ICOS Ecosystem Station Labelling Report

Station: BE-Maa ()

Viterbo (), Antwerp (), Bordeaux (France), May 4th 2020 Description of the Labelling procedure The Step2 procedure has the aims to organize the building the station in accordance with the ICOS Instructions, to establish the link with the ETC, and to validate all the data formats and submission. Furthermore, it also involves defining the additional steps needed after the labelling to complete the station construction according to the station Class. During the Step2 a number of steps are required and organized by the ETC in collaboration with the PI.

Preparation and start of the Step2 The station started the Step1 of the labelling on June 13th 2016 and got the official approval on May 31th 2017. The Step2 started officially on July 19th 2018 with a specific WebEx between the ETC members and the station team members where the overall procedure was discussed and explained.

Team description The station PI has to describe the station team and provide the basic information about the proposed station using the BADM system. The submission is done using a specific ICOS interface.

Sampling scheme implementation The sampling scheme is the distribution of points in the ecosystem where a number of measurements must be done. It is composed by two different types of sampling locations: the Sparse Measurement Plots (SP) that are defined by the ETC following a stratified random distribution based on information provided by the PI and the Continuous Measurement Plots (CP) where continuous measurements are performed.

Measurements implementation The measurement of a set of variables must be implemented in the Step2 labelling phase. The compliance of each proposed sensor and method is checked by the ETC and discussed with the PI in order to find the optimal solution. In case specific reasons make it impossible to follow the ICOS agreed protocols and Instructions an alternative solution, equally valid, is defined and discussed with the MSA if needed. Once the sensors and methods have been agreed upon the station Team has to implement the measurements using calibrated sensors, submit the metadata to the ETC and start to submit data Near Real Time for the continuous measurement. Also vegetation samples must be collected and shipped to the ETC chemical laboratory in France. The list of variables to be implemented during Step2 is reported in Table 1. Adaptation of the table to specific ecosystem conditions are possible and always discussed with the PI and the MSA. In addition to the variables reported in Table 1 there is an additional set of measurements that is requested and that must be implemented after the labelling in the following 1-2 years. For all these variables (in particular for the soil sampling) an expected date and specific method to be used is discussed and agreed upon before the end of the Step2 process.

Group Variable Turbulent fluxes EC fluxes CO2-LE-H Storage fluxes SW incoming LW incoming SW outgoing Radiations LW outgoing PPFD incoming PPFD outgoing Air temperature Relative humidity Air pressure Meteorological above ground Total precipitation Snow depth Backup meteo station Soil temperature profiles Soil water content profiles Soil climate Soil heat flux density Groundwater level History of disturbances Site characteristics History of management Site description and characterization Green Area Index Biometric measurement Aboveground Biomass Sample of leaves Foliar sampling Leaf Mass to Area Ratio

Additional variables for Class1 stations Radiation SW/PPFD diffuse Meteorological Precipitation (snow) Biometric measurement Litterfall

Table 1 – Variables requested for Step2

Data evaluation Stations entering Step2 have already been analyzed during Step1 of the labelling but the optimal configuration and the possible presence of issues can be checked only by looking to the first data measured. For this reason, a number of tests will be performed on the data collected during the Step2 (NRT submissions, that can be integrated if needed by existing data) and the results will be discussed with the PI in order to find the best solution to ensure the maximum quality that is expected by ICOS stations. Four tests are performed: Test 1 - Percentage of data removed During the fluxes calculation the raw data are checked by a number of and some of them will lead to data exclusion and gaps. The number of half hours removed by these QAQC filters will be calculated and the target value is to have less than 40% of data removed. If the test fails, an in depth analysis of the reasons is performed in order to find solutions and alternatives. Test 2 – Footprint and Target Area The Target Area is the area that we aim to monitor with the ICOS station. The test will analyze the estimated contribution area for each half hour using a footprint model (Klijun et al. 2015) and will check how many records have a contribution coming mainly from the target area. The target is to have at least 70% of measurements that are coming mainly (70% of the contribution) from the Target Area. If the test fails, a discussion with the PI is started in order to find solutions and alternatives, in particular changing the measurement height or wind sectors to exclude. Test 3 – Data Representativeness in the Target Area The aim is to identify areas that are characterized by different species composition or different management (and consequently biomass and density) and analyze, using the same footprint model (Kljun et al. 2015), the amount of records coming from the different ecosystems, checking their representativeness in terms of day-night conditions and in the period analyzed. The target is to get, for the main ecosystem types, at least 20% of the data during night and during day and also distributed along the period analysed. If not reached, a discussion with the PI is started in order to find solutions and alternatives, in particular changing the measurement height or wind sectors to exclude. Test 4 – CP Representativeness in the Target Area The CPs must be as much as possible representative of the Target Area and this will be checked on the basis of the results of the site characterization, in particular in relation to species composition, biomass and management. The target is to have the percentage of the two main species and their biomass in the CP not more than 20% different respect to the measurements done in the SP plots. In case the CPs proposed do not represent a condition present in the Target Area they are relocated or one or more additional CPs can be added.

