The microwave radiometer HATPRO has seven K-band (22.235-31.4 GHz) channels that are sensitive to the column amounts of cloud liquid water (clwvi) and water vapor (precipitable water, prw). The files in this data set contain quality-controlled, 10-minute-resolved statistics (mean, median, standard deviation, percentiles) for zenith clwvi and prw, which have been calculated from the original level 2 HATPRO data having a temporal resolution of about 2–3 seconds (Nomokonova et al., 2019; Ebell et al., 2022). prw and clwvi are based on multi-variate regression retrievals (see also Löhnert and Crewell, 2003; doi:10.1029/2002RS002654; Nomokonova et al, 2019; doi:10.5194/acp-19-4105-2019). To calculate the 10-minute-resolved statistics, only prw and clwvi data that have passed certain quality checks are considered. In particular, during and after a rain event, measurements can be unreliable due to a wet radome of the instrument. Within a 30-minute moving window, it is thus checked whether liquid precipitation is detected and whether spectral inconsistencies occur in the brightness temperature spectrum of the seven frequencies (Chellini et al., 2023). If this is the case, the whole 30-minute window is excluded from the analysis. If no liquid precipitation is present, only data points with spectral inconsistencies are removed. Liquid precipitation is assumed when the HATPRO weather station reports temperatures above 3.6 K (as all observed precipitation is expected to be liquid only; Ebell et al., 2025) and the precipitation sensor indicates precipitation (i.e., the rain flag is set in the original data). Finally, sudden jumps in prw that are remaining after the above-mentioned quality check are removed. The quality-checked prw time series is smoothed with a 30-minute moving window and then subtracted from itself. If the difference between individual values is equal to or higher than 0.7 kg/m² (reflecting sudden jumps in prw), these are removed. The threshold value of 0.7 kg/m² was set after a thorough assessment of these cases. The corresponding clwvi values affected by these jumps are also removed. prw and clvwi statistics are only calculated if at least 50% of the original data within the 10-minute time interval is reliable.