Meteorological measurements at Auchencorth Moss from 1995 to 2016

The Auchencorth Moss atmospheric observatory has being measuring meteorological parameters since 1995. The site was originally set‐up to measure the deposition of sulphur dioxide at a site that represented the vegetation and climate typical of NW Europe, in relatively clean background air. It is one of the longest running flux monitoring sites in the region, over semi‐natural vegetation, providing infrastructure and support for many measurement campaigns and continuous monitoring of air pollutants and greenhouse gases. The meteorological sensors that are used, data processing and quality reviewing procedures are described for a set of core measurements up to 2016. These core measurements are essential for the interpretation of the other atmospheric variables.


| INTRODUCTION
Auchencorth Moss was first instrumented in 1995 due to its suitability for micrometeorological studies, specifically for sulphur dioxide (SO 2 ) deposition under the European funded LIFE Programme (Appendix A). The first meteorological measurements at the site were wind speed, wind direction, air temperature, total solar radiation, net radiation, soil temperature, soil heat flux, surface wetness, rainfall and dewpoint temperature. Measurements continued and have been enhanced with new sensors for many parameters, such as relative humidity, air pressure, direct and diffuse photosynthetically active radiation (PAR) and soil water content. The site was inspected by a representative of the UK Meteorological Office in 2013 and subsequently assigned the status of a regional station as part of the World Meteorological Organisation, Global Atmosphere Watch (WMO GAW, n.d.), in March 2014. This led to an enhancement of management systems and measurement protocols in line with GAW requirements. In 2016, integration into the pan-European Integrated Carbon Observation System (ICOS, n.d.) research infrastructure was also initiated which has further enhanced operations and instrumentation at the site. Future meteorological data will be made available via these programs.

| Site description
Auchencorth Moss (55°47′32°N, 3°14′35°W, 267 m a.s.l.) is a low-lying peatland situated ca. 20 km south-west of Edinburgh (Scotland, UK) as shown in Figure 1. The site has an extensive fetch of open moorland in all directions except the easterly sector. The dominant wind direction is SW (Helfter et al., 2015; Figure 2). The vegetation consists of mixed grass species, heather and mosses (Sphagnum spp. and Polytrichum spp.). The site was drained for agricultural use over 100 years ago but since then the drains have naturally refilled and the site rewetted (Leith et al., 2014). The land is now mainly used for sheep grazing at a low stocking density of ca. 1/ha. The photographs in Figure 3 show the instrumentation and exposure of the site.

| DATA PRODUCTION AND DESCRIPTION
The sensors deployed at the site have been changed and expanded over its 25 years of operation; Table 1 lists which sensors are used and when. All the sensors have been logged using Campbell Scientific data loggers which have been periodically replaced, as shown in Table 1. The sensors were sampled at either 1, 5 or 10 s intervals and 15 or 30 min averages (or sums for rainfall) calculated and stored in the logger. The dataset described here has been through a range of assessments and adjustments to provide a time series that is consistent in terms of instrument calibrations or other changes, and as complete as possible. The following quality control steps are applied to every time series: Periods of known downtime are removed, such as servicing periods, instrument or power failures.
A filter is applied using bounds appropriate to the parameter to exclude spurious data points; the data are plotted and examined visually to ensure they are random outliers and not part of consistent change in the parameter.
The data are plotted alongside measurements of the same parameters from alternative instruments in-place at Auchencorth or at nearby sites also operated by CEH Edinburgh (Appendix B). If a consistent divergence is observed between the time series, conditions are examined in more detail, for example by looking at synoptic weather patterns, to ascertain whether the differences are likely to be real or caused by an instrument fault. If a problem can be clearly identified, the data are excluded.

