Abstract. The exchange ratio (ER) between atmospheric O2 and CO2 is a useful tracer for better understanding the carbon budget on global and
local scales. The variability of ER (in mol O2 per mol CO2) between terrestrial ecosystems is not well known, and there is no consensus on how to derive
the ER signal of an ecosystem, as there are different approaches available, either based on concentration (ERatmos) or flux measurements
(ERforest). In this study we measured atmospheric O2 and CO2 concentrations at two heights (23 and 125 m) above
the boreal forest in Hyytiälä, Finland. Such measurements of O2 are unique and enable us to potentially identify which forest carbon
loss and production mechanisms dominate over various hours of the day. We found that the ERatmos signal at 23 m not only represents the diurnal cycle of the forest exchange but also includes other factors, including entrainment of air masses in the atmospheric boundary layer before midday, with different thermodynamic and atmospheric composition characteristics. To derive ERforest, we infer O2 fluxes
using multiple theoretical and observation-based micro-meteorological formulations to determine the most suitable approach. Our resulting
ERforest shows a distinct difference in behaviour between daytime (0.92 ± 0.17 mol mol−1) and nighttime
(1.03 ± 0.05 mol mol−1). These insights demonstrate the diurnal variability of different ER signals above a boreal forest, and we
also confirmed that the signals of ERatmos and ERforest cannot be used interchangeably. Therefore, we recommend measurements on
multiple vertical levels to derive O2 and CO2 fluxes for the ERforest signal instead of a single level time series of the
concentrations for the ERatmos signal. We show that ERforest can be further split into specific signals for respiration
(1.03 ± 0.05 mol mol−1) and photosynthesis (0.96 ± 0.12 mol mol−1). This estimation allows us to separate the
net ecosystem exchange (NEE) into gross primary production (GPP) and total ecosystem respiration (TER), giving comparable results to the more
commonly used eddy covariance approach. Our study shows the potential of using atmospheric O2 as an alternative and complementary method to
gain new insights into the different CO2 signals that contribute to the forest carbon budget.