A comparison of manual and automated systems for soil CO2 flux measurements: trade-offs between spatial and temporal resolution

Abstract

Soil respiration is affected by distributions of roots and soil carbon substrates and by temperature and soil water content, all of which vary spatially and temporally. The objective of this paper was to compare a manual system for measuring soil respiration in a temperate forest, which had a greater spatial distribution of measurements (n=12), but poorer temporal resolution (once per week), with an automated system which had poorer spatial distribution (n=3) but superior temporal frequency of measurements (hourly). Soil respiration was measured between 18 June and 21 August, 2002, at the Harvard Forest in central Massachusetts, USA. The fluxes measured within 1 h of each other by these systems were not significantly different. However, extrapolations of the mid‐morning manual measurements to daily flux values were consistently lower (averaging 13% lower) than the daily estimates obtained from summing the 24 hourly measurements of the automated system. On the other hand, seasonal flux estimates obtained by interpolating between weekly manual sampling dates or by summing the hourly automated measurements were nearly identical. Underestimates by interpolated weekly manual measurements during some periods were cancelled by overestimates during other periods. Hence, a weekly sampling schedule may be sufficient to capture the most important variation of seasonal efflux of CO₂ from the soil. The larger number of chambers that could be measured with the manual system (larger n) resulted in a smaller 95% confidence interval for characterizing spatial variability within the study area on most dates. However, the greater sampling frequency of the automated system revealed rapid responses of soil respiration to wetting events, which permitted better empirical modelling of the effects of soil temperature and moisture on soil respiration than could have been achieved with the manual sampling system. Most of the positive residuals of a function that predicts soil respiration based on temperature were from fluxes measured within 12 h of a rain event, and the residuals were positively correlated with water content of the O horizon. The automated system also demonstrated that Q₁₀ values calculated for diel variation in soil temperature over a few days were not significantly different than Q₁₀ values for the entire 3 month summer sampling period. In summary, a manual system of numerous, spatially well‐distributed flux chambers measured on a weekly basis may be adequate for measuring seasonal fluxes and may maximize confidence in the characterization of spatial variance. The high temporal frequency of measurements afforded by automation greatly improves the ability to measure and model the effects of rapidly varying water content and temperature. When the two approaches can be combined, the temporal representativeness of the manual measurements can be tested with the automated measurements and the spatial representativeness of the automated measurements can be tested by the manual measurements.