives convection in this system and that this may be parameterized by a boundary layer Rayleigh number. The authors test this theory by conducting both a linear stability analysis and direct numerical simulations of a diffusive interface. Their linear stability analysis reveals that the transition to instability always occurs as an oscillating diffusive convection mode and at boundary layer Rayleigh numbers much smaller than previously thought. However, these findings are based on making a quasi-steady assumption for the growth of the interfaces by molecular diffusion. When diffusing interfaces are modeled (using direct numerical simulations), the authors observe that the time dependence is significant in determining the instability of the boundary layers and that the breakdown is due to a purely convective-type instability. Their findings therefore demonstrate that the relevant instability in a DC staircase is purely convective.
Carpenter, J. R.; Sommer, T.; Wüest, A. (2012) Stability of a double-diffusive interface in the diffusive convection regime, Journal of Physical Oceanography, 42(5), 840-854, doi:10.1175/JPO-D-11-0118.1, Institutional Repository

Diffusive-type of double diffusion in lakes - a review

This chapter is a contribution in honour of Gerhard H Jirka, who has been fascinated by the amazing variety of small-scale structures that nature surprises us with, particularly in stratified natural waters. Here, we focus on the diffusive regime of double-diffusive convection that occasionally occurs in lakes. Preconditions are a permanent stratification by dissolved constituents - such as salinity and carbon-dioxide – and convective forcing by deep sources of heat. After setting the stage for double diffusion to occur, possible genesis processes of the diffusive regime are reviewed by explaining specific examples of this unusual stratification such as (i) the flushing of fossil seawater by freshwater, (ii) the solar-pond phenomenon in ice-covered lakes in Antarctica, and (iii) the deep release of salt and gases in volcanic regions. In particular, the two most prominent examples of natural waters in which double diffusion occurs, Lakes Nyos and Kivu, are reviewed in more detail. The generation and evolution of staircase layering are discussed in relation to experiences gained from laboratory experiments, DNS modelling, and analysis of data from natural waters.
Wüest, A.; Sommer, T.; Schmid, M.; Carpenter, J. R. (2012) Diffusive-type of double diffusion in lakes - a review, In: Rodi, W.; Uhlmann, M. (Eds.), Environmental fluid mechanics. Memorial volume in honour of Prof. Gerhard H. Jirka, 271-284, doi:10.1201/b12283-18, Institutional Repository

Double-diffusive convection in Lake Kivu

Double-diffusive staircases with a total of 230–350 mixed layers and sharp interfaces were observed in nine microstructure temperature profiles measured during February 2004 in Lake Kivu. The presence of these staircases at depths > 120 m indicates that diapycnal turbulent mixing is weak and vertical diffusive transport is dominated by double diffusion. Contrary to previously investigated natural or laboratory double-diffusive systems, the dissolved gases CO2 and CH4 contribute significantly to the density stratification, thereby influencing the formation and the structure of the staircases. The density ratio (i.e., the ratio of the stabilizing effect of dissolved substances to the destabilizing effect of temperature) ranges between 2.0 and 4.5 in large sections of the deep waters, implying a high susceptibility to the formation of staircases. The mixed layers (average thickness 0.48 m) are shown to be in a state of active convection. The average thickness of the interfaces (0.18 m) is surprisingly constant and independent of the large-scale stratification. The vertical heat fluxes correlate well with the temperature steps across the interfaces. Lake Kivu receives inflows from subaquatic springs at several depths that maintain the large-scale structure of the density stratification and disturb the staircases. In comparison to earlier observations from 1972, the double-diffusive heat fluxes appear to have been reduced, leading to a heat accumulation in the deep waters. Conversely, the strengthening of the main chemocline indicates an increased discharge of the subaquatic springs that could be responsible for recent changes in the nutrient cycling and methane production in the lake.
Schmid, M.; Busbridge, M.; Wüest, A. (2010) Double-diffusive convection in Lake Kivu, Limnology and Oceanography, 55(1), 225-238, doi:10.4319/lo.2010.55.1.0225, Institutional Repository

Lake sediments and sub-aquatic sources


In the project “Lake Kivu: Learning from the past for managing its future”,  we investigated the following topics:
 

  • The lake floor of the northern part of the basin was mapped using a side scan sonar. The resulting map showed old shorelines, which indicated that the lake surface was at least 300 m below its present level about 15’000 years ago. A large number of volcanic structures were observed on the lake floor, which had been formed during or after the lake level rise.
     
