He elevation cutoff angle and observation weighting, which contribute drastically for the sensitivity of ZTD estimates for the many elevationdependent error sources (mapping functions, antenna PCO/PCV models, and multipath). The objective of employing a reduced elevation cutoff angle in CODE would be to contain a lot more observations, i.e., enhance the precision of your estimated parameters. Having said that, multipath is commonly higher at low elevations. To mitigate it, the reduced elevation observations are downweighted. The JPL/NASA processing technique was different as they applied a 7cutoff angle and no downweighting. Possibly, this approach might be additional sensitive to multipath and anomalous propagation effects at low elevations. The ZTD estimates from each data sets were screened for outliers following the methodology described in Bock [41] and Stepniak et al. [42], and converted to IWV working with either ERAInterim or ERA5 reanalysis. The 6hourly IWV information had been in comparison to the reanalysis IWV information and further screened for outliers (for every station, the IWV variations exceeding the median five typical deviations had been removed). Afterward, the IWV values from GNSS and reanalysis, along with the IWV Redaporfin In Vivo differences amongst GNSS and reanalysis, have been aggregated into day-to-day and monthly estimates and created publicly accessible around the AERIS data center [43,44]. Figure 1 shows the location from the GNSS station obtainable in the two data sets. In this study, we selected 81 frequent stations, for which the time series in each information sets covered a period of a minimum of 15 years.Figure 1. Map in the GNSS stations out there from the two reprocessed data sets: IGS repro1 (empty circles), CODE REPRO2015 (modest dots), as well as the 81 common stations (complete circles) made use of within this study.two.two. Reference IWV Information Our homogenization process operates on IWV differences involving a GNSS series and also a reference series. Because the IGS network is pretty sparse, we can not use a nearby station as is normally carried out by climatologists (as in Venema et al. [9]). Alternatively, for everyAtmosphere 2021, 12,8 ofGNSS station, a series of IWV from every of the two reanalyses is derived, and each day IWV differences are formed, as explained above. In earlier studies, we located that ERAInterim and GNSS IWV had significant representativeness differences in Antarctica and in regions of steep topography (Andes, Himalayas, and so forth.) or close to the oceans [17]. Within this study, we are going to investigate the impact of representativeness errors on the segmentation final results by comparing the results in the two reanalyses. The spatial resolution from the reanalyses is 0.750.75for ERAInterim and 0.250.25for ERA5. Decreased representativeness errors are, as a result, expected from ERA5 data. Additionally, the IWV values computed from ERA5 are also anticipated to be of higher good quality because this reanalysis employed a far more current model and Ganoderic acid DM custom synthesis assimilation program, and assimilated a much larger quantity of observations, specifically in current years [18]. 2.three. Homogenization Approach Figure two shows the data flow chart starting together with the GNSS ZTD information and ending with all the corrected IWV series. The first two steps (Conversion and Comparison) are described in the previous subsection.Figure 2. Flowchart of the basic homogenization procedure.The third step would be the segmentation, i.e., the detection of changepoints in the imply of the IWV difference time series. Here, we use the rapidly version of the GNSSfast R package published by Quarello et al. [45]. This version is readily available on https://github.com/arq1 6/GNSSfast.gi.