Station Description

The station with ICOS code BE-Maa, is located in Maasmechelen, 10 Km northeast of the city of on the Belgian plateau or High Campine, a region that stretches over north-eastern Belgium and the south-eastern . A national park of 60 Km2 now encompasses part of this region, founded in 2006. The site is situated in a 400ha heathland called “Mechelse heide”, and it is a shrubland, having coordinates in WGS84 system Latitude 50.979868 °N and Longitude 5.631851 °E, at an elevation of 87 m above sea level and with an offset with respect to the Coordinated Universal Time (UTC) equal to +01. The site is marked by the following climate characteristics: Mean Annual Temperature 10.3 °C, Mean Annual Precipitation 839.1 mm, Mean Annual Radiation 116.7 W m-2. The shrubland is dominated by common heather (Calluna vulgaris L.), that grows up to one meter high. Other typical species are cross-leaved heath (Erica tetralix L.), bell heather (Erica cinerea L.) and common broom (Cytisus scoparius L.).

Figure 1: the BE-Maa station.

Team description The staff of the site has been defined and communicated in October 2018 and updated at a later date. It includes in addition to the PI the technical-scientific staff. Below the summary table of the Team members is reported.

MEMBER_NAME MEMBER_INSTITUTION MEMBER_ROLE MEMBER_MAIN_EXPERT

Marilyn Roland University of Antwerp PI

Jan Segers University of Antwerp SCI

Lodewijk Lefevre SCI-ANC

Tim De Meulder SCI-ANC

Jasper Van Look University of Antwerp TEC-FLX Table 2 - Description of team members roles at BE-Maa

Spatial sampling design For the spatial sampling design at Maasmechelen, the PI originally proposed a very small target area (TA), designed to exclude a burned area at West of the proposed TA (in red in Fig. 2). ETC replied that such a small TA would 1) imply the risk that test 2 will not be passed because a large number of flux will likely have most of the contribution from outside and 2) it will likely imply exceptions in the SP-I number and disposition (both reducing the minimum inter-distances and their number). Consequently, suggested to extend the TA to the sum of the two sub-areas (red + white in Fig. 2). The station PI, accepted the proposal and confirmed that the burned area will not be managed differently from the rest of the TA.

Figure 2: BE-Maa aerial map showing the original proposal for target and exclusion areas and EC tower position. The current (official) target area is the sum of the two.

The sampling design was achieved considering the TA and EA as shown in Fig. 3 along with the sampled SP-I positions. It has been agreed that CPs will not be circular plots with gridded sampling points, but transects of about 27 m long with sampling points at 3 m distances. It was also agreed to move the EC tower so as to optimize its fetch (Fig. 3).

Figure 3: Proposed spatial features at BE-Maa according to the reported target area (TA), exclusion area (EA) and ICOS requirements. The TA surface is 1.83 Ha, the total excluded area is of 0.06 Ha. Note that the position of the EC tower have been changed from the original PI proposal. The station team mapped the points in the field, used 4 reservoir SP-II points, and report the coordinates to ETC. The check on the field coordinates has been performed and all points were accepted, having a maximum offset from the randomly extracted locations of 153.2 and 111.6 cm for SP-I and SP-II respectively.