| Instrument specific data processing
The surface soil heat flux (G) is calculated from the measurements of average soil temperature and average soil heat flux measured by two plates (SHF, see Appendix C for instrument setup), using the standard formula: C s = heat capacity of moist soil; B d = soil bulk density 100 kg/m 3 ; C d = specific heat dry soil, 840 J kg −1 K −1 ; fSWC = fractional soil water content, measured when available or 0.9; C w = specific capacity heat of water, 4,190 J kg −1 K −1 ; SC = storage correction; ΔTs = change in average soil temperature from start to end of measurement period (first and last 2 min); d = plate depth 0.2 m; Δtime = measurement period, 1,800 s; G = net surface soil heat flux; SHF = average measured soil heat flux at 0.2 m (Campbell Scientific, 2016a). The wind speed at 10 m is estimated using turbulence measurements from the 3D sonic anemometers and wind speed measured by a 2D sonic located, all at 3 m height, as follows (Sutton et al., 1993): The NR-Lite net radiometer used from 2001 requires a correction for wind speed to be applied at speeds above 5 m/s (Campbell Scientific, 2016b): where u(1 m) is calculated using the same equations as for 10 m, as above.  For modelling applications, complete time series are often requested so a set of core measurements will be gap-filled for this purpose and included in the CEDA database (ambient air temperature, barometric pressure, relative humidity, total solar radiation, rainfall, wind speed and direction) as described in a later paper. Evaporation of water vapour from the peat-bog will also be added to the database when it has been calculated from the measurements of water-vapour flux which are currently being assessed for quality.

| SUMMARY METEOROLOGY
The site's climate is typical of an exposed rural area in the lowlands of Scotland. The prevailing wind direction is south-westerly and wind speeds averaging 4 ± 0.1 m/s ( Figure 2, Table 2), with occasional strong winds~26 m/s, typically over the winter months. Temperatures average 7.5 ± 0.6°C with a minimum around −7°C although in some years it has been as low as −15°C and a maximum 30°C, with typical summer day time temperatures around 13°C. No temperatures below −10°C have been recorded since winter 2010-2011 ( Figure 4, Table 2).
Total solar and photosynthetically active radiation show similar seasonal patterns with levels over 500 W/m 2 and 1,000 μmol m −2 s −1 respectively in the summer months. The time series do appear to show a slight decrease in levels which is likely due to degradation of the thermopile sensors. The sensors have been sent for calibration and this information will be added to the database when it is available. In contrast, the net radiation sensors appear stable but there is a step change in the measurement range when the REBS Q7 (Campbell Scientific, 1996) was replaced with a Kipp and Zonen NR-Lite in 2001 (minimum readings decreased from~−92 to −175 W/m 2 ).
The soil heat flux and temperature measurements show a decline in the range of observed values over the first 3-4 years of operation which may be due to the sensors taking a long period to bed-in or the build-up of the vegetation and peat layer above them. Regular measurements of the vegetation above the installation were not made but when the sensors were replaced in 2006, it was noted that they appeared to be 4-5 cm further below the moss layer than when initially installed. Soil water content measurements began in 2013 and also show a change in the range of seasonal variation which is believed to be due to the sensors bedding in, as any air gaps around the probes would cause low readings when the soil is dry.
The tipping bucket measures~1,000 mm of rainfall every year which is slightly more than other stations in the region. For example, Edinburgh receives~850 mm a year, but the local topography tends to enhance the local rainfall. In March 2016, a second Casella tipping bucket rain gauge was installed in a pit to reduce wind shear effects on the amount of water collected (http://www.case           llasolutions.com/uk/en/products/met/met-environmental/prod ucts/tbrg.aspx, Figure 3), as specified by the WMO. The old and new gauge gave very similar totals, 775 and 778 mm respectively from 22nd March to 31st December 2016, when scaled for data capture. From 2017, data from the new gauge will be reported and the old gauge decommissioned. The Vaisala FD12P Present Weather Sensor records weather and visibility, as well as estimating precipitation amounts, by measuring the amount of infrared light scattered at an angle of 33°. It also includes a heated capacitive sensor to detect surface water. The precipitation amount is estimated by scaling the optical intensity while the type of precipitation is inferred using air temperature, optical intensity and capacitive signal. Its estimates of rainfall amounts do not agree well with the tipping buckets and, although they detect precipitation at the same times, data capture was relatively low for the FD12P in most years due to data communication and logging issues. The 'present weather' and 'past weather' codes are included in the database, where they represent the most frequently recorded weather type over the last 15 min and one hour respectively. The FD12P can report either WMO or National Weather Service (NWS) codes, in this case the WMO 4,680 codes are recorded (Appendix D).
On hourly, daily or weekly time scales, the soil water content and water table depth mirror the rainfall pattern as would be expected. On several occasions, the water table was measured as being above the surface level and these coincide with periods of heavy or persistent rain. In reality, this was due to the nature of the vegetation and soil; the site was not flooded although the peat and surface vegetation were waterlogged and pools of standing water appeared in some places.
The relative humidity measurements obtained from the Vaisala HMP50 probes show some inconsistencies over the years. Initially, in 2006, the sensor regularly measured over 100% humidity. After the sensor was replaced in 2007, the maximum was less than 95% but gradually increased up to 2010, when it was again replaced. Measurements then appeared quite consistent until the end of 2013 when the system developed a logging fault and measurements below 73% were not recorded properly (in 2015 this cut-off increased to 78%). In 2016, a new sensor and serviced data logger were installed and measurements returned to a more normal range. From 23/03/2016, measurements from the Rotronics HC2S3 are reported in the database and the HMP50 was logged alongside it until 2017. Comparison of the two show they were very closely matched, giving an average of 87%, and maximum 100%, although their minima were slightly different with the Rotronics reading 29% and the Vaisala 27%. Despite the uncertainty on the measured values for some periods, all the data are included in the database because the measurements correlate well with those from the other sites, giving an indication of variations at least. The actual values for the periods below should be treated with caution:   Extensive uniform fetch over the prevailing wind direction, making it ideal for flux measurements using the gradient and eddy-covariance techniques.
Relatively clean air with low or background concentrations of the key pollutants, particularly in the main wind sector.

