G23B-1074
Noise characteristics of Continuous GPS time series of Central and Eastern Himalaya 

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Jagat Dwipendra Ray1, M.Sithartha Muthu Vijayan2 and Ashok Kumar1, (1)Tezpur University, Tezpur, India, (2)Center for Mathematical Modelling and Computer Simulation, Bangalore, Karnataka, India
Abstract:
Global positioning system measurements with its millimetre level accuracy have been widely used to monitor the crustal dynamics. Geodetic crustal deformation studies require accurate estimate of the parameters which demands realistic estimate of the uncertainties in order to constrain the signal. GPS based crustal deformation studies in tectonically active region, such as Central and Eastern Himalaya have been carried out by several groups however, proper noise characteristics of GPS time series of this study region are unknown. In this work, we attempt to address the noise characteristics of GPS position time series by analysing the GPS time series of 22 stations from North-East India, Bhutan and Nepal Himalaya spanning 2002-2013. We have employed Spectral analysis and Maximum Likelihood Estimation (MLE) to study the noise characteristics. Power spectrum obtained by using Lomb-Scargle method reveals characteristics of white noise at the high frequencies and power law noise at lower frequencies. Estimation of the spectral index by finding the slope of the spectral curve suggests fractal white noise with overall index of -0.61. MLE was performed in two ways. First, by assuming the time series to be composed of (a) white (WN), (b) white plus flicker (FL) and (c) white plus random walk noise (WRN) and then by estimating spectral index assuming the noise to be composition of white and power law noise (WPN). The comparison of MLE values of three noise model suggest that white plus flicker noise model (FL) is the most preferred noise model. Comparison of velocity uncertainties between white noise and white plus flicker noise, obtained from MLE, suggest that velocity uncertainty is under estimated by factor of ~8 when simple white noise model is used. The spectral index estimated using MLE is -1.1 (~1) which suggests that flicker noise is the main power law noise in time series of all 22 GPS stations. A slight difference of noise amplitudes of two different monument types (Concrete Pillar and Shallow drilled braced monuments) is also observed which rules out the preference of one over the another.