A practical approach for uncertainty quantification of high‐frequency soil respiration using Forced Diffusion chambers

Abstract

This paper examines the sources of uncertainty for the Forced Diffusion (FD) chamber soil respiration (R s ) measurement technique and demonstrates a protocol for uncertainty quantification that could be appropriate with any soil flux technique. Here we sought to quantify and compare the three primary sources of uncertainty in R s : (1) instrumentation error; (2) scaling error, which stems from the spatial variability of R s ; and (3) random error, which arises from stochastic or unpredictable variation in environmental drivers and was quantified from repeated observations under a narrow temperature, moisture, and time range. In laboratory studies, we found that FD instrumentation error remained constant as R s increased. In field studies from five North American ecosystems, we found that as R s increased from winter to peak growing season, random error increased linearly with average flux by about 40% of average R s . Random error not only scales with soil flux but scales in a consistent way (same slope) across ecosystems. Scaling error, measured at one site, similarly increased linearly with average R s , by about 50% of average R s . Our findings are consistent with previous findings for both soil fluxes and eddy covariance fluxes across other northern temperate ecosystems that showed random error scales linearly with flux magnitude with a slope of~0.2. Although the mechanistic basis for this scaling of random error is unknown, it is suggestive of a broadly applicable rule for predicting flux random error. Also consistent with previous studies, we found the random error of FD follows a Laplace (double-exponential) rather than a normal (Gaussian) distribution.

The value of soil respiration measurements for interpreting and modeling terrestrial carbon cycling

“…This spatial variability makes measuring R S representatively and robustly, and using these measurements to compare and evaluate R ECO derived from EC, more challenging. One way researchers can help to identify the presence of hot spots and hot moments to other data users is to report the full probability distribution function (or at least the second and third moments-variance and skewness) of observed soil respiration in addition to site means (Lavoie et al 2014). Ideally, complete data sets that include full time series of raw data and all sampling locations would be made available in archived repositories, allowing users to define hot spots and hot moments using different algorithms, and re-analyze older data as new techniques or insights become available.…”

“…Hashimoto 2012), instantaneous R S measures from autochambers and survey campaigns have rarely been synthesized (Bahn et al 2010;Vargas et al 2010a;Lavoie et al 2014;Cueva et al 2015), and are not easily available to modelers because they are not in organized data repositories (see Vargas and Leon 2015 for an exception). These small but geographically widespread datasets are generally in the hands of individual investigators, and are part of what Dietze et al (2013) refers to as the 'long tail' of data.…”

Observations of 14CO2 in ecosystem respiration from a temperate deciduous forest in Northern Wisconsin

“…Periodic side‐by‐side comparisons between the FD system and a Licor‐8100 soil flux system showed that the FD was biased to lower measurements than the Licor (FD µmol m −2 s −1 = 0.74 × Licor µmol m −2 s −1 − 0.34, R 2 = 0.65, N = 19); however, under field conditions it is difficult to ascertain which system is more accurate. Under laboratory conditions, the instrumental error of the FD system has been shown to be lower than the Licor system [ Lavoie et al , ], but at low flux rates we expect that the FD system may have been less accurate than the Licor, based on observations of more high flux “spikes” in the FD time series and on the fact that the FD flux is based on a differential measurement between two CO 2 sensors, each of which is susceptible to instrumental noise and drift. (Newer versions of the FD chamber remedy this problem by using a single CO 2 sensor.…”

An inversion approach for determining distribution of production and temperature sensitivity of soil respiration

“…The top constraint (measurements at 5, 10, and 15 cm) had an average error of 2.01 %, and the top six combinations all had error less than 3 %. These errors are small compared to the degree of random error in CO 2 -flux studies (Lavoie et al, 2015). These results are summarized in Table 4, where the top and bottom five combinations are listed individually and overall.…”