  • An analysis of sediment cores indicated a cyclic behaviour of the lake, where long periods of permanent stable stratification were occasionally interrupted by mixing events that were probably caused by volcanic activity. It remains unclear whether these mixing events could also have caused gas eruptions from the lake.
     
  • Several subaquatic groundwater sources feeding the lake were located and the inflowing waters were characterized by sampling at the inflow locations. In agreement with previous predictions from lake modelling, two different types of groundwater sources were found: cool and fresh sources, which are probably fed by rainwater infiltrating in the volcanic ground, and warm and salty sources, which are fed by one or several hydrothermal systems.
     
  • The water balance of the lake was assessed, using a simple modelling approach. The model showed that the observed lake level variations could be largely explained by variations in precipitation. The results also highlighted the lack of recent meteorological and hydrological data from the region.
     

This project was financed by the Swiss National Science Foundation and the Swiss Agency for Development and Cooperation. The project was done in collaboration with the Institut Supérieur de Bukavu (DR Congo) and the Kigali Institute of Science and Technology (now University of Rwanda). Besides the scientific aims, it also aimed at improving the local research capacities in both countries, the Democratic Republic of the Congo and Rwanda.
 


This figure is a schematic representation of the subaquatic springs feeding Lake Kivu. Fresher and less saline springs, probably fed by rainwater infiltrating to the north of the lake, enter the lake along its northern shore as indicated by the white discs. Warm saline groundwater springs enter volcanically active region in the norhteastern part of the lake. Their water is denser than the lake water and therefore flows downward following the topography and feeds the deepest part of the lake (Figure from Methane Extraction from Lake Kivu – Scientific background)
 

Publications

Characterisation of the subaquatic groundwater discharge that maintains the permanent stratification within Lake Kivu; East Africa

Warm and cold subaquatic groundwater discharge into Lake Kivu forms the large-scale density gradients presently observed in the lake. This structure is pertinent to maintaining the stratification that locks the high volume of gases in the deepwater. Our research presents the first characterisation of these inflows. Temperature and conductivity profiling was conducted from January 2010 to March 2013 to map the locations of groundwater discharge. Water samples were obtained within the lake at the locations of the greatest temperature anomalies observed from the background lake-profile. The isotopic and chemical signatures of the groundwater were applied to assess how these inflows contribute to the overall stratification. It is inferred that Lake Kivu’s deepwater has not been completely recharged by the groundwater inflows since its turnover that is speculated to have occurred within the last ~1000 yrs. Given a recent salinity increase in the lake constrained to within months of seismic activity measured beneath the basin, it is plausible that increased hydrothermal-groundwater inflows into the deep basin are correlated with episodic geologic events. These results invalidate the simple two-component end-member mixing regime that has been postulated up to now, and indicate the importance of monitoring this potentially explosive lake.
Ross, K. A.; Gashugi, E.; Gafasi, A.; Wüest, A.; Schmid, M. (2015) Characterisation of the subaquatic groundwater discharge that maintains the permanent stratification within Lake Kivu; East Africa, PLoS One, 10(3), e0121217 (21 pp.), doi:10.1371/journal.pone.0121217, Institutional Repository

Lake-level rise in the late Pleistocene and active subaquatic volcanism since the Holocene in Lake Kivu, East African Rift

The history of Lake Kivu is strongly linked to the activity of the Virunga volcanoes. Subaerial and subaquatic volcanoes, in addition to lake-level changes, shape the subaquatic morphologic and structural features in Lake Kivu's Main Basin. Previous studies revealed that volcanic eruptions blocked the former outlet of the lake to the north in the late Pleistocene, leading to a substantial rise in the lake level and subsequently the present-day thermohaline stratification. Additional studies have speculated that volcanic and seismic activities threaten to trigger a catastrophic release of the large amount of gases dissolved in the lake. The current study presents a bathymetric mapping and seismic profiling survey that covers the volcanically active area of the Main Basin at a resolution that is unprecedented for Lake Kivu. New geomorphologic features identified on the lake floor can accurately describe related lake-floor processes for the first time. The late Pleistocene lowstand is observed at 425 m depth, and volcanic cones, tuff rings, and lava flows observed above this level indicate both subaerial and subaquatic volcanic activities during the Holocene. The geomorphologic analysis yields new implications on the geologic processes that have shaped Lake Kivu's basin, and the presence of young volcanic features can be linked to the possibility of a lake overturn.
Ross, K. A.; Smets, B.; De Batist, M.; Hilbe, M.; Schmid, M.; Anselmetti, F. S. (2014) Lake-level rise in the late Pleistocene and active subaquatic volcanism since the Holocene in Lake Kivu, East African Rift, Geomorphology, 221, 274-285, doi:10.1016/j.geomorph.2014.05.010, Institutional Repository

