Data quality

Figure A1 shows the CO2, CH4 and CO mole fractions in target cylinder 1 measured at

Westmaas after data processing. The mixing ratio of Targ1 as established in the laboratory is given by the red dashed line. All results for Targ1 and Targ3 at both locations are summarised in Table A1. The difference between the average Targ1 and Targ3 mole fractions measured by the instrument and the mole fractions determined in the laboratory gives us the bias in the observations. The CO2 measurements are very accurate, with a bias

of 0.01 ppm in all cases. The bias is largest for CO (up to 1.6 ppb). The precision of the measurements, based on the standard deviation, is satisfactory for all species. CO shows the largest relative variation.

APPENDIX A

163

Figure A1: Time series of CO2, CH4, and CO mixing ratio for target cylinder 1 after data processing at location Westmaas for the year 2014. The mole fraction of Targ1 as established in the laboratory is given by the dashed line.

The measurements at both locations are made continuously, although they are sometimes interrupted by technical malfunctions. For example, if the cavity temperature deviates too much from the regulated temperature, the measurements are shut down until the instrument has rebooted. Moreover, other activities near the measurement sites that could disturb the observations are written down in a log. Data affected by these malfunctions or activities are either corrected or flagged, such that the data quality is not affected. The resulting time series after all data processing are shown in Figure A2.

At the start of the measurements at Zweth a 10m mast was unavailable. Therefore, the sampling height was 3m until July 2014. After installing the 10m mast, we continued to take a few samples per day at 3m during one month (July). Previously observed vertical profiles show that the CO2 mixing ratio can change with several ppm moving only a few

tens of meters upwards (Vermeulen et al., 2011). This effect is especially clear during the night, but is dependent on the season due to the role of vegetation as CO2 source and

sink. Similarly, the CH4 mixing ratio can vary with height due to the presence of local

sources. Note that the inlet height is explicitly included in the data files provided for download. We examine the difference between the 3 and 10 m observations by binning observations by hour. This results in an average daily cycle from the irregular samples (N = 263 for both heights). The daily cycles of the CO2, CH4, and CO mixing ratio at both heights

and the differences between the two heights (given by δ) relative to the mixing ratio at 10m are shown in Figure A3.

164

Figure A2: Time series of ambient CO2, CH4, and CO mixing ratio at Westmaas (left) and Zweth (right). The vertical dashed lines indicate the transition from 3 to 10m sampling height.

The diurnal cycle is mainly determined by atmospheric mixing and for CO2 the diurnal

cycle is stronger at 3m than at 10m. The mixing ratio is generally larger at 3m, except from about 8-15h UTC (10-17h LT) when δCO2 fluctuates around zero. For CO2, emissions from

vegetation and the soil play an important role in this diurnal cycle. During the night respiration causes the CO2 mixing ratio to increase. Since there is little mixing in a stable

boundary-layer this increase is more pronounced closer to the surface. During daytime photosynthesis has the opposite effect, being an important sink of CO2. However, due to

convective mixing the difference between the two heights is smaller than at night time. CH4 shows a similar difference between the two heights as CO2. However, vegetation

does not affect CH4, so there are other local sources that cause this difference. One

potentially important source is the peat soil, which is a CH4 producer due to the high

ground water level at this location (Le Mer and Roger, 2001; Schrier-Uijl et al., 2014; Smith et al., 2003). Estimates of CH4 emissions are highly variable, as they depend on many

environmental factors. Yet, emissions of 10-30 μmol m-2 hr-1 have been estimated for Scottish wetlands with soil water depths of 0-10 cm (Smith et al., 2003), which could significantly enhance the CH4 mixing ratio close to the surface. During daytime convection

causes this CH4 flux to be mixed quickly through the boundary-layer, causing δCH4 to be

close to zero.

CO has no local sources and this mixing ratio is mainly determined by advection. This may explain the increase in CO in the morning, which has about the same timing as the

APPENDIX A

165 morning rush hour. Due to the absence of local influences we find no clear diurnal pattern in δCO. Yet the CO mixing ratio generally seems to be higher at 10m. This can be explained by the presence of large trees and bushes that surround the measurement area and potentially reduce the effect of advection at 3m. The 10m mast reaches to the tree tops and is therefore more exposed to advected air.

To conclude, users of this data set should be aware that data for May and June 2014 at location Zweth are more affected by local influences due to the lower sampling height. Considering this time series as homogeneous can therefore result in a bias. However, between 6 and 18h UTC the impact appears to be limited.

Figure A3: Left: Hourly averaged observed CO2, CH4, and CO mixing ratio at 3 (blue circles) and 10m (red triangles) height at Zweth; Right: the difference between 10 and 3m CO2, CH4, and CO mixing ratio relative to the mixing ratio at 10m. LT is the local wintertime (UTC+1).