Investigation of subsidence in the northeastern of Iran by estimating the velocity vector and uncertainty of permanent GPS stations

Document Type : Research Article

Authors

1 Department of Surveying Engineering. Faculty of Basic Sciences and Technical Engineering. Takestan Branch. Islamic Azad University. Takestan 34815-1515. Iran

2 University of Tehran, School of Surveying and Geospatial Engineering, College of Engineering, Tehran, Iran

3 Associate Professor. Department of Geodesy and Surveying Engineering. Tafresh University. Tafresh 39518-79611. Iran

Abstract

This study presents a new estimate for subsidence in the northeast of the country through the time series analysis of 11-year (from beginning of 2006 till the end of 2016) of 31 stations of Khorasan network, as part of Iranian Permanent Geodynamic & GNSS Network (IPGN), located in northeastern Iran. The mentioned estimation is obtained from the velocity vector of network stations in the International Terrestrial Reference Frame of ITRF2014 based on time series analysis in two realms, i.e., deterministic model analysis and stochastic model analysis. The deterministic model analysis is comprised of jump detection, determination of station motion model parameters, the study of station trend, outlier detection, and statistical significance test to check the jumps magnitude. Due to the interdependence of these steps, the related calculations are performed iteratively. Noise analysis includes two phases, namely, spatial filtering and temporal filtering. In the first phase, the Common Mode Error (CME) parameter is calculated using the weighted stacking method and taking into account the data correlation coefficient and stations distance. In the second phase, using the maximum likelihood estimation (MLE) method, the optimal noise model is derived as a combination of white noise and flicker noise. As a result, the reliable velocities of the stations (resulting from a complete analysis of the deterministic model) and their realistic uncertainties (resulting from the selection of optimal stochastic models) are calculated. Based on this study we found that: (1) Each station during the 11-year study period has on average nine jumps, all of which of non-seismic origin. (2) Including the data from all IPGN stations in spatial filtering, leads better results and on average reduces the norm of post-fit residual vectors for east, north and up coordinate components by 30.17%, 29.40%, and 17.90%, respectively. (3) Concerning the temporal filtering, we found that the noise of the up component is significantly higher than the noise of the horizontal components. (4) Stochastic model analysis showed the realistic uncertainties of the east, north, and up components are 4.33, 4.44, and 3.70 times, respectively, greater than the uncertainties which are derived without application of stochastic modeling (optimistic uncertainties). (5) The vertical velocity of most of stations found to be in the normal range of -5 to 5 mm/yr. (6) Five stations, namely, GOLM, GRGN, NFRD, NISH and SHRN are having anomalous subsidence (up to 9 mm/yr). (7) The proximity of the three stations GOLM, NFRD and NISH allow us to infer a regional subsidence for the area of their location. (8) The station GRGN, in addition to anomalous subsidence, shows distinctive features such as the nonlinear trend as well as large periodic signals in the up component of the station. Therefore, to find the reason for such vertical behavior of the earth's crust more permanent GNSS stations must be established in that area. (9) The estimated parameters of periodic signal of the stations demonstrate that the annual and draconitic year signals have the largest amplitudes in the three coordinate components. In addition, amplitude of the periodic signals of the up-component is significantly larger than the other components.

Keywords

Main Subjects