Representative of the climate throughout North Western
Europe, that is, moderately high rainfall and low temperature regime.
Vegetation representative of open moorland found in large areas of the United Kingdom, Ireland, north-western France and parts of Scandinavia.
Mains power available at an accessible location where infrastructure could be established.
Several papers came out of this first period from 1995 to 2001, that enhanced the understanding of the processes controlling SO 2 , NH 3 and O 3 deposition, many of which are related to meteorological conditions.
In subsequent years, additional instruments for other pollutants and trace gases were added, such as particulates, carbon dioxide and methane, for which the meteorological data were also essential to allow interpretation of the F I G U R E 4 Continued. measurements and production of many other publications. The site has proved very useful in the study of greenhouse gases as it represents a significant ecosystem type for NW Europe.
A full list of published papers that have used Auchencorth data can be found at http://www.auchencorth.ceh.ac.uk/biblio.
The site is one of a few rare examples of long-term flux measurements over any ecosystem, comparable to sites such as Hyytiälä in Finland and others in the Fluxnet database, http://fluxnet.fluxdata.org/.
At present, the meteorological measurements continue to be an essential adjunct to all the pollutant and trace-gas measurements, allowing researchers to make full use of the results and enhance our understanding of the land-atmosphere interaction and atmospheric chemistry. The long time series has the potential for many studies into the changes in the pollution climate and interactions between factors such as SO 2 to NH 3 ratios or climate and greenhouse gases, as well as the climate itself.
The infrastructure at the site is well established and has gradually improved so that it can now accommodate teams of scientists for intensive field campaigns or the addition of longer term measurements where the supporting resources are also available. It is an ideal location for the study of European clean, background air and exchange processes over peaty moorland.

| DATASET LOCATION AND FORMAT
The data are stored by the Centre for Environmental Data Analysis (CEDA, http://www.ceda.ac.uk/) in BADC CSV format, as annual files containing all variables. It is publicly available under the licence: http://www.nationalarc hives.gov.uk/doc/open-government-licence/version/3/. When using these data, you must cite them correctly using the citation given on the CEDA Data Catalogue record.
Annual updates will be made, with a year's new data being added as soon as it has been fully quality controlled and checked.

OPEN PRACTICES
This article has earned an Open Data badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at http://catalogue.ceda.ac.uk/uuid/8e6cbb111cfd41a19c92aadc b2d040fd. Learn more about the Open Practices badges from the Center for Open Science: https://osf.io/tvyxz/wiki.