Station implementation Eddy covariance

EC System

MODEL GA_CP-LI-COR LI-7200RS SA-Gill HS-50

SN 72H-0824 H161810

HEIGHT (m) 3 3

EASTWARD_DIST (m) 0.6 0.7

NORTHWARD_DIST (m) -0.4 -0.7

SAMPLING_INT 0.05 0.05

LOGGER 1 1

FILE 1 1 GA_FLOW_RATE 15 -

GA_LICOR_FM_SN FM1-0318 -

GA_LICOR_AIU_SN AIU-1725 -

SA_OFFSET_N - 145

SA_WIND_FORMAT - U, V, W

SA_GILL_ALIGN - Spar

ECSYS_SEP_VERT -0.02 ECSYS_SEP_EASTWARD -0.16 ECSYS_SEP_NORTHWARD 0.14

ECSYS_WIND_EXCL 332.5

ECSYS_WIND_EXCL_RANGE 15

The station has installed the ICOS instruments for eddy covariance measurements, Gill HS sonic anemometer and LICOR LI7200 gas analyser. The calibration of both the sonic anemometer and the IRGA is valid until September 2020. After the agreement between ETC and PI on the extension of the TA (see previous section), the PI asked to lower the sonic height from 3 to 2 m and to move it in a different location, and the ETC accepted. However, in the Step1 the PI proposed to orient the sonic at 234° from N, while it is at 145° from N. The PI explained that this choice was made to include a cabin in the exclusion sector of the sonic. The excluded wind sector however is the second more represented one (NW). At the beginning of December the first 3 weeks of data at the height of 2.5 m were processed. After the test the issue was a large percentage of data discarded due to non- stationarity and the wind sector exclusion. It has been decided that the exclusion sector can be safely reduced to 15°. A further increase of the measurement height to 3 m was also agreed on Jan 2020 to try to reduce the non-stationarity problem. This strategy was successful (see “Data check and test” section)

The station experienced a water leakage in the IRGA system in wintertime. The issue seems indeed to be related to the presence of fog which likely enters into the tube. The station team cleaned and replaced the filter. However, other stations had similar (but more serious) issues of leakage, which were mostly solved by replacing the O-rings. LICOR was also contacted. The station team however checked the O-rings, that are OK. The issue seemed occasional, and the team will keep the eyes open on this point, and also on the communication between the flow module and the IRGA, that in one occasion raised anomalies.

An episode of intensive rainfall also led to sonic values out-of-range. The station team replaced the whole sonic system (including cables), which didn't help. When the rain stopped, the issue was no longer present. ETC suggested to keep an eye on it and shared the info with Gill on the issue. When Gill will provide some suggestions on that, the ETC will inform the station.

Further details on EC setup are reported in the table above. Storage: Considering the lower measurement height (3 m), and the particular vegetation characteristics at site (very dense with a maximum height of about 70 cm), the system to measure the storage flux was not requested for this station.

Radiations

HEIGHT EASTWARD_DIST NORTHWARD_DIST MODEL SN VARIABLE_H_V_R (m) (m) (m) SW_IN_1_1_1

SW_OUT_1_1_1 RAD_4C-K&Z CNR4 120744 2.8 11.5 20.1 LW_IN_1_1_1

LW_OUT_1_1_1

RAD_PAR-K&Z PQS1 140387 2.1 11.3 20.3 PPFD_IN_1_1_1 RAD_PAR-K&Z PQS1 140365 2.8 11.3 20.3 PPFD_OUT_1_1_1

For short- and long-wave radiations the CNR-4 (Kipp & Zonen) radiometer will be used in combination with the CNF4 ventilation and heating unit, for PPFD radiations the PQS1 quantum sensors (Kipp & Zonen) will be used, and for diffuse radiation the Delta-T Devices SPN1 pyranometer. All the sensors proposed for radiation measurements are ICOS compliant.

Precipitation

HEIGHT EASTWARD_DIST NORTHWARD_DIST MODEL SN VARIABLE_H_V_R (m) (m) (m) PREC-OTT Pluvio2 398868 1 -48.37499 26.90917 P_2_1_1

Total precipitation will be measured by OTT PLuvio2 (OTT Hydromet) weighing gauge in combination with an Alter type windshield. For the snow depth, the PI proposed to use a digital camera and snow stake and, since the station team demonstrated that at the site there is snow only for a few days per year, the ETC accepted it as exception.