Modelling Lake Kivu water level variations over the last seven decades

This study aimed at analysing the hydrological changes in the Lake Kivu Basin over the last seven decades with focus on the response of the lake water level to meteorological factors and hydropower dam construction. Historical precipitation and lake water levels were acquired from literature, local agencies and from global databases in order to compile a coherent dataset. The net lake inflow was modelled using a soil water balance model and the water levels were reconstructed using a parsimonious lake water balance model. The soil water balance shows that 370 mm yr−1 (25%) of the precipitation in the catchment contributes to the runoff and baseflow whereas 1100 mm yr−1 (75%) contributes to the evapotranspiration. A review of the lake water balance resulted in the following estimates of hydrological contributions: 55%, 25%, and 20% of the overall inputs from precipitation, surface inflows, and subaquatic groundwater discharge, respectively. The overall losses were 58% and 42% for lake surface evaporation and outflow discharge, respectively. The hydrological model used indicated a remarkable sensitivity of the lake water levels to hydrometeorological variability up to 1977, when the outflow bed was artificially widened.
Muvundja, F. A.; Wüest, A.; Isumbisho, M.; Kaningini, M. B.; Pasche, N.; Rinta, P.; Schmid, M. (2014) Modelling Lake Kivu water level variations over the last seven decades, Limnologica, 47, 21-33, doi:10.1016/j.limno.2014.02.003, Institutional Repository

Local conditions structure unique archaeal communities in the anoxic sediments of meromictic Lake Kivu

Meromictic Lake Kivu is renowned for its enormous quantity of methane dissolved in the hypolimnion. The methane is primarily of biological origin, and its concentration has been increasing in the past half-century. Insight into the origin of methane production in Lake Kivu has become relevant with the recent commercial extraction of methane from the hypolimnion. This study provides the first culture-independent approach to identifying the archaeal communities present in Lake Kivu sediments at the sediment-water interface. Terminal restriction fragment length polymorphism analysis suggests considerable heterogeneity in the archaeal community composition at varying sample locations. This diversity reflects changes in the geochemical conditions in the sediment and the overlying water, which are an effect of local groundwater inflows. A more in-depth look at the archaeal community composition by clone library analysis revealed diverse phylogenies of Euryarchaeota and Crenarachaeota. Many of the sequences in the clone libraries belonged to globally distributed archaeal clades such as the rice cluster V and Lake Dagow sediment environmental clusters. Several of the determined clades were previously thought to be rare among freshwater sediment Archaea (e.g., sequences related to the SAGMEG-1 clade). Surprisingly, there was no observed relation of clones to known hydrogentrophic methanogens and less than 2 % of clones were related to acetoclastic methanogens. The local variability, diversity, and novelty of the archaeal community structure in Lake Kivu should be considered when making assumptions on the biogeochemical functioning of its sediments.
Bhattarai, S.; Ross, K. A.; Schmid, M.; Anselmetti, F. S.; Bürgmann, H. (2012) Local conditions structure unique archaeal communities in the anoxic sediments of meromictic Lake Kivu, Microbial Ecology, 64(2), 291-310, doi:10.1007/s00248-012-0034-x, Institutional Repository

Nutrient cycling and methane production

This project examined the cycles of the main nutrients phosphorus, nitrogen, and silica in Lake Kivu. In particular, the external nutrient sources, their internal fluxes and the export to the sediment were quantified. The study showed that the internal recycling is by far the most important source of the limiting nutrients nitrogen and phosphorus for the biological production in the surface layer of the lake.

Furthermore, the production and consumption of methane in the lake was examined. A previous study had proposed that methane concentrations in the lake may have increased by up to 15% within only 30 years. The present project yielded further indications for increasing methane concentrations.

