CAMELS-LUX: Highly Resolved Hydro-Meteorological and Atmospheric Data for Physiographically Characterized Catchments around Luxembourg

CAMELS-LUX: Dataset description Judith Nijzink1,2, Davide Zoccatelli1, Laurent Gourdol1, Jean François Iffly3, Ralf Loritz4, and Laurent Pfister1,2 1Catchment and Ecohydrology Group (CAT), Environmental Research and Innovation, Luxembourg Institute of Science and Technology (LIST), Belvaux, 4422, Luxembourg 2Faculty of Science, Technology and Medicine (FSTM), University of Luxembourg, Esch-sur-Alzette, 4365, Luxembourg 3Observatory for Climate, Environment and Biodiversity (OCEB, Environmental Research and Innovation, Luxembourg Institute of Science and Technology (LIST), Belvaux, 4422, Luxembourg 4Institute of Water and Environment, Karlsruhe Institute of Technology (KIT), Karlsruhe, 76131, Germany Correspondence: Judith Nijzink ([email protected]) Contents 1 Introduction 2 2 Time series data 2 2.1 Generalinformation................................................. 2 2.2 Hydrologicparameters ............................................... 35 2.3 Air temperature and potential evapotranspiration parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.4 Thunderstorm relevant atmospheric parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.5 SoilMoistureparameters.............................................. 6 2.6 Precipitationparameters .............................................. 6 3 Static catchment attributes 810 3.1 Metacatchmentattributes.............................................. 8 3.2 Climaticcatchmentattributes............................................ 11 3.3 Geologiccatchmentattributes ........................................... 12 3.4 Landusecatchmentattributes ........................................... 12 3.5 Topographiccatchmentattributes.......................................... 1315 4 Spatial data - shapefiles 14 1 1 Introduction The CAMELS-LUX dataset encompasses hydro-meteorological time series and catchment attributes for 56 partly nested stream gauges feeding into the Luxembourgish stream network. The data is available at three temporal resolutions: daily, hourly and at a 15 minute resolution and spans the hydrological years from 2004-11-01 to 2021-10-31. The static catchment attributes cover20 parameters classifying the topography, geology and land use as well as climatic and hydrologic annual statistics of the 17-year time period. While an in depth description of the dataset as well as background information on catchments, the environment and exact calculation methods is provided in the accompanying publication, this dataset description isolates information on the available parameters and data structure contained in the provided files.25 Please note that the dataset might not include data corrections or validations that are subject to a date later than the date of the retrieval of the data for the processing of this dataset. The data was last retrieved and updated for version two of CAMELS-LUX in October 2025. 2 Time series data 2.1 General information30 Dynamic time series are available in the folder timeseries. This folder is split into the three subfolders separating the three different time steps to which the data have been aggregated: 15Min,daily and hourly. Each of these folders contains one .csv-file per catchment, ID 1 to ID 56, named CAMELS_LUX_hydromet_timeseries_ID_<ID>.csv. The time series data are in the time zone UTC+1 with a date format of "YYYY-mm-dd" for the daily data and "YYYY-mmdd HH:MM:SS" for the hourly and 15 minute data. The time stamps are set at the end of accumulation periods. For each time35 step ("Date"), 26 parameters are provided: three hydrologic parameters (Subsection 2.2), six parameters related to precipitation (Subsection 2.6), two parameters for air temperature (Subsection 2.3), two for potential evapotranspiration (Subsection 2.3), nine parameters to characterize thunderstorm prone atmospheric conditions (Subsection 2.4) and eventually four parameters to characterize soil moisture at four different depths (Subsection 2.5). 