Air temperature, relative humidity and air pressure

HEIGHT EASTWARD_DIST NORTHWARD_DIS MODEL SN VARIABLE_H_V_R (m) (m) T (m) RHTEMP-Vaisala K3010011 2.12 -1.3 0.5 TA_1_1_1 HMP155 RH_1_1_1

PRES-Vaisala J0330012 2.12 -1.2 1.1 PA_1_1_1 PTB210 WD-Vector 60286 4.13 0 -0.33 WD_1_1_1 W200P WS-Vector 16247 4.13 0 0.33 WS_1_1_1 A100(x)

The sensors installed at the station for TA, RH and PA are all ICOS-compliant: Vaisala HMP155 and Vaisala PTB210 respectively. Their installation respects the ICOS standards. The station team communicated that the original sensors, not calibrated, were removed and spare sensors mounted to go on with the labelling. These spare sensors are also out of calibration, but were unused since purchase (6-7 years ago). Original sensors will be sent for calibration in winter 2019. This plan is OK for the ETC, and actually replacing the sensors with spare sensors was not needed, but not an issue too. The station team will have to decide, once the main sensors are back from calibration, whether to replace them again, or put them in parallel to the spare ones to check their calibration.

In addition, the station also installed two sensors for measuring wind direction and wind speed (Vector V200P and Vector A100 respectively). The sensors selected are ICOS compliant. Their calibration is expired, but as these sensors, while recommended, are not mandatory for Class2 stations, this will not prevent the labelling.

Backup meteorological station

HEIGHT EASTWARD_DIST NORTHWARD_DIST MODEL SN VARIABLE_H_V_R (m) (m) (m)

RHTEMP-Vaisala TA_3_1_1 K4550057 1.8 -33.6 16.2 HMP155 RH_3_1_1

PRES-Setra 278 4095442 1.8 -33.6 16.2 PA_3_1_1 RAD_SW-APO SP110 38487 2 -34.2 16.4 SW_IN_3_1_1 PREC-EML ARG100 1438044 0.7 -29.4 20.1 P_3_1_1

WDWS-Young WD_3_1_1 WM133510 2.55 -32.52 -16.4 WindMonitor 0510x WS_3_1_1

The sensors selected by the PI to be used as backup sensors for TA, RH and P are ICOS compliant: the Rotronic HC2(A)-S family, and the raingauge ARG100. The Apogee SP110, measuring shortwave incoming radiation, is not fully ICOS compliant, but it is accepted as a backup sensor. In addition, a barometer Setra 278 and a 2D sonic anemometer (Young WindMonitor 0510x family) are installed at the backup station, even if not mandatory. Their need for calibration will be established by comparison with the main sensors. The backup station is independent from the main one both in terms of power and data logging. However, the backup station is powered by solar panels, and has no backup. The station team confirmed to the ETC that this solution is robust, providing also consumption details that are way lower than the battery capacity. ETC accepted, but in case of missing data will ask to update the system.

The station team is going to have an independent connection for data submission from backup station to CP: this is not required, but it is an even safer strategy to have backup data submitted in case of issues to the main connection.

Soil temperature, soil water content, soil heat flux density and water table depth

