Original scientific paper For all scientific studies, for processing analytical data the subject of outliers is of critical importance. The accuracy of Total Electron Content (TEC) models obtained from ground based Global Positioning Systems (GPS) data is strongly affected by some factors such as satellite geometry, cut-off angle, ionospheric shell height, quantity of GPS data etc. Probable outliers within the observables have a significant effect on the quality of the estimated TEC modelling. These should be removed from the TEC time series in order to get more reliable results for further process. Although some analytical methods have been developed for the code-outliers, phase-slips and ionosphere disturbances, significant reduction in the number of measurement could adversely affect the quality of the generated TEC values using geometry-free combinations of those. In this study, we proposed a novel approach for outlier detection to be used at geometry-free combinations in estimation of vertical TEC without reducing the number of code observables. Here the GPS data obtained from 22 permanent stations are used to estimate the VTEC on May 14-15, 2005. The research results from the detailed processing clearly show the efficiency of the method. Our research shows that the proposed approach appears to be a fast, effective and objective method to detect outlying effects which cause improper modelling. Moreover the resulting VTEC maps from the new approach differ reliably with respect to the data including outlying observables. Consequently, the proposed method could provide an alternative to outlier rejection in estimation of VTEC, in which outlying observations are retained in geometry-free combinations.
Keywords: discordant data; GPS; median; outlier; total electron content
Novi pristup vanjskim (outlying) podacima u procjeni vertikalnog ukupnog sadržaja elektronaIzvorni znanstveni članak U svim vrstama znanstvenih studija, za obradu analitičkih podataka pitanje outliera je od bitne važnosti. Na točnost Total Electron Content (TEC) modela koja se postiže na temelju podataka Sustava za Globalno Pozicioniranje (GPS) sa zemlje uvelike djeluju čimbenici kao što su geometrija satelita, kut prekida, visina ionosferske ljuske, količina GPS podataka i sl. Vjerojatni outlieri među promatranima imaju značajan učinak na kvalitetu procijenjenog TEC modeliranja. Njih je potrebno ukloniti iz TEC vremenskih serija kako bi se dobili pouzdaniji rezultati za daljnji postupak. Iako su neke analitičke metode razvijene za code-outliers, phase-slips i poremećaje u ionosferi, značajno smanjenje broja mjerenja negativno bi moglo utjecati na kvalitetu nastalih TEC vrijednosti primjenom njihovih kombinacija bez geometrije. U ovom istraživanju, predložili smo novi pristup za otkrivanje outliera koji će se koristiti u kombinacijama bez geometrije u procjeni vertikalnog TEC-a bez smanjenja broja promatrača koda. Ovdje se GPS podaci dobiveni iz 22 stalnih postaja koriste za procjenu VTEC-a na dan 14-15 svibnja 2005. Rezultati istraživanja dob...