On October 6th, Eosense and the Lundholm Lab set up 4 eosAC automated soil flux chambers with a Los Gatos Research Ultraportable Greenhouse Gas Analyzer (UGGA) on the Atrium Green Roof at Saint Marys University in Halifax, Nova Scotia. Using the eosAC/UGGA system, carbon dioxide and methane fluxes were monitored during a week-long period on two treatment types, vascular plants and mosses, and with one replicate plot for each treatment. During the deployment we also happened to capture the remnant of hurricane Matthew as it passed, which came with high winds and significant rainfall.

Examples of the moss (left) and vascular plant treatments on the Atrium Green Roof at Saint Marys University.

Examples of the moss (left) and vascular plant treatments on the Atrium Green Roof at Saint Marys University.

Preliminary Results and Discussion

We pulled the data from the UGGA after a week of deployment and filtered the raw analyzer files by hand to remove effects caused by power flickers during the hurricane (mostly signified by widely variable pressure as measured in the UGGA optical cavity). Shown in the plot below are the smoothed time series data for both CO2 (lines, left y-axis) and CH4 (lines and symbols, right y-axis).

Smoothed time series data for both CO2 (lines, left y-axis) and CH4 (lines and symbols, right y-axis). Also shown on the right y-axis in the large, blue squares is the rainfall (dm) recorded by a nearby Environment Canada rain gauge.

While we did not see an apparent treatment effect (vascular plants are in black and blue, and mosses are in green and red) we did see a very obvious effect of the high rainfall on both CO2 and CH4. Fluxes of CO2 deceased by approximately two-thirds immediately following heavy rain and methane fluxes mainly increased from near-zero to a peak of approximately 0.1 nmol/m2/s. In one case, on a moss plot, methane fluxes actually became significantly negative suggesting an uptake of CH4 by the moss plot (although not replicated in the other moss plot). Interestingly we observed almost no correlation between temperature and CO2 efflux rate during the period of these observations, but a closer examination of the time series shows that temperature and CO2 emissions are significantly phase shifted from one another.

Carbon dioxide fluxes for all plots shown along side chamber temperatures (orange circles) recorded by the eosAC internal thermistors.

Using the statistical software package R a cross-correlation analysis was performed that suggests peak correlation between temperature and CO2 flux at approximately 9.87 hours; in other words, carbon dioxide flux lags peak temperature by 9.87 hours on the green roof. This effect is quite common where insulating plant layers (which trap stagnant air) and soil properties cause slow propagation of air temperature into the soil profile. Alternatively, if the majority of flux from the system is coming from plant respiration, the apparent temperature response can be lagged due to transport time of gases and photosynthesis products through the plant. When examining the two treatments separately, the vascular plant treatment had a lag approximately 1 hour shorter than the moss treatment, consistent with the moss forming an insulating layer of more stagnant air, thereby limiting thermal transfer. Still, the long lag between peak temperature is consistent with a parent material (soil) with a low thermal diffusivity or a significant plant respiratory component.

Methane fluxes (shown below) did not show any significant correlation with temperature, but showed a clear response to the rain events associated with the hurricane. The increase in methane production in three out of four chambers is expected as the soil wets and oxygen supply is decreased making the environment more favourable for methanogenic organisms.

Methane fluxes from the mosses (red and green) and the vascular plants (black and blue) with rainfall (blue squares) showing the strong response to high rainfall.

What is more interesting is the strong uptake in methane in one moss plot during the rainfall event. Because this green roof is a constructed environment it might be associated with water drainage patterns from the rooftop during the storms (water may have been pooling at another location but draining quickly here) causing a physical uptake of methane from the chamber headspace. Similarly it may have been a favourable environment for methanotrophs, and could represent an actual biological consumption of methane for a short period where conditions were favourable.

Next Steps

Overall the eosAC and LGR UGGA system performed well during the field deployment, offering almost no data loss (except a few missing points during power flickers) and week-long continuous measurements of CO2 and methane fluxes from the site. Eosense and the Lundholm lab will continue working on this data to see what else can be uncovered about the CO2 and methane dynamics at the site. Of particular interest in this data set is the long lag time between peak temperature and peak flux; which is not common in ecosystems that have small stature plant species. One avenue of potential future collaboration will be to deploy both transparent and opaque chamber on the Atrium Green Roof to get a more holistic picture of both the photosynthetic uptake of these greenhouse gases, as well as estimate of the soil to atmosphere fluxes in order to estimate the Net Ecosystem Exchange and get closer to estimating the annual C-budget of the green roof environment.

St. Marys University Atrium Green Roof

Automated CO2 and CH4 flux measurements with the eosAC and the ABB-LGR Microportable Greenhouse Gas Analyzer

Automated CO2 & CH4 flux measurements with the UGGA