HEIGHT EASTWARD_DIST NORTHWARD_DIST MODEL SN VARIABLE_H_V_R (m) (m) (m) TEMP-Campbell BE-Maa_TS_1_1_1 -0.015 -10.5 46.7 TS_1_1_1 CS10X TEMP-Campbell BE-Maa_TS_1_2_1 -0.05 -10.5 46.7 TS_1_2_1 CS10X TEMP-Campbell BE-Maa_TS_1_3_1 -0.1 -10.5 46.7 TS_1_3_1 CS10X TEMP-Campbell BE-Maa_TS_1_4_1 -0.2 -10.5 46.7 TS_1_4_1 CS10X TEMP-Campbell BE-Maa_TS_1_5_1 -0.5 -10.5 46.7 TS_1_5_1 CS10X TEMP-Campbell BE-Maa_TS_1_6_1 -1 -10.5 46.7 TS_1_6_1 CS10X TEMP-Campbell BE-Maa_TS_2_1_1 -0.015 16.8 43.1 TS_2_1_1 CS10X TEMP-Campbell BE-Maa_TS_2_2_1 -0.05 16.8 43.1 TS_2_2_1 CS10X TEMP-Campbell BE-Maa_TS_2_3_1 -0.1 16.8 43.1 TS_2_3_1 CS10X TEMP-Campbell BE-Maa_TS_2_4_1 -0.2 16.8 43.1 TS_2_4_1 CS10X TEMP-Campbell BE-Maa_TS_2_5_1 -0.5 16.8 43.1 TS_2_5_1 CS10X TEMP-Campbell BE-Maa_TS_2_6_1 -1 16.8 43.1 TS_2_6_1 CS10X TEMP-Campbell BE-Maa_TS_3_1_1 -0.015 -24.3 6.2 TS_3_1_1 CS10X TEMP-Campbell BE-Maa_TS_3_2_1 -0.05 -24.3 6.2 TS_3_2_1 CS10X TEMP-Campbell BE-Maa_TS_3_3_1 -0.1 -24.3 6.2 TS_3_3_1 CS10X TEMP-Campbell BE-Maa_TS_3_4_1 -0.2 -24.3 6.2 TS_3_4_1 CS10X TEMP-Campbell BE-Maa_TS_3_5_1 -0.5 -24.3 6.2 TS_3_5_1 CS10X TEMP-Campbell BE-Maa_TS_3_6_1 -1 -24.3 6.2 TS_3_6_1 CS10X TEMP-Campbell BE-Maa_TS_4_1_1 -0.015 -49.15 -20.46 TS_4_1_1 CS10X TEMP-Campbell BE-Maa_TS_4_2_1 -0.05 -49.15 -20.46 TS_4_2_1 CS10X SWCTEMP-Campbell 13053 -0.05 -10 46.4 SWC_1_1_1 CS65X SWCTEMP-Campbell 13067 -0.1 -10 46.4 SWC_1_2_1 CS65X SWCTEMP-Campbell 13073 -0.2 -10 46.4 SWC_1_3_1 CS65X SWCTEMP-Campbell 13070 -0.5 -10 46.4 SWC_1_4_1 CS65X SWCTEMP-Campbell 13057 -1 -10 46.4 SWC_1_5_1 CS65X SWCTEMP-Campbell 13081 -0.05 17.6 42.9 SWC_2_1_1 CS65X SWCTEMP-Campbell 13102 -0.1 17.6 42.9 SWC_2_2_1 CS65X SWCTEMP-Campbell 13103 -0.2 17.6 42.9 SWC_2_3_1 CS65X SWCTEMP-Campbell 13080 -0.5 17.6 42.9 SWC_2_4_1 CS65X SWCTEMP-Campbell 13090 -1 17.6 42.9 SWC_2_5_1 CS65X SWCTEMP-Campbell 13100 -0.05 -25 6.8 SWC_3_1_1 CS65X SWCTEMP-Campbell 13091 -0.1 -25 6.8 SWC_3_2_1 CS65X SWCTEMP-Campbell 13092 -0.2 -25 6.8 SWC_3_3_1 CS65X SWCTEMP-Campbell 13083 -0.5 -25 6.8 SWC_3_4_1 CS65X SWCTEMP-Campbell 13084 -1 -25 6.8 SWC_3_5_1 CS65X SWCTEMP-Campbell 13114 -0.05 -49.5 -20.9 SWC_4_1_1 CS65X SOIL_H-Hukseflux 3206 -0.05 -10.3 46.5 G_1_1_1 HFP01SC SOIL_H-Hukseflux 3207 -0.05 17.4 43 G_2_1_1 HFP01SC SOIL_H-Hukseflux 3198 -0.05 -24.7 6.5 G_3_1_1 HFP01SC SOIL_H-Hukseflux 3209 -0.05 -49.29 -20.68 G_4_1_1 HFP01SC

The station team has installed the full set of soil meteo sensors required for their Class 2 station, i.e. three fully equipped soil plots and one extra soil heat flux plate with accessory sensors. The sensors have been installed at locations in the target area that are accepted by the ETC (see Figure 4). The set-up of the installed soil plots and flux plate is compliant with the ICOS Instructions in terms of sensor models, number of sensors, and sensor depths (see Table above and Figure 5). Furthermore, the station team has submitted all requested metadata on the installed sensors.

The station is exempt from water table depth measurements, because data from a nearby monitoring well have shown that the groundwater table is located at a depth of more than 5 m. Measurements of water table depth are hence considered irrelevant for BE-Maa.