However, it also showed that the previously estimated 15% increase within 30 years was at the upper end of what could have been supported by the carbon cycling within the lake. An increased upward flux of nutrients due to increased subaquatic groundwater discharge into the lake was estimated to be a possible cause for the increased methane production. However, a follow-up study (see below) then indciated that no significant increase in methane concentrations had occurred since the 1970s.

This project was done in collaboration with the Institut Supérieur de Bukavu (DR Congo) and the National University of Rwanda. It was financed by the Swiss National Science Foundation and the Swiss Agency for Development and Cooperation. Besides the scientific research, it also supported the build-up of research capacities in Rwanda and Congo and provided a continuous scientific exchange. This was an important base for developing the present monitoring of the lake during methane extraction activities by the Lake Kivu Monitoring Program.

Figure from Methane Extraction from Lake Kivu – Scientific background
 

Publications

Modelling Lake Kivu water level variations over the last seven decades

This study aimed at analysing the hydrological changes in the Lake Kivu Basin over the last seven decades with focus on the response of the lake water level to meteorological factors and hydropower dam construction. Historical precipitation and lake water levels were acquired from literature, local agencies and from global databases in order to compile a coherent dataset. The net lake inflow was modelled using a soil water balance model and the water levels were reconstructed using a parsimonious lake water balance model. The soil water balance shows that 370 mm yr−1 (25%) of the precipitation in the catchment contributes to the runoff and baseflow whereas 1100 mm yr−1 (75%) contributes to the evapotranspiration. A review of the lake water balance resulted in the following estimates of hydrological contributions: 55%, 25%, and 20% of the overall inputs from precipitation, surface inflows, and subaquatic groundwater discharge, respectively. The overall losses were 58% and 42% for lake surface evaporation and outflow discharge, respectively. The hydrological model used indicated a remarkable sensitivity of the lake water levels to hydrometeorological variability up to 1977, when the outflow bed was artificially widened.
Muvundja, F. A.; Wüest, A.; Isumbisho, M.; Kaningini, M. B.; Pasche, N.; Rinta, P.; Schmid, M. (2014) Modelling Lake Kivu water level variations over the last seven decades, Limnologica, 47, 21-33, doi:10.1016/j.limno.2014.02.003, Institutional Repository

Methane sources and sinks in Lake Kivu

Unique worldwide, Lake Kivu stores enormous amounts of CH4 and CO2. A recent study reported that CH4 concentrations in the lake have increased by up to 15% in the last 30 years and that accumulation at this rate could lead to catastrophic outgassing by ∼2100. This study investigates the present-day CH4 formation and oxidation in Lake Kivu. Analyses of 14C and 13C in CH4 and potential carbon sources revealed that below 260 m, an unusually high ∼65% of the CH4 originates either from reduction of geogenic CO2 with mostly geogenic H2 or from direct inflows of geogenic CH4. Aerobic CH4 oxidation, performed by close relatives of type X CH4-oxidizing bacteria, is the main process preventing CH4 from escaping to the atmosphere. Anaerobic CH4 oxidation, carried out by CH4-oxidizing archaea in the SO42−-reducing zone, was also detected but is limited by the availability of sulfate. Changes in 14CCH4 and 13CCH4 since the 1970s suggest that the amount of CH4 produced from degrading organic material has increased due to higher accumulation of organic matter. This, as well as the sudden onset of carbonates in the 1960s, has previously been explained by three environmental changes: (1) introduction of nonnative fish, (2) amplified subaquatic inflows following hydrological changes, and (3) increased external inputs due to the fast growing population. The resulting enhancement of primary production and organic matter sedimentation likely caused CH4 to increase. However, given the large proportion of old CH4 carbon, we cannot exclude an increased inflow of geogenic H2 or CH4.
Pasche, N.; Schmid, M.; Vazquez, F.; Schubert, C. J.; Wüest, A.; Kessler, J. D.; Pack, M. A.; Reeburgh, W. S.; Bürgmann, H. (2011) Methane sources and sinks in Lake Kivu, Journal of Geophysical Research G: Biogeosciences, 116(G3), G03006 (16 pp.), doi:10.1029/2011JG001690, Institutional Repository

Abrupt onset of carbonate deposition in Lake Kivu during the 1960s: response to recent environmental changes

