Economical, scalable measurements of soil respiration in eddy covariance footprints using forced diffusion

Background/Question/Methods

Eddy covariance (EC) systems have been widely used across the globe for more than 20 years, offering researchers invaluable measurements of parameters including Net Ecosystem Exchange and ecosystem respiration. However, evidence suggests that EC assumptions and technical obstacles may be causing biased gas exchange estimates. Measurements of soil respiration (R_S) at the ground level may help alleviate these biases; for example, by allowing researchers to reconcile problems with nocturnal EC flux data and provide a means to inform gap-filling models. R_S measurements have been used sparingly alongside EC towers because of the large cost to scale chamber systems to the EC footprint and data integration and processing burdens. Here we present the Forced Diffusion (FD) method for measurement of R_S at EC sites. The FD method allows for inexpensive and autonomous measurements, providing a scalable approach to matching the EC footprint compared to other R_S systems.

Results/Conclusions

Results from a pilot study at the Howland Forest AmeriFlux site (Maine) carried out in Spring 2016 using EC and FD chambers in tandem will be presented, with an emphasis on how R_S measurements can identify decoupling of above and below canopy air masses and assist in gap filling. Uncertainty in nocturnal EC fluxes has already been characterized at Howland Forest, in addition to R_S using automated chambers which have been used to inform gap-filling models at Howland Forest. This study has been designed to replicate these findings using the FD approach.

We will also present data collected using the FD chambers in Fall 2015 in southwestern Nova Scotia demonstrating methods for quality assurance and quality control (QA/QC) of data. For the QC metric, a simple Q₁₀ model where periods can be cross-compared is used. The range of CO₂ flux observed at this site was 0-3 mmol m^-2 s^-1 with a temperature range from 20 C to -5 C. We effectively modeled the data using only temperature to yield a Q₁₀ of 2.24. During periods where sensors were not performing due to insufficient power, the measurements produced a Q₁₀ below the ideal modeled value (Q₁₀ of 1.64). During this period, the Q₁₀ curve also fell outside the 95% error envelopes of the before and after brownout periods. This work provides a first-order QA/QC approach that allows the eosFD chamber measurements to be easily integrated with EC measurements, thereby lessening the burden of including R_S data for individual operators and EC networks.