Figure 4: Location of the three soil plots and the extra soil heat flux plate in the target area. The white line is the target area boundary. The square indicates the location of the EC tower. Yellow = exclusion area.

Figure 5: Set-up of the three soil plots (all sensors) and the extra heat flux plate with accessory sensors (grey box). SWC = volumetric soil water content, G = soil heat flux density, and TS = soil temperature. Spatial heterogeneity characterization The station team has collected in the course of the 2019 growing season all data required to characterize the spatial heterogeneity of the vegetation at the site. These data entail measurements of species cover, vegetation height, and GAI at the 100 SP-II-order points (and at the 40 installed CPs). All data have been quality-checked by the ETC (see Figure 6 for the SP-II-order measurements). Note: GAI has been measured by means of spectral reflectance. This method has been discussed and agreed upon by the station team and the ETC. The station team has done substantial efforts in collecting measurements to calibrate a relationship between GAI and several spectral indices, the best of which was selected for GAI calculation (Enhanced Vegetation Index 2; Jiang et al, 2008, Remote Sens. Environ., 112(10), 3833-3845; GAI = 20.334 EVI2 - 0.8611).

Based on these characterization measurements and the history of management, the target area is divided in three Land Cover Typologies (LCTs), i.e. areas with a different vegetation composition and/or density that may contribute differently to the sensed EC fluxes (see Figures 7 and 8):

● LCT1: an area vegetated with old common heather (Calluna vulgaris (L.) Hull). This area includes the EC tower. ● LCT2: an area managed by controlled burning in 2009 and vegetated with common heather (i.e. younger and shorter than in LCT1). ● LCT3: an area managed by controlled burning in 2014 and vegetated with a mix of young common heather and purple moor-grass (Molinia caerulea (L.) Moench) (much less heather cover than in LCT2)

Note: Since the LCTs will all undergo the same future management, the differences in vegetation between the LCTs will likely become insignificant in a few years. The target area can from then on be treated as one single LCT. At present, the differences between the LCTs are considered too large to treat the target area as one single LCT.

Figure 6: Measurements of a) species cover, b) vegetation height, and c) Green Area Index (GAI) at the 100 SP-II-order points. The points are grouped per LCT that is distinguished in the target area. The three leftmost bars in each panel show the LCT average with standard deviation.

Figure 7: Division of the target area in three Land Cover Typologies (LCTs): LCT1, LCT2, and LCT3. The white line and the dashed lines indicate the boundaries of the target area and the LCTs. White points show the locations of the 100 SP-II-order points, while squares indicate the locations of the 40 installed CPs (with ID). Letters a, b, and c indicate the locations at which the images of the LCTs shown in Figure 8 are taken.

Figure 8: Images of the three LCTs distinguished in the target area: a) LCT1, b) LCT2, and c) LCT3. Green Area Index The station team has collected the minimum number of two GAI datasets that are requested as part of the step 2 labelling procedure. It has in fact measured GAI with the spectral reflectance method on three different dates between March and June 2019 in all 40 CPs. All measurements have been quality-checked by the ETC - see Figure 9 for results.

Figure 9: GAI measured in the 40 CPs on three dates in 2019, shown as the average GAI per transect (10 CPs per transect). Error bars show the standard deviation to the transect average.

Vegetation sampling and analysis The first set of Calluna leaf samples were received and analysed. The values of foliar nutrient content showed no anomaly but the LMA found was twice the value expected so we are checking this value with the station team.