This study interprets the recent history of Lake Kivu, a tropical lake in the East African Rift Valley. The current gross sedimentation was characterized from a moored sediment trap array deployed over 2 years. The past net sedimentation was investigated with three short cores from two different basins. Diatom assemblages from cores were interpreted as reflecting changes in mixing depth, surface salinity and nutrient availability. The contemporary sediment trap data indicate seasonal variability, governed by diatom blooms during the annual mixing in the dry season, similar to Lakes Malawi and Tanganyika. The ratio of settling fluxes to net sediment accumulation rates implies mineralization rates of 80–90% at the sediment-water interface. The sediment cores revealed an abrupt change ~40 years ago, when carbonate precipitation started. Since the 1960s, deep-water methane concentrations, nutrient fluxes and soil mineral inputs have increased considerably and diatom assemblages have altered. These modifications probably resulted from a combination of three factors, commonly altering lake systems: the introduction of a non-native fish species, eutrophication, and hydrological changes inducing greater upwelling. Both the fish introduction and increased rainfall occurred at the time when the onset of carbonate precipitation was observed, whereas catchment population growth accompanied by intensified land use increased the flux of soil minerals already since the early twentieth century due to more intense erosion.
Pasche, N.; Alunga, G.; Mills, K.; Muvundja, F.; Ryves, D. B.; Schurter, M.; Wehrli, B.; Schmid, M. (2010) Abrupt onset of carbonate deposition in Lake Kivu during the 1960s: response to recent environmental changes, Journal of Paleolimnology, 44(4), 931-946, doi:10.1007/s10933-010-9465-x, Institutional Repository

Physical and biogeochemical limits to internal nutrient loading of meromictic Lake Kivu

Lake Kivu is one of the large African Rift lakes situated between the Democratic Republic of the Congo and Rwanda. In its permanently stratified hypolimnion, unusually high methane concentrations have increased further in recent decades. Because methanogenesis is, in part, dependent on supply of organic material from the photic zone, it is necessary to quantify upward nutrient fluxes from the saline, nutrient-rich deep waters. These upward fluxes are mainly driven by advection caused by subaquatic springs. Biogenic calcite precipitation drives surface-water depletion and deep-water enrichment of Ca2+, Sr2+, and Ba2+. Methane is mainly oxidized aerobically at the redox interface at 60 m, with a small contribution of anaerobic methane oxidation. A subaquatic spring that sustains the major chemocline at 250 m depth was depleted of N, P, and CH4, and concentrations of major ions were slightly lower than in the lake water of the same depth. Enrichment of the deep waters with nutrients and CH4 are driven by mineralization of settling organic material, whereas SiO2 is influenced by uptake and mineralization of diatoms and inputs through subaquatic springs. Dissolved inorganic phosphorus and Si fluxes supplied by internal loading through upwelling were found to be lower than the estimations for Lakes Malawi and Tanganyika. In contrast, N flux was within the lower range for Lake Malawi, whereas it was assumed to be totally lost by denitrification in Lake Tanganyika. In Lake Kivu, nutrient uptake by primary production is three times higher than nutrient upward fluxes.
Pasche, N.; Dinkel, C.; Müller, B.; Schmid, M.; Wüest, A.; Wehrli, B. (2009) Physical and biogeochemical limits to internal nutrient loading of meromictic Lake Kivu, Limnology and Oceanography, 54(6), 1863-1873, doi:10.4319/lo.2009.54.6.1863, Institutional Repository