2 2.2 Hydrologic parameters40 Discharge data is provided at 56 partly nested stream gauges that contain reasonably long time series and contribute to the Luxembourgish stream network. This stream network extends over most of Luxembourg as well as areas on the Belgian and French side of Luxembourg, and streams on the German side (Fig. 1). Discharge (Q) is provided in m3s-1 as well as normalized to the specific discharge (Qspec) in mm per respective time step. Additionally, we provide a flag column (Q_flag) that indicates interpolated values (Table 1). The original time step of the discharge measurements is 15 minutes. The observations were45 aggregated to an hourly and daily temporal resolution. In most catchments, data is available from 2004-11-01 – 2021-10-31. Ten catchments (ID 12, 20, 31, 36, 37, 38, 46, 54, 55, 56) do not have data for the entire time period but mostly start one or two years later. Precise start and end dates of each time series are noted in the meta-attributes file (Subsection 3.1). Two time series are significantly shorter (ID 55-56). These catchments were nonetheless included in the dataset, as flash floods were observed there.50 Discharge data in Luxembourg is measured at stations of the Luxembourgish Water Agency (AGE) and the Luxembourg Institute of Science and Technology (LIST). The discharge stations on the German side are maintained by the LfU-RLP - Rhineland-Palatine State Office for Environment (2022). The exact mapping of stations to agencies is listed in the meta attributes (Subsection 3.1). All gaps in the discharge time series were filled. After individual examination of the data gaps, they were compared to55 neighbouring stream gauges. For each data gap an individual decision was taken following the "guidelines": If the gaps only last a few time steps, in which the discharge was in a recession period and no or barely any precipitation was observed, data was interpolated linearly. If streamflow was not in the recession period and precipitation was recorded, the specific discharge was transferred from a surrounding stream gauge, that reacted similar to the stream gauge around the time of the data gap. The stream gauges used to fill the data gaps are listed in Supplement S2 of the accompanying publication. The discharge60 (Q [m3s-1]) was then recalculated according to the basin area. In the CAMELS-LUX dataset, a flag column is provided to indicate interpolated values. “0” represents original measurements, “1” indicates linear interpolation. All other numbers represent specific surrounding stream gauges. Table 1. Hydrologic time series in the CAMELS-LUX dataset. The start end end dates for each catchment are listed in the meta-attribute file. Proxy for Parameter Abbr. Unit Level Source Hydrology Discharge Q m3s-1 Gauge AGE - Luxembourgish Water Agency (2025); LIST - Luxembourg Institute of Science and Technology, Hydro-Climatological Observation network (HOST) (2025); LfU-RLP - Rhineland-Palatine State Office for Environment (2022) Specific discharge Qspec mm Gauge calculated Flag for Q Q_flag - - determined 3 2.3 Air temperature and potential evapotranspiration parameters Air temperature data (T_stn [°C]) was extracted with a 10 minute resolution from 67 stations run by AGE - Luxembourgish65 Water Agency (2025); ASTA - Luxembourgish Ministry of Agriculture (2025); DWD - German Weather Service (2025b); LIST - Luxembourg Institute of Science and Technology, Hydro-Climatological Observation network (HOST) (2025). The data was spatially interpolated using Voronoi polygons and clipped to the catchments. It was moreover aggregated to meet this datasets temporal resolution. A list of the stations is enclosed in the accompanying publication’s Supplement S1. The air temperature (t2m [°C]) is extracted from the ERA5 reanalysis dataset (Hersbach et al., 2023b). The 0.25° × 0.25°70 grid cells were clipped to the catchments and a weighted mean was built. The ERA5 data is available at an original hourly resolution and was dis-/aggregated to the 15 minute and daily time steps. For potential evapotranspiration (PET) two time series are provided: PET_Oudin [mm] was calculated based on the air temperature based method by Oudin et al. (2005). Moreover, PET_PM [mm] is the potential evapotranspiration calculated according to the Penman-Monteith approach (Allen et al., 1998; Penman, 1947; Monteith, 1965). The equations’ variable75 parameters are all derived from the ERA5 reanalysis datasets (Hersbach et al., 2023a, b). Table 2. Air temperature and potential evapotranspiration parameters in the CAMELS-LUX dataset. Proxy for Parameter Abbr. Unit Level Source Temperature Air temperature t2m °C single, 2m Hersbach et al. (2023b) Air temperature T_stn °C Station AGE - Luxembourgish Water Agency (2025); LIST - Luxembourg Institute of Science and Technology, HydroClimatological Observation network (HOST) (2025); ASTA - Luxembourgish Ministry of Agriculture (2025); DWD - German Weather Service (2025b) Water balance Potential evapotranspiration PET_Oudin mm - Calc. according to Oudin et al. (2005) Potential evapotranspiration PET_PM mm - Calc. according to (Allen et al., 1998; Penman, 1947; Monteith, 1965) 4 2.4 Thunderstorm relevant atmospheric parameters According to Meyer et al. (2022), proxy parameters were chosen to characterize thunderstorm relevant atmospheric conditions: Sufficient atmospheric instability, high atmospheric moisture and low wind speeds and shear winds. The atmospheric parameters (Table 3) are derived on either single levels or at different pressure levels from the ERA5 reanalysis datasets (Hersbach80 et al., 2023a, b). The proxy parameters are listed in Table 3. The original spatial resolution of the ERA5 reanalysis parameters is 0.25° × 0.25° and the data was averaged respecting the overlap of the catchments with the ERA5 grid cells. This coarse grid as well as the fact that the data is extracted from a global model limits the data quality despite the data being assimilated to measurements and regridded. From the original temporal resolution of hourly time steps, the data was linearly disaggregated to the 15 minute interval and aggregated to the daily time step. The values at the full hour mark the original values.85 The wind speed (WS10 m a.g.l.-500 hPa) and shear winds (LLS, DLS [m s-1]) were calculated based on the wind vectors uand v provided in the ERA dataset. The wind speed is the mean between the wind speeds at 10 m a.g.l. and at the pressure level of 500 hPa. The wind shear is the difference between the wind speed at 10 m a.g.l. and at the pressure level of 850 hPa and 500 hPa respectively.90 Table 3. Atmospheric parameters in the CAMELS-LUX dataset. Proxy for Parameter Abbr. Unit Level Source Instability Convective available potential energy cape J kg-1 single Hersbach et al. (2023b) Convective inhibition cin J kg-1 single Hersbach et al. (2023b) K-index kx °C single Hersbach et al. (2023b) Moisture Total column water vapour tcwv kg m-2 single Hersbach et al. (2023b) Specific humidity q kg kg-1 700 hPa Hersbach et al. (2023a) Relative humidity rh % 700 hPa Hersbach et al. (2023a) Storm motion Wind speed ws10500 m s-1 10 m & 500 hPa Calculated from and organisation Low-level wind shear lls m s-1 10 m & 850 hPa Hersbach et al. (2023a) Deep-layer wind shear dls m s-1 10 m & 500 hPa 5 2.5 Soil Moisture parameters The soil moisture data was extracted from the ERA5-Land data (Muñoz Sabater, 2019). It is available at a spatial resolution of 0.1° × 0.1°. The grids were clipped to the catchments and averaged according to the percentage of the cell within the catchments. The original hourly temporal resolution was disaggregated and aggregated in order to calculate the 15 minute and daily data. The soil moisture is provided at four different depths as indicated in Table 4. Note, that also the ERA5-Land data is95 extracted from a global model and its quality is limited to tendencies. Yet, we found the data to be sufficiently accurate on the catchment scale. Table 4. Soil moisture parameters in the CAMELS-LUX dataset. Proxy for Parameter Abbr. Unit Level Source Catchment wetness state Volumetric soil water layer 1 swvl1 m3m-3 0-7 cm Muñoz Sabater (2019) Volumetric soil water layer 2 swvl2 m3m-3 7-28 cm Muñoz Sabater (2019) Volumetric soil water layer 3 swvl3 m3m-3 28-100 cm Muñoz Sabater (2019) Volumetric soil water layer 4 swvl4 m3m-3 100-289 cm Muñoz Sabater (2019) 2.6 Precipitation parameters The time series files contain different rainfall products complementing each other (Table 5). The allegedly most accurate rain fall product is the rain rate based on radar data (RR_rad [mm]). The radar data is extracted from the radar product RadKlim100 (“Radar-based Precipitation Climatology” - YW) by the German Weather Service (DWD, version 2017.002 - Winterrath et al., 2018). RadKlim is a reprocessed and quasi gauge-adjusted version of the DWD’s operational Radolan (“Radar-OnlineAdjustment”) product. A decent quality is therefore assumed. That data has a temporal resolution of 5 minutes and was accumulated to this dataset’s 15 minute, hourly and daily time step with the time stamps at the end of the