de.hamburg.icdc/amd.de.fval/fub.messi.l2.isd en SAMD v2.2, https://doi.org/10.25592/uhhfdm.9902 FESSTVAL fval_fub_messi_l2_isd Meteorological network observations by MESSI weather stations during FESSTVaL 2021 https://doi.org/10.25592/uhhfdm.17022 Sobottke, Vincent; Rust, Henning; Goeber, Martin; Boettcher, Christopher; Lehmke, Jonas Freie Universität Berlin; Deutscher Wetterdienst, Meteorological Observatory Lindenberg – Richard-Aßmann-Observatory, Lindenberg, Germany CF-1.7 where applicable external NETCDF4 de 2.5 daily 1 This data is subject to the FESSTVaL data policy to be found at https://doi.org/10.25592/uhhfdm.9890 and in the SAMD Observation Data Product Standard https://doi.org/10.25592/uhhfdm.9902. Freie Universität Berlin Henning Rust henning.rust@fu-berlin.de 2021-06-10 2021-09-30 60 none none 56 postions within bounding box between 52° 5' 31.2'' N / 13° 43' 28.56'' E and 52° 18' 4.68'' N / 14° 25' 59.519'' E. Altitude between 38m and 98m above MSL. Clusters in Bad Saarow, Lindenberg and Beeskow. The data set contains meteorological data from a citizen science network that was part of the Field Experiment on Sub-mesoscale Spatio-Temporal Variability in Lindenberg (FESSTVaL) from May to September 2021. The network consists of 56 stations that were mostly set up in gardens of citizen participants of the project around the Meteorological Observatory Lindenberg - Richard-Aßmann-Observatory (MOL-RAO eastern Germany; 52.16°N, 14.12°E). A map of the positions of the stations is attached (png- and html-version). One of the goals was to investigate to what extent a data set gathered from a citizen science network can add value in addition to the network of the MOL-RAO and professional networks in general. The main subjects of investigation were sub-mesoscale structures such as cold pools. The 56 stations were equipped with devices called MESSI, which is a low-cost, autonomous device planned and built at Freie Universität Berlin for the use case of building meteorological Citizen Science networks. The MESSI measures air temperature, air pressure, surface downwelling illuminance and humidity (humidity data is not being published due to issues with the respective sensor during the campaign). The original time resolution was 10 seconds. As a part of the data processing a statistical model was used to establish a relation to a reference station and use this to adjust the measurements such that bias and conditional bias with respect to a DWD reference is reduced. For this we used reference data of a 1-minute resolution from a ventilated station at MOL-RAO provided by Deutscher Wetterdienst (DWD, no final quality control applied) and similarly aggregated observation data of one particular MESSI position next to the DWD device. This model was applied to all MESSI data to get more accurate values for the air temperature than measured directly. As another step of preprocessing several quality controls were applied. All measurements that did not pass these quality controls have been excluded from the dataset (values set to missing values). At some stations the MESSI had to be exchanged during the campaign due to technical issues. In some cases it can not be guaranteed that the device was measuring at the respective station position (possible untracked movement of device by citizen participant). The MESSIs did send their GPS-position to the MESSI database via LoRaWan and the The Things Network once per day. But in some cases this data transfer failed, leading to some uncertainty of the station position. These and other information can be found in station_list_messi_fesstval.csv The MESSI device and data processing including the statistical modelling and the applied quality controls is described in more detail in [http://dx.doi.org/10.17169/refubium-42772]. Two versions are being published - v00 / v01 (as indicated in the file names): v00 is the data with the original 10-sec time resolution. The script that applied statistical models and quality controls to the aggregated 1-minute resolution data was applied to this higher resolution dataset. Since the script was optimized for 1-minute data, there is no guarantee that all filters and especially the statistical model (trained on 1-minute data) is optimal for the 10 seconds resolution data. v01 is the version with time resolution aggregated to 1 minute. The script for quality control and statistical modelling has been optimized for this time resolution. The data of all network stations is stored in daily files separated by station type and measurement variable. The variable are: ta - Air temperature inside device ta2 - Air temperature outside device ta3 - Air temperature derived from statistical model pa - Unnormalized air pressure at station location isd - Surface downwelling illuminance The total data availability is 53%. This is because of the filters mentioned above, failures in data transfer and mainly the fact that measurements started and ended at different times for the different stations. Best data availability is given in July and August (72% and 73%). Absolute accuracy according to manufacturers of sensors: ta/ta2: ±0.15K, pa: ±50Pa The actual accuracy is probably lower due to assembly of the devices by the participants and other factors. In the modelling procces a root-mean-square-error of 0.64K was estimated when comparing the ta-value (Air temperature inside device) of the MESSI with the reference sensor. MESSI (Mein Eigenes SubSkalen Instrument) 52° 5' 31.2'' N - 52° 18' 4.68'' N 13° 43' 28.56'' E - 14° 25' 59.519'' E 38-98m none none 56 2 time ('time',) time seconds since 1970-01-01 00:00:00 UTC time_bnds ('time', 'nv') start and end of time averaging intervals seconds since 1970-01-01 00:00:00 UTC station_id ('station',) station ID lat ('station',) latitude latitude of instrument location degrees_north lon ('station',) longitude longitude of instrument location degrees_east zsl ('station',) altitude altitude of instrument location above mean sea level m zag ('station',) height height of sensor above ground m isd ('time', 'station') surface downwelling illuminance lx