Balancing nutrient inputs to Lake Kivu

The primary production in meromictic Lake Kivu is sustained by external nutrient inputs and by internal loading due to upwelling caused by sub-aquatic sources. We present here the results of external loading of phosphorus (P), nitrogen (N) and silica (Si) by rivers and atmospheric deposition measured from 2006 to 2008. These external inputs are compared to internal loading. The input of soluble-reactive P (SRP), supplied in equal parts from rivers and atmospheric deposition, adds up to 230 t P yr–1, 20 times less than total P load. Ammonium (mainly via rainwater) and nitrate (mainly via rivers) are primary sources of the dissolved N load (5400 t N yr–1), with both species contributing ∼50%. Dissolved Si input (40,000 t Si yr–1) is unique in that only ∼60% enters by rivers, while the remaining ∼40% comes from sub-aquatic sources and atmospheric deposition is negligible. Based on the molar nutrient ratios, we identify P as the limiting factor for algae production. Despite the strong anthropogenic impact on the catchment and the high particle erosion (74 t km–2 yr–1), the area-specific nutrient mobilization is rather low. The external nutrient input is therefore not the cause for the reported increase of methane production in the last decades. External loading to the epilimnion plays a lesser role for all three nutrients (∼10% for SRP, ∼25% for dissolved N and ∼45% for dissolved Si), as compared to the lake-internal loading by upwelling (90%, 75% and 55%, respectively). Lake Kivu, therefore, is similar to other East African large lakes in that the internal loading exceeds the external loading. Despite the substantial uncertainty of the load estimates of up to 50%, we can conclude that the observed nutrient input is consistent with the primary production of 260 g C m–2 yr–1 recently measured by Sarmento et al. (2006) and also consistent with the lake-internal fluxes established by Pasche et al. (in press).
Muvundja, F. A.; Pasche, N.; Bugenyi, F. W. B.; Isumbisho, M.; Müller, B.; Namugize, J.-N.; Rinta, P.; Schmid, M.; Stierli, R.; Wüest, A. (2009) Balancing nutrient inputs to Lake Kivu, Journal of Great Lakes Research, 35(3), 406-418, doi:10.1016/j.jglr.2009.06.002, Institutional Repository

Managing Lake Kivu: moving from a steady-state to a dynamic modelling approach

This project aimed at updating the scientific background for the assessment of different strategies for methane extraction from Lake Kivu.

As part of the project, we adapted the portable field mass-spectrometer recently developed by the Environmental Isotopes research group at Eawag for vertical profiling of gas concentrations in deep lakes. With this instrument, we contributed to an international intercalibration campaign with the aim of updating the concentrations of dissolved methane and carbon dioxide in Lake Kivu. The results showed that there was no significant increase of gas concentrations since the measurements that were performed in the 1970s.

The same instrument was also used to measure a vertical profile of the concentrations of dissolved noble gases and their isotopic composition in the lake. The results showed a a striking lack of atmospheric noble gases as well as a distinct non-atmospheric isotopic signal in the deep waters. The most likely explanation for these observations is that concentrations of atmospheric noble gases are depleted in the groundwater sources feeding the lake.

Finally, we developed a new transient model for Lake Kivu that considers the dynamic properties of the lake and can account for densitiy-driven stratification of the groundwater inflows. For this purpose, we coupled our successful one-dimensional physical lake model Simstrat  to the Aquatic EcoDynamics Modelling Library AED2  of the University of Western Australia. The long-term simulations performed with this model showed that the previously postulated and subsequently partially detected groundwater inflows are fully sufficent to allow the emergence of the current density structure of the lake within a time frame of about 2000 years even from a completely mixed lake.

Publications

Dynamic modelling provides new insights into development and maintenance of Lake Kivu's density stratification

Lake Kivu is a 485 m deep, Central-East African rift lake with huge amounts of carbon dioxide and methane dissolved in its stably stratified deep waters. In view of future large-scale methane extraction, one-dimensional numerical modelling is an important and computationally inexpensive tool to analyze the evolution of stratification and the content of gases in Lake Kivu. For this purpose, we coupled the physical lake model Simstrat to the biogeochemical library AED2. Compared to an earlier modelling approach, this coupled approach offers several key improvements, most importantly the dynamic evaluation of mixing processes over the whole water column, including a parameterization for double-diffusive transport, and the density-dependent stratification of groundwater inflows. The coupled model successfully reproduces today's near steady-state of Lake Kivu, and we demonstrate that a complete mixing event ∼2000 years ago is compatible with today's physical and biogeochemical state.
Bärenbold, F.; Kipfer, R.; Schmid, M. (2022) Dynamic modelling provides new insights into development and maintenance of Lake Kivu's density stratification, Environmental Modelling and Software, 147, 105251 (15 pp.), doi:10.1016/j.envsoft.2021.105251, Institutional Repository
Schmid, M.; Bärenbold, F.; Wüest, A. (2021) Methane extraction from Lake Kivu. Scientific background, 15 p, Institutional Repository

No increasing risk of a limnic eruption at Lake Kivu: intercomparison study reveals gas concentrations close to steady state

