Modelling methane emissions from lakes and reservoirs: case study of Lake Kinneret
MODSIM2021, 24th International Congress on Modelling and Simulation.
Aquatic systems are responsible for 6-16% of global emissions of methane, a potent greenhouse gas. Methane concentration in the atmosphere is rising continuously, prompting the need for a better understanding of freshwater methane sources and sinks. To date, the global methane budget is based on upscaling emissions from individual lakes resulting in erroneous freshwater methane emission estimates. Methane is produced in the anoxic sediments of lakes and is emitted to the atmosphere via
... pathways (Figure 1 ). These include diffusion, ebullition (bubbling), storage flux and advection through aquatic vegetation, all regulated by different physical, chemical, and biological factors. As a result, methane fluxes significantly vary both within and across systems. Due to its stochastic nature, ebullition is ever challenging to quantify and is generally disregarded from the global methane budget. To improve methane emission estimates from freshwater, process-based models can be used that enable the simulation of each pathway separately. The aim of this study is to simulate methane emissions from a focus site. This is done through the development of a methane module in the Aquatic Ecodynamics (AED) modelling library which is applied to Lake Kinneret. Whilst Lake Kinneret is a single focus of this study, applying the developed methane model to a site with previously existing monitoring and modelling data will contribute to the building of an open-source general methane model. In this study, the one-dimensional General Lake Model (GLM) was coupled to the AED modelling library to simulate the thermal dynamics and dissolved oxygen concentration of Lake Kinneret. The methane sources and sinks in the AED carbon module include diffusion, aerobic oxidation, and atmospheric gas exchange. To simulate ebullition, a new algorithm was added to the model. The ebullitive flux in this algorithm varies according to temperature and water level changes. The bubbles released from the sediments either dissolve in the water column or are directly emitted to the atmosphere. The vertical temperature profile of the lake was successfully reproduced, with a Root Mean Square Error (RMSE) of 1.5 °C (~5%). The model has appropriately resolved the oxygen dynamics, except for the metalimnetic oxygen minimum, with an RMSE of 2.99 mg/L (~20%). The RMSE between the observed and simulated methane concentrations was 56 mmol/m 3 (~10%). The great annual and within system variations of the methane fluxes are captured well by the model. However, surface methane concentrations are repeatedly overestimated during holomixis. Lastly, the model simulates the seasonality of ebullition well, however, it doesn't capture its stochastic nature reported in the literature. Process-based models have the ability to account for the variability in lake characteristics and capture the interannual and within system variations in methane emissions. However, to move away from site-specific ebullition models, it is crucial to develop new general ways to parametrise ebullition, enabling its applicability across a gradient of systems.