Modeling of Residual GNSS Station Motions through Meteorological Data in a Machine Learning Approach

Long-term Global Navigation Satellite System (GNSS) height residual time series contain signals that are related to environmental influences. A big part of the residuals can be explained by environmental surface loadings, expressed through physical models. This work aims to find a model that connects raw meteorological parameters with the GNSS residuals. The approach is to train a Temporal Convolutional Network (TCN) on 206 GNSS stations in central Europe, after which the resulting model is applied to 68 test stations in the same area. When comparing the Root Mean Square (RMS) error reduction of the time series reduced by physical models, and, by the TCN model, the latter reduction rate is, on average, 0.8% lower. In a second experiment, the TCN is utilized to further reduce the RMS of the time series, of which the loading models were already subtracted. This yields additional 2.7% of RMS reduction on average, resulting in a mean RMS reduction of 28.6% overall. The results suggests that a TCN, using meteorological features as input data, is able to reconstruct the reductions almost on the same level as physical models. Trained on the residuals, reduced by environmental loadings, the TCN is still able to slightly increase the overall reduction of variations in the GNSS station position time series.

General Geometric Model of GNSS Position Time Series for Crustal Deformation Studies – A Case Study of CORS Stations in Vietnam

In processing of position time series of crustal deformation monitoring stations by continuousGNSS station, it is very important to determine the motion model to accurately determine the displacementvelocity and other movements in the time series. This paper proposes (1) the general geometric model foranalyzing GNSS position time series, including common phenomena such as linear trend, seasonal term,jumps, and post-seismic deformation; and (2) the approach for directly estimating time decay ofpostseismic deformations from GNSS position time series, which normally is determined based on seismicmodels or the physical process seismicity, etc. This model and approach are tested by synthetic positiontime series, of which the calculation results show that the estimated parameters are equal to the givenparameters. In addition they were also used to process the real data which is GNSS position time series of4 CORS stations in Vietnam, then the estimated velocity of these stations: DANA (n, e, u = -9.5, 31.5, 1.5mm/year), HCMC (n, e, u = -9.5, 26.2, 1.9 mm/year), NADI (n, e, u = -10.6, 31.5, -13.4 mm/year), andNAVI (n, e, u = -13.9, 32.8, -1.1 mm/year) is similar to previous studies.

Spatial Variations of Stochastic Noise Properties in GPS Time Series

The noise in position time series of 568 GPS (Global Position System) stations across North America with an observation span of ten years has been investigated using solutions from two processing centers, namely, the Pacific Northwest Geodetic Array (PANGA) and New Mexico Tech (NMT). It is well known that in the frequency domain, the noise exhibits a power-law behavior with a spectral index of around −1. By fitting various noise models to the observations and selecting the most likely one, we demonstrate that the spectral index in some regions flattens to zero at long periods while in other regions it is closer to −2. This has a significant impact on the estimated linear rate since flattening of the power spectral density roughly halves the uncertainty of the estimated tectonic rate while random walk doubles it. Our noise model selection is based on the highest log-likelihood value, and the Akaike and Bayesian Information Criteria to reduce the probability of over selecting noise models with many parameters. Finally, the noise in position time series also depends on the stability of the monument on which the GPS antenna is installed. We corroborate previous results that deep-drilled brace monuments produce smaller uncertainties than concrete piers. However, if at each site the optimal noise model is used, the differences become smaller due to the fact that many concrete piers are located in tectonic/seismic quiet areas. Thus, for the predicted performance of a new GPS network, not only the type of monument but also the noise properties of the region need to be taken into account.

Implication between Geophysical Events and the Variation of Seasonal Signal Determined in GNSS Position Time Series

The seasonal signal determined by the Global Navigation Satellite System (GNSS), which is captured in the coordinate time series, exhibits annual and semi-annual periods. This signal is frequently modelled by two periodic signals with constant amplitude and phase-lag. The purpose of this study is to explore the implication of different types of geophysical events on the seasonal signal in three stages—in the time span that contains the geophysical events, before and after the geophysical event, but also the stationarity phenomena, which is analysed on approximately 200 reference stations from the EPN network since 1995. The novelty of the article is demonstrated by correlating three different types of geophysical events, such as earthquakes with a magnitude greater than 6° on the Richter scale, landslides, and volcanic activity, and analysing the variation in amplitude of the seasonal signal. The geophysical events situated within a radius of 30 km from the epicentre showed a higher seasonal value than when the timespan did not contain a geophysical event. The presence of flicker and random walk noise was computed using overlapping Hadamard variance (OHVAR) and the non-stationary behaviour of the time series of the CORS coordinates in the time frequency analysis was done using continuous wavelet transform (CWT).