Lake Kivu, East Africa, is well known for its huge reservoir of dissolved methane (CH4) and carbon dioxide (CO2) in the stratified deep waters (below 250 m). The methane concentrations of up to ~ 20 mmol/l are sufficiently high for commercial gas extraction and power production. In view of the projected extraction capacity of up to several hundred MW in the next decades, reliable and accurate gas measurement techniques are required to closely monitor the evolution of gas concentrations. For this purpose, an intercomparison campaign for dissolved gas measurements was planned and conducted in March 2018. The applied measurement techniques included on-site mass spectrometry of continuously pumped sample water, gas chromatography of in-situ filled gas bags, an in-situ membrane inlet laser spectrometer sensor and a prototype sensor for total dissolved gas pressure (TDGP). We present the results of three datasets for CH4, two for CO2 and one for TDGP. The resulting methane profiles show a good agreement within a range of around 5-10% in the deep water. We also observe that TDGP measurements in the deep waters are systematically around 5 to 10% lower than TDGP computed from gas concentrations. Part of this difference may be attributed to the non-trivial conversion of concentration to partial pressure in gas-rich Lake Kivu. When comparing our data to past measurements, we cannot verify the previously suggested increase in methane concentrations since 1974. We therefore conclude that the methane and carbon dioxide concentrations in Lake Kivu are currently close to a steady state.
Bärenbold, F.; Boehrer, B.; Grilli, R.; Mugisha, A.; von Tümpling, W.; Umutoni, A.; Schmid, M. (2020) No increasing risk of a limnic eruption at Lake Kivu: intercomparison study reveals gas concentrations close to steady state, PLoS One, 15(8), e0237836 (14 pp.), doi:10.1371/journal.pone.0237836, Institutional Repository

Missing atmospheric noble gases in a large, tropical lake: the case of Lake Kivu, East-Africa

Lake Kivu is a 485 m deep tropical rift lake in East-Africa and well-known for its very high concentrations of dissolved carbon dioxide and methane in the stratified deep waters. In view of future large-scale methane extraction for power production, there is a need for predicting the evolution of gas concentrations and lake stability using numerical modelling. However, knowledge about the geochemical origin and transport processes affecting dissolved gases in the lake is still partially missing. Due to their inert nature, the analysis of dissolved noble gases can help to shed light on such questions. To learn more about transport processes in Lake Kivu, we extended a well-established sampling method for dissolved noble gases to work in the lake's high gas pressure waters. The results of our analysis show a distinct non-atmospheric isotopic signal in the deep waters (below 250 m) with 3He/4He and 40Ar/36Ar ratios ~250% and ~20% higher than air saturated water (ASW). Moreover, the gas concentration profiles reveal a striking lack of atmospheric noble gases in the deep waters with respect to ASW. While Ne is depleted by ~45%, the more soluble 36Ar and Kr even decrease by ~70%. In contrast, 4He concentrations increase with depth by a factor of up to ~600. We attribute this excess He and the increases in 3He/4He and 40Ar/36Ar to the inflow of magmatic gases into Lake Kivu, along with a significant contribution of radiogenic 4He. To explain the depletion of atmospheric noble gases, we present and discuss three different scenarios, namely continuous outgassing, the inflow of depleted groundwater and a large past outgassing event. Due to the best agreement with our observations, we conclude that the inflow of depleted groundwater is likely responsible for the observed atmospheric noble gas depletions.
Bärenbold, F.; Schmid, M.; Brennwald, M. S.; Kipfer, R. (2020) Missing atmospheric noble gases in a large, tropical lake: the case of Lake Kivu, East-Africa, Chemical Geology, 532, 119374 (9 pp.), doi:10.1016/j.chemgeo.2019.119374, Institutional Repository

