Ebullition Regulated by Pressure Variations in a Boreal Pit Lake

Abstract

Methane ebullition from lakes is an important contributor to atmospheric greenhouse gases. However, ebullition is typically sampled at intervals greater than the duration of ebullition events, limiting our understanding of the factors controlling this flux. Here, we present high-frequency ebullition data from a single site in a boreal pit lake during the open-water season between June 24 and 21 October 2018. We record ebullition every 30 min for the first 2 months, and then every minute for the next 2 months. During the 4-month period, 24 ebullition events were recorded. These events generally lasted 2–4 days in response to low atmospheric pressure systems. The peaks in ebullition corresponded to troughs in atmospheric pressure. We provide empirical equations that incorporate a pressure threshold to model the time-series of ebullition events. Minor and gradual variations in mud temperature had no apparent effect on the observed ebullition events.

Exploring the temporal dynamics of methane ebullition in a subtropical freshwater reservoir

“…On overall, the models from previous studies failed in predicting the accumulated fluxes (Rel error from −346.9% to 289.5%, S1 Table ). The model proposed by [ 31 ] based on total pressure could estimate the accumulated fluxes with an underestimation of 10%, however the model parameters were site specific (to Passaúna), which explain the improved model performance ( Table 1 ). Nonetheless, the range of relative errors obtained from the prediction of accumulated fluxes were lower (from −2.8% to 48.1%) for the new models implemented, which can result from the combination of the addition of new ebullition predictors, the model’s capability of capturing the short-term temporal dynamics of ebullition (and thus, reducing accumulated errors), and the tuning of model parameters to our data.…”

“…The characteristics of ebullition time-series of episodic/pulse events and being zero-inflated, in combination with the complex connection to various drivers, pose difficulties in reproducing the temporal dynamics of ebullition fluxes and its magnitude using empirical approaches. Nevertheless, previous studies could successfully reproduce the temporal dynamics of ebullition by considering only atmospheric pressure (R 2 = 0.87) [ 31 , 58 ], or a combination of more variables. For instance, [ 29 ] reproduced ebullition time series with an autoregressive model using sediment temperature, wind speed, change in atmospheric pressure with good agreement (R 2 = 0.86).…”

“…Statistical and data-driven models are less demanding in terms of input variables, however the quality and quantity of input data directly affect the model performance. These empirical models have been applied in previous studies for both testing the dependency of ebullition on diverse environmental variables and for estimating the ebullition flux, and considered temperature [ 18 , 25 ], chlorophyll-a [ 26 ], littoral area and radiance [ 27 ], sediment organic matter content [ 28 ], wind speed [ 29 ], pressure, and pressure changes [ 29 – 31 ]. Nevertheless, the applicability of these statistical and data-driven models across diverse systems and timescales for reproducing ebullition dynamics has rarely been addressed.…”