Analysis of Noise and Velocity in GNSS EPN-Repro 2 Time Series

This study evaluates the EUREF Permanent Network (EPN) station position time series of approximately 200 GNSS stations subject to the Repro 2 reprocessing campaign in order to characterize the dominant types of noise and amplitude and their impact on estimated velocity values and associated uncertainties. The visual inspection on how different noise model represents the analysed data was done using the power spectral density of the residuals and the estimated noise model and it is coherent with the calculated Allan deviation (ADEV)-white and flicker noise. The velocities resulted from the dominant noise model are compared to the velocity obtained by using the Median Interannual Difference Adjusted for Skewness (MIDAS). The results show that only 3 stations present a dominant random walk noise model compared to flicker and powerlaw noise model for the horizontal and vertical components. We concluded that the velocities for the horizontal and vertical component show similar values in the case of MIDAS and maximum likelihood estimation (MLE), but we also found that the associated uncertainties from MIDAS are higher compared to the uncertainties from MLE. Additionally, we concluded that there is a spatial correlation in noise amplitude, and also regarding the differences in velocity uncertainties for the Up component.

Comparison and Assessment of Three ITRS Realizations

A terrestrial reference frame (TRF) is derived based on historical geodetic data and is normally updated every 5–6 years. The three most recent International Terrestrial Reference System (ITRS) realizations, ITRF2014, DTRF2014, and JTRF2014, were determined with different strategies, which has resulted in different signals in the reference frame parameters. In this paper, we used the continuous site position time series of International GNSS Service (IGS) from 1995 to 2020 as a benchmark to investigate the characteristics of the three frames. In the comparison, the ITRS realizations were divided into the determination and prediction sections, where the site coordinates of the TRFs were extrapolated in the prediction period. The results indicated that the orientation and scale parameters of the ITRF2014, and the IGS solutions showed excellent agreement during the determination period of ITRF2014, while, during the prediction period, the orientation parameter diverged from IGS with rates of 11.9, 5.5, and 8.4 as/yr, and the scale degraded with a rate of −0.038 ppb/yr. The consistency of the origin parameters between the DTRF2014 and the IGS solutions during the two periods changed from 0.07, 0.11, and −0.15 mm/yr to −0.17, −0.18, and −0.12 mm/yr; the consistency of orientation parameters from −3.6, −1.9, and 2.9 as/yr to 15.9, −2.3, and 13.2 as/yr; and the consistency of scale from 0.007 to −0.005 ppb/yr. In the comparison between the JTRF2014 and IGS solutions, annual signals in the origin differences were 1.5, 3.0, and 2.4 mm in the X, Y, and Z components, respectively, and the temporal variation trends in different periods disagreed with their long-term trends. Obvious trend switches in the rotation parameters were also observable, and the complex temporal variation characteristics of the scale offsets may be related to the scale definition strategy applied in different TRFs.

Remote Detection of Surge-Related Glacier Terminus Change across High Mountain Asia

High Mountain Asia (HMA) hosts the largest glacier concentration outside of polar regions. It is also distinct glaciologically as it forms one of two major surge clusters globally, and many glaciers there contradict the globally observed glacier recession trend. Surging glaciers are critical to our understanding of HMA glacier dynamics, threshold behaviour and flow instability, and hence have been the subject of extensive research, yet many dynamical uncertainties remain. Using the cloud-based geospatial data platform, Google Earth Engine (GEE) and GEE-developed tool, GEEDiT, to identify and quantify trends in the distribution and phenomenological characteristics of surging glaciers synoptically across HMA, we identified 137 glaciers as surging between 1987–2019. Of these, 55 were newly identified, 15 glaciers underwent repeat surges, and 18 were identified with enhanced glaciological hazard potential, most notably from Glacier Lake Outburst Floods (GLOFs). Terminus position time series analysis from 1987–2019 facilitated the development of a six-part phenomenological classification of glacier behaviour, as well as quantification of surge variables including active phase duration, terminus advance distance and rate, and surge periodicity. This research demonstrates the application of remote sensing techniques and the GEE platform to develop our understanding of surging glacier distribution and terminus phenomenology across large areas, as well as their ability to highlight potential geohazard locations, which can subsequently be used to focus monitoring efforts.