Concentrations and fluxes of isoprene and oxygenated VOCs at a French Mediterranean oak forest

Abstract. The CANOPEE project aims to better understand the biosphere–atmosphere exchanges of biogenic volatile organic compounds (BVOCs) in the case of Mediterranean ecosystems and the impact of in-canopy processes on the atmospheric chemical composition above the canopy. Based on an intensive field campaign, the objective of our work was to determine the chemical composition of the air inside a canopy as well as the net fluxes of reactive species between the canopy and the boundary layer. Measurements were carried out during spring 2012 at the field site of the Oak Observatory of the Observatoire de Haute Provence (O3HP) located in the southeast of France. The site is a forest ecosystem dominated by downy oak, Quercus pubescens Willd., a typical Mediterranean species which features large isoprene emission rates. Mixing ratios of isoprene, its degradation products methylvinylketone (MVK) and methacrolein (MACR) and several other oxygenated VOC (OxVOC) were measured above the canopy using an online proton transfer reaction mass spectrometer (PTR-MS), and fluxes were calculated by the disjunct eddy covariance approach. The O3HP site was found to be a very significant source of isoprene emissions, with daily maximum ambient concentrations ranging between 2–16 ppbv inside and 2–5 ppbv just above the top of the forest canopy. Significant isoprene fluxes were observed only during daytime, following diurnal cycles with midday net emission fluxes from the canopy ranging between 2.0 and 9.7 mg m−2 h1. Net isoprene normalized flux (at 30 °C, 1000 μmol quanta m−2 s−1) was estimated at 7.4 mg m−2 h−1. Evidence of direct emission of methanol was also found exhibiting maximum daytime fluxes ranging between 0.2 and 0.6 mg m−2 h−1, whereas flux values for monoterpenes and others OxVOC such as acetone and acetaldehyde were below the detection limit. The MVK+MACR-to-isoprene ratio provided useful information on the oxidation of isoprene, and is in agreement with recent findings proposing weak production yields of MVK and MACR, in remote forest regions where the NOx concentrations are low. In-canopy chemical oxidation of isoprene was found to be weak and did not seem to have a significant impact on isoprene concentrations and fluxes above the canopy.

where, I(RH + i ) is the ion count signal at mass m/z i (units:cps), I(RH + i ) zero is the ion count signal at mass m/z i from the zero air (cps), m/z 21 and m/z 37 the counts of the primary (H 3 18 O + ) and reagent clusters (H 3 16 O + H 2 16 O + ) (cps), m/z 21 zero and m/z 37 zero the counts of the primary and reagent clusters when measuring from the zero air (cps), I(RH + i ) norm the normalized ion count rate of I(RH + ) (ncps) and VMR the volume mixing ratios (ppbv).

VOC not present in the calibration gas standard
We calculated the normalized sensitive S norm-calculated for VOC not present in our calibration gas standard (acetic acid and hydroxyacetone) using the procedure described by De Gouw and Warneke (2006 (2.8 cm 2 Vs -1 ) is the reduced mobility of the primary ion and V m (22400 cm 3 ) is the molar volume.
In both cases, VMR were calculated as the ratio of the normalized count rate of ions detected to the normalized sensitivity.

or (S4)
However, due to the uncertainties associated to the proton transfer reaction rate coefficients k i and the relative transmission curve, the accuracy of the calculated data using the calculated sensitivity S norm-calculated is significantly lower than the accuracy of the data based on the measured normalized sensitivities after gas calibration.

PTR-MS based water vapour flux measurements and comparison with a reference system.
Water vapour concentrations and fluxes were measured using a standard reference system based on the combination of a closed path infrared gas analyzer (IRGA, Model 7000, Li-COR) and the sonic anemometer. Both instruments were set to a sampling frequency of 20 Hz. Ambient air close to the sonic sensor head was continuously sampled through the main line (inlet at 10 m) leading to the IRGA instrument. Fluxes were calculated by the eddy covariance (EC) method, as implemented before by Loubet et al., (2011). Additionally a high frequency losses corrections was implemented based on the co-ogive method as in (Ammann et al., 2006) EC water vapour fluxes from the standard reference system were also compared to DEC water vapour fluxes derived from the signal at m/z 37 of the PTR-MS (Ammann et al., 2006). The same anemometer was used for the DEC flux measurements with the PTR-MS as for the IRGA system.
In the PTR-MS, water vapour was detected at m/z 37, a mass corresponding to the cluster ion H 3 O + H 2 O + present in the drift tube. Water clusters in the drift tube originate from the ion source but also from the water vapour in the sample air. It is expected that the contribution of the ion source to the m/z 37 is relatively constant and thus, there is a quantitative relationship between the signal at m/z 37 and the concentration of the water vapour (Ammann et al., 2006).
Ion counts at m/z 37 were calibrated against the reference IRGA concentration, in order to investigate the relationship between the m/z 37 signal and the water vapour concentration in the sampled air (Fig. S1).      Water vapour fluxes obtained with the PTR-MS were highly comparable to the results of the IRGA reference system. A linear relation was found between the latent heat measured by DEC (PTR-MS) and conventional EC (IRGA), with a coefficient of correlation, R 2 , of 0.75. This good agreement supports our PTR-MS eddy flux measurements of VOCs.

Comparison between DEC fluxes and vertical concentration gradients
Isoprene fluxes derived by DEC were also compared to the vertical gradient of isoprene concentration inside the canopy multiplied by the friction velocity: Although one cannot quantitatively derive a flux from the gradient method, because the lower measurement height was not only within the roughness sublayer, but also located below some of the sources inside the canopy, the correlation found was fairly strong (R 2 = 0.6), lending further confidence to the DEC flux measurements ( Figure S5). Here, the measured gradient stands for a proxy of the above-canopy gradient and u * as a proxy for the eddy-diffusivity, which in reality depends further on atmospheric stability. Figure S5. Correlation between isoprene fluxes measured above the canopy by DEC and the gradient of isoprene concentration (between 2 m and 10 m) multiplied by the friction velocity.

Time for diffusion transport of a trace gas
The turbulent transport time between the measurement height ( ) and the ground surface can be expressed as the transfer resistance through each layer multiplied by the layer height (Garland, 1977).
where is the canopy displacement height and is the canopy roughness length. Estimates from the literature gives =0.7* where is the canopy height, and = 0.13* .
Turbulent resistances within and above the canopy, and respectively, are expressed as: where k(=0.4) is the von Kármán constant, and are the canopy roughness lengh for temperature and momentum, (=0.02 m; (Personne et al., 2009)) is the ground surface roughness length below the canopy; is the attenuation coefficient for the decrease of the wind speed inside the canopy, defined as LAI/2 (Yi, 2008), is the eddy diffusivity at the canopy height; and and are dimensionless stability correction functions for heat and momentum (Dyer, 1974).
In the current study the transport time was estimated to be in the range of 30-60 s in daytime.