Extended series. Nevertheless, 80.eight on the changepoints are related (i.e., inside 62 days). Figure

Extended series. Nevertheless, 80.eight on the changepoints are related (i.e., inside 62 days). Figure eight shows the instance of station VILL (Villafranca, Spain), where the imply noise and stdf are very similar for the twotime series, but the segmentation benefits are fairly different: six changepoints are detected in the short time series and two inside the extended time series. Two changepoints are comparable in both series and are validated. The further changepoints identified inside the brief time series capture quick but considerable variations inside the mean. These changepoints are not retained inside the extended time series simply because other such variations are seen all along within the extended time series. The penalty criterion avoids choosing all these changepoints, as well as the final optimal resolution sooner or later has only one changepoint. This remedy appears additional affordable. three.1.3. Influence of the Reference Data Set The third section of Table two summarizes the segmentation final results when either ERAI or ERA5 is employed as a reference, i.e., the segmentation is run for CODEERAI or CODEERA5 IWV differences, when exactly the same auxiliary data is utilized in the GNSS ZTD to IWV conversion (here from ERA5). Globally, each the mean noise along with the stdf with ERA5 are lowered by 25 when compared with ERAI. Figure 9c,e show that the reduction in noise and stdf is observed at 95 and 75 of all stations, respectively. This difference may be explained by the reduce representativeness error in ERA5 due to greater spatial resolution, at the same time as larger top quality of the IWV temporal variations in this reanalysis, possibly due to the assimilation of much more satellite observations. Figure 10 shows one of the most striking case of increase in noise with time for ERAI. The impact on the segmentation benefits is fairly important. Only a single changepoint is detected in the extra noisy series, whilst seven changepoints are detected inside the less noisy a single. At many stations, there is also an excess noise in ERAI Myristoleic acid Apoptosis during the moist period of each and every year; see the example of station KIRU (Kiruna, Sweden) in Figure 11. At this station, the CODEERAI series has a great deal larger seasonal variations inside the noise and within the functional than CODEERA5. This results in additional changepoints in more noisy series (6 versus two) due to the sharp enhance in the noise during some years, which can be not effectively represented by the periodic bias and monthly variance. Because of this, 4 outliers are detected inside the CODEERAI segmentation benefits. Within the CODEERA5, the two changeAtmosphere 2021, 12,16 ofpoints are validated by the metadata, that is, once more, greater than within the CODEERAI results. Other examples with big modifications in stdf are stations KIT3, POL2, and SANT (Figure 9d). These stations have been also pointed as N-tert-Butyl-α-phenylnitrone Epigenetics extreme cases of representativeness errors in ERAI by Bock and Parracho [17] as a result of steep orography.Figure six. Equivalent to Figure 4, but for station GOPE (Ondrejov, Czech Republic).Atmosphere 2021, 12,17 ofFigure 7. Related to Figure 3, but comparing segmentation results from two unique lengths in the CODE data set (extended time series, from 1994 to 2018, brief time series, from 1994 to 2010). The mean noise and common deviation with the functional (stdf) are representative in the complete times series, but the quantity of breakpoints are counted inside the common period only (1994010).From Table two, we see that the total variety of changepoints is bigger when ERA5 is used as a reference. This can be understood because the consequence in the common reduce with the noise and periodic bias with this reanaly.