Data check and test Data quality analysis (Test 1) The test aims at quantifying the availability of NEE half-hourly data after the application of Quality Control (QC) procedures. The requirement expected for the Step 2 of the station labelling is that the total percentage of missing and removed data after the QC filtering does not exceed the 40% threshold value. This threshold was agreed between the ETC and the ecosystem MSA, and has the goal to maintain the ICOS data quality standard. Tests involved in the QC procedure are described in Vitale et al. 2020 and aim at detecting NEE flux estimates contaminated by the following sources of systematic error: (i) EC system malfunction occurring when fluxes originate from unrepresentative wind sectors or evidenced by diagnostics of sonic anemometer (SA) and gas analyzer (GA); (ii) instruments malfunction as provided by Vickers and Mahrt (1997) statistical tests; (iii) lack of well developed turbulence regimes (Foken and Wichura, 1996); (iv) violation of the stationary conditions (Mahrt, 1998). By comparing each test statistic with two pre-specified threshold values, flux data are identified as affected by severe, moderate or negligible evidences about the presence of specific sources of systematic error (hereinafter denoted as SevEr, ModEr and NoEr). Subsequently, the data rejection rule involves a two-stage procedure: in the first stage half-hourly fluxes affected by SevEr are directly discarded, whereas, in the second stage, those affected by ModEr are removed only if they are also identified as outliers. Concerning BE-Maa site, the testing period involves raw data sampled in 2019 from January 27th to May 14th. Of 5184 expected half-hourly files for NEE fluxes, only 56.7% were retained after QC routines in particular due to issues in stationarity and large gaps at the origin. A second three months period was collected at a different measurement height from January 24th to April 8th 2020 and 62% of data were retained as illustrated in Figure 10. In particular, about 3% of raw-data was originally missed, 36.7% of calculated half-hourly fluxes was discarded because affected by SevEr, while an additional 1.3% was discarded because identified as outliers and affected by ModEr.

References Foken T and Wichura B (1996) Tools for the quality assessment of surface-based flux measurements, Agric For Meterol, 78, 83-105 Mahrt L (1998) Flux sampling errors for aircraft and towers, J Atmosph Ocean Techn, 15, 416-429 Vickers D and Mahrt L (1997) Quality control and flux sampling problems for tower and aircraft data, J Atmosph Ocean Techn, 14(3), 512-526 Vitale D., Fratini G., Bilancia M., Nicolini G., Sabbatini S., Papale D (2020). A robust data cleaning procedure for eddy covariance flux measurements. Biogeosciences, 17, 1367-1391

Figure 10: Summary of the data cleaning procedure applied to the Net Ecosystem Exchange (NEE) of CO2 flux collected at BE-Maa site from 2020/01/24 to 2020/04/08. The original half-hourly flux time series is exhibited in the top panel. Panels b-f display the sequential removal of data affected by severe evidences of error according to the following criteria: (b) wind sectors to exclude and diagnostics provided by sonic anemometer (SA) and gas analyser (GA); (c) instrumental problems detection; (d) integral turbulence characteristics test (ITC, Foken and Wichura, 1996); (e) stationarity test by Mahrt (1998). Bottom panel displays the time series of retained high-quality NEE after the additional removal of outlying fluxes affected by moderate evidence of error.

Footprint analysis (Test 2) The test aims to evaluate whether half-hourly flux values are sufficiently representative of the target area (TA) or not. It was performed on about 3 months of data (77 days), after QC filtering procedures (previous Section) were achieved. The model by Klijun et al. (2015) was used to obtain the 2- dimensional flux footprint for each half-hour, which was compared to the TA spatial extent. After the QC procedure and additional filtering according to footprint model requirements, the 48.7 % of the data was used for the test.

Results showed that basically the 100 % of the whole data have a cumulative contribution of at least 70% from the TA (Figure 11, leftmost bars block), and this holds for daytime and nighttime periods too. In addition, the test was performed on 4 sub-periods of equal length and results confirmed the percentages obtained on the whole period (Figure 11).

Figure 11: Test 2 results obtained over the whole period (leftmost block) and sub-periods, showing the percentage of half-hours with a footprint cumulative contribution of at least 70% from the target area. The target value (dashed horizontal line) is that the 70% of data (half-hourly fluxes) must hold this condition. The analysis was done considering the whole day (‘24H’) and daytime and nighttime separately (‘D’ and ’N’ respectively). The footprint climatology for BE-Maa, estimated over the period under consideration is reported in Figure 12, by which it is possible to notice that the 70% footprint cumulative contribution (even 80% actually) is always included in the TA. According to these results, the test is passed.

Figure 12: Footprint climatology in relation to the TA, the EC tower (EC), and the excluded areas (EA, see the spatial sampling Section). The 50, 70 and 80 % cumulative contribution isopleths are reported. Data representativeness analysis (Test 3) This test aims to evaluate the representativeness of the possible different land cover tipologies (LCT) inside the Target area (TA). It is run on LCTs which contribute with at least 70% of the fluxes in at least 20% of the data (good data after filtering for QA/QC). For such LCTs the number of records collected during daytime, nighttime and for each of two periods obtained dividing the dataset in two parts, must be the 20% of data at least.