Intercalibration campaign for gas concentration measurements in Lake Kivu

1.The 2018 intercalibration campaign aimed at quantifying the methane (CH4) and carbon dioxide (CO2) gas content and the recharge rate of CH4 in Lake Kivu using a range of different measurement methods. Measurements were performed by research teams from the Helmholtz Centre for Environmental Research (UFZ) in Magdeburg (Germany), the Swiss Federal Institute for Aquatic Science and Technology (Eawag, Switzerland), the French National Centre for Scientific Research (CNRS) in Grenoble (France), and from KivuWatt Ltd (Kigali, Rwanda).
2. The following measurement methods were applied: two sensors for the in-situ observation of total dissolved gas pressure; two sensors for the in-situ observation of the partial pressure of dissolved CH4; and two methods for quantifying the concentrations of dissolved CH4 and CO2 in samples retrieved from the lake either using sampling bags or a tubing system. These methods were specifically customized for the application under the special conditions in Lake Kivu. Since some of the methods quantify partial pressures of CH4 and/or CO2, and other methods quantify their concentrations, a procedure for converting between partial pressures and concentrations was developed and implemented.
3. The observations yielded a consistent picture of the vertical profiles of dissolved concentrations of CH4 and CO2 as well as the total gas pressures in Lake Kivu. The observed variability between the datasets is related to the limited accuracy of the different measurement methods.
4. The observed CH4 concentrations were within the range of previous observations. However, in the resource zone (below 260 m depth), they were approximately 5-20 % below the concentrations measured by M. Halbwachs and J.-C. Tochon in 2003, which had previously been used as the standard for estimating the CH4 content in the lake.
5. The CH4 content in the resource zone (between 260 and 480 m depth) was estimated to ~40 km3 STP (volume of gas at a temperature of 0°C and a pressure of 1 atm) (range, 36.4 - 42.2 km3). These numbers are somewhat lower than the previous estimate of 44.7 km3 by Wüest and Schmid (2012) based on the measurements of M. Halbwachs and J.-C. Tochon in 2003. The CH4 content in the potential resource zone (200-260 m) was estimated to range between 8.2 and 8.6 km3 for all methods, which agrees with the previous estimate of 8.5 km3. The whole-lake CO2 content was estimated to 285 km3 STP.
6. The differences in CH4 concentrations and content compared to previous estimates are due to the limited accuracy of the different measurement methods and were not caused by the comparably small amount of CH4 removed by the past and ongoing gas extraction operations.
7. The 2018 measurements do not confirm the previous hypothesis that the CH4 concentrations were increasing during the last decades in Lake Kivu. They rather indicate approximately constant concentrations since the first observations in the 1950's within the uncertainty range of the present and previous measurements.
8. Recommendations for monitoring the evolution of gas concentrations in the lake cover two important aspects. Required monitoring equipment and frequency are defined based on different monitoring purposes. In addition, a need for building up sampling routines and skills is identified.
Schmid, M.; Bärenbold, F.; Boehrer, B.; Darchambeau, F.; Grilli, R.; Triest, J.; von Tümpling, W. (2019) Intercalibration campaign for gas concentration measurements in Lake Kivu, 64 p, Institutional Repository

Methane Harvesting

Management for the safe extraction of methane from Lake Kivu

The Democratic Republic of the Congo and Rwanda decided to reduce the risk of a sudden eruption of the gases dissolved in Lake Kivu. This is intended to be done by extracting the gases from the deep water of the lake and using the methane for power production in a safe, environmentally sustainable, and economically beneficial way. A group of experts, including Eawag representatives, has developed management prescriptions for the safe extraction of methane from Lake Kivu.

A first pilot plant, Kibuye Power 1 (KP1) has started extracting methane from the lake in January 2009. A larger plant with an installed capacity of 26 MW by KivuWatt/ContourGlobal has started operation in December 2015. Several further projects for methane extraction are under development in both countries.

In Rwanda, the Lake Kivu Monitoring Program was established in 2008. It is responsible for monitoring the methane to avoid any impacts on the lake and on the safety of the riparian population. Since 2015, the LKMP is supported by an international expert advisory group with the participation of Eawag representatives. An important next step would be the build-up of a binational authority involving both Rwanda and the DR Congo for the joint management of methane extraction and lake monitoring.

Documents to download:

  • Intercalibration campaign for gas concentration measurements in Lake Kivu, Report, 2019, M. Schmid, F. Bärenbold, B. Boehrer, F. Darchambeau, R. Grilli, J. Triest, W. von Tümpling [pdf, 1.74MB]
  • Modelling the reinjection of deep-water after methane extraction in Lake Kivu, Report, December 2009, A. Wüest, L. Jarc, M. Schmid [pdf, 2.52MB]
  • Management prescriptions for the development of Lake Kivu Gas Resources, Expert Working Group, June 2009 [pdf, 675KB]