acumulation interval. Apart from the aggregated rain rate, we extracted the maximum and the minimum rain rate (RR_max_rad, RR_min_rad) within105 one grid cell (1 km x 1 km) and one time step (in the original 5 minute resolution). The 15 minute, hourly and daily datasets still contain the minimum and maximum rain rate per 5 minutes within the respective time steps. The radar data was averaged per catchment. A 1 km × 1 km sized grid cell was counted to belong to a catchment if its center point lies within the catchment area. As the Weierbach catchment (ID 50) is very small (0.4575 km2) it does not contain any grid cell center. In this case, the four surrounding grid cells were averaged. In case missing values persisted in the final precipitation radar time series per110 catchment, these data gaps were filled with interpolated station data (RR_stn [mm]). To fill gaps of the 5 minute minimum and maximum rain rates in the dataset, the minimum and maximum of the station data are taken and calculated from a further disaggregation. While they do not represent the real rainfall intensities, this disaggregation was all we could extract from the station data. All replaced values within the radar precipitation time series are indicated as "1" within a flag column (P_flag). 6 In addition, we added spatially interpolated station data as an extra time series (RR_stn [mm]). The station data encom-115 passes a network of 75 stations (listed in Supplement S1 of the accompanying publication) that are spread over Luxembourg and Germany. The stations are maintained by the AGE - Luxembourgish Water Agency (2025), the ASTA - Luxembourgish Ministry of Agriculture (2025), the LIST - Luxembourg Institute of Science and Technology, Hydro-Climatological Observation network (HOST) (2025), MeteoLux (2025) or the DWD - German Weather Service (2025a). The original data is available at differing temporal resolutions: the stations maintained by AGE and LIST at a 15 minute resolution, the stations by ASTA at120 a 10 minute resolution, the DWD-stations at a 5 minute resolution and the MeteoLux station at a 1 minute resolution. To reach the 15 minute, hourly and daily temporal resolutions of CAMELS-LUX, the data was aggregated. The stations were spatially interpolated using Voronoi polygons and afterwards clipped to the catchments. Lastly, the total precipitation (tp [mm]) of the ERA5 reanalysis data (Hersbach et al., 2023b) was added as a parameter. Its original temporal resolution is one hour. The data was dis-/aggregated respectively to reach the 15 minute and daily time steps.125 Table 5. Precipitation time series in the CAMELS-LUX dataset. Proxy for Parameter Abbr. Unit Level Source Precipitation Rain rate (radar) RR_rad mm Radar Winterrath et al. (2018), gaps filled with station data (RR_stn) Min. rain rate (radar) RR_min_rad mm 5Min-1 km-2 Radar Max. rain rate (radar) RR_max_rad mm 5Min-1 km-2 Radar Flag for P RR_flag_rad - - Rain rate (station) RR_stn mm Station AGE - Luxembourgish Water Agency (2025); ASTA - Luxembourgish Ministry of Agriculture (2025); LIST - Luxembourg Institute of Science and Technology, Hydro-Climatological Observation network (HOST) (2025); DWD - German Weather Service (2025a) Rain rate (ERA5) tp mm single Hersbach et al. (2023b) 7 3 Static catchment attributes A general overview of the catchments’ meta data is given in Subsection 3.1. The static catchment attributes further characterize the catchments by climatological (Subsection 3.2), geologic (Subsection 3.3), land use (Subsection 3.4) and topographic (Subsection 3.5) attributes. For each category there is one .csv-file named CAMELS_LUX_<category>_attributes.csv. 3.1 Meta catchment attributes130 The file CAMELS_LUX_meta_attributes.csv contains the catchments’ meta data including the parameters listed in Table 7. A list of catchments is added below in Table 6 and the catchments are mapped in Fig. 1. Figure 1. The 56 nested catchments of the CAMELS-LUX dataset grouped in a few basins of higher order. Flash flood catchments are the two catchments in which a flash flood was observed. 