Improving TNFGN velocities via utilizing time series analysis of continuous GPS stations

Sabit GPS istasyonları zaman serileri analizi ile TUTGA hızlarının iyileştirilmesi

Abstract

Campaign type GPS measurements provide higher resolution spatial monitoring of region by supplying to the observer more station than of his receiver number. On the other hand, campaign type measurements return baseline measurements with very coarse temporal sampling. This may not be a major problem for studying only slow, steady interseismic deformations, but changes in this behavior which occur over timescales less than the separation in observations will be missed by field campaigns. It is possible to make geodetic and geodynamic calculations, carry out kinematic modeling of crustal movements via using continuous GPS measurements obtained from CGPS stations. In this study, an algorithm is investigated to receive more accurate velocity field and improve velocity obtained from campaign type GPS measurements by adding corrections to the velocities via utilizing time series analysis of continuous GPS stations. Another important usage of velocities is the densification of geodetic networks. According to the Big Scale Mapping Guide, velocities of C1 and C2 points are produced using TNFGN velocities via interpolating at related sites. Thus, obtaining accurate and reliable velocities at networks like TNFGN becomes essential. In the event that TNFGN velocities are evaluated from campaign type measurements, it is more than normal to encounter errors at densification networks due to error propagation rule. Public data of Ankara, İstanbul, Trabzon, Mersin, Gebze and Antalya stations is used as test measurements.13 ITRF stations are included to the processes for realization of datum in ITRF2000. GPS processes were done by GAMIT-GLOBK software and time series are created in ITRF2000. Also, outliers were removed at this stage. Pope test was applied to the time series and they became ready to time series analysis. Time series of related stations are prepared by completing the GPS process until the beginning of the year 2006 that constitutes the scope of this study. As with many other geophysical phenomena, noise in GPS position time series can be described as a power law process. Maximum Likelihood Estimation (MLE) was applied to investigate stochastic noise properties of GPS time series. Time series were tested by different noise models and a combination of white noise+flicker noise was found best describing model. In case only white noise model used, in other words using solely Least Squares Adjustment, uncertainties would be underestimated 7 times at north and east component and 8 times at up component. Missing values cause gaps in what otherwise would be evenly sampled data, and an even sampling may sometimes be impossible in the first place, for technical reasons. When Fourier analysis is used, and some data points are missing, they have to be provided in a more or less arbitrary way. The Lomb- Scargle algorithm has several computational advantages over more common approaches, such as Fourier analysis, including direct treatment of missing values and an estimate of a false discovery rate. Lomb-Scargle algorithm which is suitable for unequally spaced time series was applied to investigate the amplitudes and frequencies of dominant signals on a station by station basis. Seasonal signals are a potentially important part of the error budget for continuous GPS sites. Annual and semi-annual signals were determined dominant common to all sites. Thus, these periodical signals were chosen to be brought as corrections to TNFGN velocity field. An algorithm is suggested on how to apply annual and semi-annual effects in order to improve both deformation measurements and TNFGN velocities. Firstly, annual and semi-annual quantities derived from spectral analysis of all three components of continuous GPS stations, are interpolated at selected grid points. Periodic effects related to north and east component found inadequate for modeling throughout Turkey. Instead, it is proposed to use the horizontal periodic effects in an area of 10-20 km around the CGPS station. Annual and semi-annual periodics of vertical component are found convenient for interpolating at grid points. After obtaining the measurement epochs of TNFGN sites, corrections are computed at those related epochs. Finally, after adding corrections to coordinates, new velocities are obtained. New velocities are found consistent with the velocities that are derived from a complete time series analysis. An assessment is given to evaluate the results obtained from this study on how to apply both locally and in the vicinity of Turkey for horizontal and vertical velocity fields.

Özet

Kampanya tipi GPS gözlemleri kullanıcıya elindeki alıcı sayısından daha fazla sayıda noktada gözlem yapma imkânı vererek ölçü bölgesinin daha yüksek konumsal çözünürlükle izlenmesine olanak sağlamaktadır. Diğer taraftan, kampanya tipi veriler zamansal olarak seyrektirler ve küçük sıçramaları tespit edemeyecek kadar yetersizdirler. Sabit GPS noktasında toplanan sürekli GPS verileri ile yer kabuğu hareketlerine yönelik kinematik modelleme çalışmalarını gerçekleştirmek mümkün olabilmektedir. Bu çalışmada; sabit GPS istasyonları zaman serileri analizi ile hesaplanan periyodik etkilerin TUTGA hızlarına düzeltme olarak getirilerek daha gerçekçi hıza ulaşılması için bir yaklaşım önerilmiştir. Verileri herkese açık 6 sabit GPS istasyonundan elde edilen veriler GAMITGLOBK yazılımı ile değerlendirilerek zaman serileri analizine hazır hale getirilmiştir. EBOK analizi ile seçilen istasyonlara ait en uygun modelin Beyaz Gürültü ve Kırpışma Gürültüsünün birleşimi olduğu görülmüştür. Zaman serileri içindeki anlamlı sinyallerin frekanslarını ve genliklerini istasyon bazında hesaplamak amacıyla GPS zaman serilerine, boşluklu veriler için oldukça uygun bir yöntem olan Lomb-Scargle algoritması uygulanmıştır. Tüm istasyonlarda ortak olarak senelik ve altı aylık etkilerin baskın olduğu tespit edilmiştir. Böylece, TUTGA hızlarına düzeltme olarak getirilecek periyodik etkilerin hangilerinin modelleneceğine karar verilmiştir. Senelik ve altı aylık etkilerin, TUTGA hızlarına nasıl ilave edilebileceğine dair bir algoritma geliştirilerek, hız alanının iyileştirilmesi için bir yöntem önerilmiştir. Çalışmadan elde edilen sonuçlara ilişkin bir değerlendirme yapılarak, Türkiye genelinde ve bölgesel olarak nasıl uygulanabileceğine ilişkin yatay ve düşey hızlar için bir değerlendirme yapılmıştır.