Overestimation of soil CO₂ fluxes from closed chamber measurements at low atmospheric turbulence biases the diurnal pattern and the annual soil respiration budget

Abstract
Precise quantification of the diurnal and seasonal variation of soil respiration (Rs) is crucial to correctly estimate annual soil carbon fluxes as well as to correctly interpret the response of Rs to biotic and abiotic factors on different time scale. In this study we found a systematic effect of low atmospheric turbulence on continuous hourly Rs measurements with closed chambers throughout one year in a temperate Danish beech forest. Using friction velocity (u⋆) measured at the site above the canopy, we filtered out chamber flux data measured at low atmospheric turbulence. The non-filtered data showed a clear diurnal pattern of Rs across all seasons with highest fluxes during night time suggesting an implausible negative temperature sensitivity of Rs. When filtering out data at low turbulence, the annually averaged diurnal pattern changed, such that the highest Rs fluxes were seen during day time, i.e. following the course of soil temperatures. This effect on the diurnal pattern was due to low turbulence primarily occurring during night time. We calculated different annual Rs budgets by filtering out fluxes for different levels of u⋆. The highest annual Rs budget was found when including all data and it decreased with an increasing u⋆ filter threshold. Our results show that Rs was overestimated at low atmospheric turbulence throughout the year and that this overestimation considerably biased the diurnal pattern of Rs and led to an overestimation of the annual Rs budget. Thus we recommend that that any analysis of the diurnal pattern of Rs must consider overestimation of Rs at low atmospheric turbulence, to yield unbiased diurnal patterns. This is crucial when investigating temperature responses and potential links between CO2 production and Rs on a short time scale, but also for correct estimation of annual Rs budgets. Acknowledgements: This study was funded by the free Danish Ministry for Research, Innovation and higher Education, the free Danish Research Council (DFF - 1323-00182).