BayCEER Workshop 2015

Analysis of Methane Emissions in a Subarctic Permafrost Region using Wavelet Transformation and Conditional Sampling

Introduction
Global warming will result in dramatic changes of Arctic ecosystems including permafrost wetlands. It benefits microorganisms decomposing the soil organic matter under formation of methane. Only little is known about ebullition, a process, where bubbles of methane were accumulated in the water-saturated layers of the permafrost, followed by a sudden release within minutes. Ebullition events might violate main requirements of eddy covariance, steady state conditions and horizontal homogeneity, resulting in a flux underestimation. The present study aims in resolving those peaks in flux.

Material and Methods
Measurements were conducted in a permafrost ecosystem south of Chersky, Sakha (Yakutia) Republic, Russia. Two eddy covariance towers were equipped with a closed-path gas analyser. The Mexican hat wavelet was used to detect the times of occurring events, while the Morlet wavelet provided information on contributing frequencies. Based on the wavelet flux with a timestep of 1 minute a test using the median absolute deviation was applied in order to detect peaks.

Results
The wavelet flux agreed very good to eddy covariance under best steady state and turbulence conditions, but conditional sampling was highly sensitive regarding the correct choice of the mean vertical wind speed. Surprisingly events mostly occurred during night time and always simultaneously at both towers. They were obviously triggered by gravity waves, low level jets or weather fronts passing the site and katabatic winds from a hill ridge nearby.

Conclusions
Wavelet analysis was proven to be a suitable method to resolve events in the order of minutes. In Chersky they were typically found at night time and not caused by ebullition or other local processes in the soil, but by different meteorological phenomenons on the mesoscale. The classical eddy covariance failed to resolve the events correctly. Normally the data would have been refused, but considering this study, a flux correction approach could now be developed.