We assess the performance of different break detection methods on three sets of benchmark data sets, each consisting of 120 daily time series of integrated water vapor differences. These differences are generated from the Global Positioning System (GPS) measurements at 120 sites worldwide, and the numerical weather prediction reanalysis (ERA-Interim) integrated water vapor output, which serves as the reference series here. The benchmark includes homogeneous and inhomogeneous sections with added nonclimatic shifts (breaks) in the latter. Three different variants of the benchmark time series are produced, with increasing complexity, by adding autoregressive noise of the first order to the white noise model and the periodic behavior and consecutively by adding gaps and allowing nonclimatic trends. The purpose of this "complex experiment" is to examine the performance of break detection methods in a more realistic case when the reference series are not homogeneous. We evaluate the performance of break detection methods with skill scores, centered root mean square errors (CRMSE), and trend differences relative to the trends of the homogeneous series. We found that most methods underestimate the number of breaks and have a significant number of false detections. Despite this, the degree of CRMSE reduction is significant (roughly between 40% and 80%) in the easy to moderate experiments, with the ratio of trend bias reduction is even exceeding the 90% of the raw data error. For the complex experiment, the improvement ranges between 15% and 35% with respect to the raw data, both in terms of RMSE and trend estimations.

With the purpose of revising World Meteorological Organization's Commission for Instruments and Methods of Observation (WMO/CIMO) Guide #8 on weather stations siting, an experiment to evaluate metrologically the maximum influence of a paved road on 2-m air temperature measurements ("road siting effect") has been designed, installed and run in Italy. It consists of a 100-m long array of seven measurement stations, at increasing distances from a local road, equipped with shielded Pt100 thermometers and ancillary sensors.Data coming from 1 year of observations have been analysed for daily climatological metrics, finding that the road mostly effects minimum temperatures, with average values of $0.30 ± 0.18 C at a distance of 1 m; then, in order to quantify the instrumental effect on the measurement, data were filtered by applying a generalized additive model, selecting only times when the effect is more intense (during nights, in presence of low winds coming from the road), and the road siting effect has been calculated by modelling the maximum temperature differences by using extreme values analysis. The 1-year return value on 10-min measurements is 1.22 ± 0.30 C at 1 m from the road, with a gradual decline ($0.1 CÁm −1 ), while an extrapolation to 100-year return level gives a value of 1.71 ± 0.79 C with a decline rate of about 0.17 CÁm −1 . This is a first step towards a redefinition of the weather station classification scheme of WMO/CIMO Guide #8, together with building and tree effects experiments which have been run in parallel with the road siting experiment here presented and which will be presented elsewhere.

This study investigates the sensitivity of the GNSSseg segmentation method to change in: GNSS data processing method, length of time series (17 to 25 years), auxiliary data used in the integrated water vapor (IWV) conversion, and reference time series used in the segmentation (ERA-Interim versus ERA5). Two GNSS data sets (IGS repro1 and CODE REPRO2015), representative of the first and second IGS reprocessing, were compared. Significant differences were found in the number and positions of detected change-points due to different a priori ZHD models, antenna/radome calibrations, and mapping functions. The more recent models used in the CODE solution have reduced noise and allow the segmentation to detect smaller offsets. Similarly, the more recent reanalysis ERA5 has reduced representativeness errors, improved quality compared to ERA-Interim, and achieves higher sensitivity of the segmentation. Only 45–50% of the detected change-points are similar between the two GNSS data sets or between the two reanalyses, compared to 70–80% when the length of the time series or the auxiliary data are changed. About 35% of the change-points are validated with respect to metadata. The uncertainty in the homogenized trends is estimated to be around 0.01–0.02 kg m−2 year−1.