8 Table 6. List of basins. ID Gauge Stream 1 Livange Alzette 2 Hesperange Alzette 3 Pfaffenthal Alzette 4 Hunsdorf Alzette 5 Mersch Alzette 6 Ettelbruck Alzette 7 Reichlange Attert 8 Useldange Attert 9 Bissen Attert 10 Livange Bibeschbach 11 Kautenbach Clerve 12 Hesperange Drosbech 13 Bettembourg Dudelingerbach 14 Hagen Eisch 15 Hunnebour Eisch 16 Sinspelt Enz 17 Huewelerbach Huewelerbach 18 Gemund Irsen 19 Kayl Kaylbach 20 Merkholtz Kirel 21 Mamer Mamer 22 Schoenfels Mamer 23 Huncherange Mierbech 24 Giesdorf Nims 25 Seffern Nims 26 Alsdorf-Oberecken Nims 27 Dasbourg Our 28 Gemund Our ID Gauge Stream 29 Vianden Our 30 Niederpallen Pall 31 Luxembourg Petrusse 32 Prüm Prüm 33 Echtershausen Prüm 34 Prümzurlay Prüm 35 Platen Roudbach 36 Cessange Ruisseau de Cessange 37 Colpach Rau de Colpach 38 Merl Ruisseau de Merl 39 Useldange Schwebich 40 Bigonville Sure 41 Heiderscheidergrund Sure 42 Michelau Sure 43 Diekirch Sure 44 Bollendorf Sure 45 Rosport Sure 46 Mertert Syre 47 Niederfeulen Wark 48 Welscheid Wark 49 Ettelbruck Wark 50 Weierbach Weierbach 51 Winseler Wiltz 52 Kautenbach Wiltz 53 Useldange Wollefsbach 54 Bous Aalbach 55 Larochette Ernz Blanche 56 Beaufort Hallerbach 9 IGN - Institut National de l’Information Géographique et Forestière: RGE ALTI: Le modèle numérique de terrain (MNT) maillé qui décrit le relief du territoire français à grande échelle, https://geoservices.ign.fr/rgealti, accessed: 2024-05-25, 2020. IT-NRW - Information und Technik Nordrhein-Westfalen: Digitales Geländemodell - Gitterweite 1 m - Paketierung: Einzelkacheln, https: //www.opengeodata.nrw.de/produkte/geobasis/hm/dgm1_xyz/dgm1_xyz/, accessed: 2024-05-25, 2021.195 LfU-RLP - Rhineland-Palatine State Office for Environment: Discharge data Rhineland-Palatine, https://wasserportal.rlp-umwelt.de, accessed: 2022-03-02, 2022. LIST - Luxembourg Institute of Science and Technology, Hydro-Climatological Observation network (HOST): Hourly Precipitation, 15minute Discharge, https://envdata.private.list.lu/, accessed: 2025-10-20, 2025. LVG - Landesamt für Vermessung und Geobasisinformation: Digitale Geländemodelle - Vermessungsund Katasterverwaltung Rheinland-200 Pfalz, https://lvermgeo.rlp.de/geodaten-geoshop/open-data, accessed: 2024-05-25, 2021. Mardhel, V., Pinson, S., and Allier, D.: Description of an indirect method (IDPR) to determine spatial distribution of infiltration and runoff and its hydrogeological applications to the French territory, Journal of Hydrology, 592, https://doi.org/10.1016/j.jhydrol.2020.125609, 2021. MECB - Ministère de l’Environnement, du Climat et du Développement durable & MEA - Ministère de l’énergie et de l’Aménagement du205 territoire: LIS-L Land Use 2018, https://data.public.lu/fr/datasets/lis-l-land-use-2007-2015-2018/, accessed: 2024-05-25, 2018. MeteoLux: precipitation data, air temperature, https://www.meteolux.lu/de/klima/wetterstation/, accessed: 2025-10-20, 2025. Meyer, J., Neuper, M., Mathias, L., Zehe, E., and Pfister, L.: Atmospheric conditions favouring extreme precipitation and flash floods in temperate regions of Europe, Hydrology and Earth System Sciences, 26, 6163–6183, https://doi.org/10.5194/hess-26-6163-2022, 2022. Monteith, J. L.: Evaporation and environment, Symposia of the Society for Experimental Biology, 19, 205–234, 1965.210 Muñoz Sabater, J.: ERA5-Land hourly data from 1981 to present, Copernicus Climate Change Service (C3S) Climate Data Store (CDS), https://doi.org/10.24381/cds.e2161bac, 2019. Oudin, L., Hervieu, F., Michel, C., Perrin, C., and Andréassian, V.: Which potential evapotranspiration input for a lumped rainfall–runoff model? Part 2 – Towards a simple and efficient potential evapotranspiration model for rainfall–runoff modelling, Journal of Hydrology, 303, 290–306, https://doi.org/10.1016/j.jhydrol.2004.08.026, 2005.215 Penman, H. L.: Natural evaporation from open water, bare soil and grass, https://royalsocietypublishing.org/, 1947. Sappington, J. M., Longshore, K. M., and Thompson, D. B.: Quantifying Landscape Ruggedness for Animal Habitat Analysis: A Case Study Using Bighorn Sheep in the Mojave Desert, The Journal of Wildlife Management, 71, 1419–1426, https://doi.org/10.2193/2005-723, 2007. SPW - Service Public de Wallonie: Relief de la Wallonie - Modèle Numérique de Terrain (MNT) 2013-2014, https://geoportail.wallonie.be/ catalogue/6029e738-f828-438b-b10a-85e67f77af92.html, accessed: 2024-05-25, 2014.220 Winterrath, T., Brendel, C., Hafer, M., Junghänel, T., Klameth, A., Lengfeld, K., Walawender, E., Weigl, E., and Becker, A.: RADKLIM Version 2017.002: Reprocessed quasi gauge-adjusted radar data, 5-minute precipitation sums (YW), Tech. rep., German Weather Service (DWD), https://doi.org/10.5676/DWD/RADKLIM_YW_V2017.002, 2018. 16