According to the spatial heterogeneity characterization and the ancillary plot representativeness (Test 4 Section below) at BE-Maa were defined 3 LCT (here LCT_01, LCT_03 and LCT_02). Exemplary half-hourly footprints at BE-Maa in relation to the TA and the different LCT are reported in Figure 13.

Figure 13: exemplary 2D half-hourly footprints at BE-Maa are related to the TA. The footprint 70% and 80% cumulative contribution isopleths are reported in red and blue respectively.

The test showed that the only LCT which substantially (more than 70% of cumulative contribution in more than 70% of cases) contributes to the fluxes is the LCT_01 (“old Calluna”), with more than the 70% cumulative contribution in basically all the half-hours. The other two LCTs, although often covered by the footprint contributing to the fluxes, do not reach the representativity threshold (Fig. 14).

Figure 14: Histogram of half-hourly footprint cumulative contributions from the envisaged land cover typologies (LCT) at BE-Maa. Values reported in brackets are the LCT’s average cumulative contribution to flux.

According to the results, additional analysis was not achieved and the test was considered as passed.

Ancillary plot representativeness (Test 4) The representativeness of the CPs has been evaluated by comparing each transect of CPs against the set of SP-II-order points located in the LCT that is supposed to be represented by the CP transect. The evaluation was done in terms of species cover (i.e. sum of the two main species: common heather and purple moor-grass), canopy height, and GAI. A CP transect is deemed representative when values are less than 20% different with respect to the LCT average, i.e. the average of the SP- II-order points located in the LCT. In order to account for the rather large variability of the considered variables in the target area, the representativity of each transect was also evaluated by means of a unpaired-samples, two-tailed t-test, testing whether the CP average differs significantly from the average of the SP-II-order points in the LCT.

As can be seen in Figure 15, all CP transects fall within the formal range for acceptance for the three tested variables, except for CP_11-20 (species cover and height) and CP_01-10 (height and GAI). In these cases, however, the difference between the CP transect and the LCT is never significant (p>0.1). It is hence concluded that the transects CP_21-30 and CP_31-40 are representative of LCT1 and that the transects CP_01-10 and CP_11-20 are representative of LCT3. Given that LCT2 lacks a CP transect, the ETC suggests to the station team to move CP_11-20 from LCT3 to LCT2 before the start of the measurements in the 2020 growing season.

Figure 15: Comparison of CP transects with LCTs in terms of a) species cover: Calluna vulgaris + Molinia caerulea, b) height, and c) GAI. Shown are averages. Error bars indicate the +/- 20% range of acceptance. Numbers in italics indicate the p-level of the t-test comparing the mean of the transect with the mean of the LCT.

Near Real Time data transmission The station got the green light for submission of EC files on 20190129. The station got the green light for file BM L05_F01 on 20190322, file L03_F01 on 20190430, file L06_F01 and L06_F02 on 20190822. The logger of radiation measurements broke and was replaced by one of the same model: ETC allowed this, and updated the BADM system accordingly. The NRT data submission is working properly for the files with the green light. An issue on the CP side caused some files to be temporarily missing, then fixed.

Plan for remaining variables Soil sampling The ETC and the station team have agreed that the soil sampling should either be done in summer or autumn 2020 or in spring 2021. Extension to 2021 is considered by ETC and the PI to cope with the delays and complications linked with the current COVID619 crisis.

Above Ground Biomass The ETC and the station team have agreed to continue the efforts started in 2019 to test and calibrate a non-destructive method to measure AGB. The candidate method is spectral reflectance. The calibration of this method for the BE-Maa vegetation shall be completed in the course of the 2020 growing season and a first assessment of the target area AGB shall be done after the end of the growing season in autumn 2020.

Labelling summary and proposal On the basis of the activities performed and data submitted and after the evaluation of the station characteristics, the quality of the data and setup, the compliance of the sensors and installations and the team capacity to follow the ICOS requirements for ICOS Ecosystem Stations we recommend that the station Maasmechelen (BE-Maa) is labelled as ICOS Class2 Ecosystem station. The AGB measurements have been postponed to 2020 for the definition of the method to apply and the ETC will ensure their realization.

Dario Papale, ETC Director May 4th 2020