Seismic buffer recognition using wavelet-based features and neural classification

One of the first operations in a seismic signal processing system applied to earthquake data is to distinguish between valid and invalid records. Since valid signals are characterized by a combination of their time and frequency properties, wavelets are natural candidates for describing seismic features in a compact way. This paper develops a seismic buffer pattern recognition technique, comprising wavelet-based feature extraction, feature selection based on the mutual information criterion, and neural classification based on feedforward networks. The ability of the wavelet transform to capture discriminating information from seismic data in a small number of features is compared with alternative feature reduction techniques, including statistical moments. Three different variations of the wavelet transform are used to extract features: the discrete wavelet transform, the single wavelet transform and the continuous wavelet transform. The mutual information criterion is employed to select a relatively small set of wavelets from the time–frequency grid. Firstly, it is determined whether wavelets can capture more informative data in an equal number of features compared with other features derived from raw data. Secondly, wavelet-based features are compared with features selected based on prior knowledge of class differences. Thirdly, a technique is developed to optimize wavelet features as part of the neural network training process, by using the wavelet neural network architecture. The automated classification techniques developed in this paper are shown to perform similarly to human operators trained for this function. Wavelet-based techniques are found to be useful, both for preprocessing of the raw data and for extracting features from the data. It is demonstrated that the definition of wavelet features can be optimized using the classification wavelet network architecture.     

Strategies to mitigate aliasing of loading signals while estimating GPS frame parameters

Although GNSS techniques are theoretically sensitive to the Earth center of mass, it is often preferable to remove intrinsic origin and scale information from the estimated station positions since they are known to be affected by systematic errors. This is usually done by estimating the parameters of a linearized similarity transformation which relates the quasi-instantaneous frames to a long-term frame such as the International Terrestrial Reference Frame (ITRF). It is well known that non-linear station motions can partially alias into these parameters. We discuss in this paper some procedures that may allow reducing these aliasing effects in the case of the GPS techniques. The options include the use of well-distributed sub-networks for the frame transformation estimation, the use of site loading corrections, a modification of the stochastic model by downweighting heights, or the joint estimation of the low degrees of the deformation field. We confirm that the standard approach consisting of estimating the transformation over the whole network is particularly harmful for the loading signals if the network is not well distributed. Downweighting the height component, using a uniform sub-network, or estimating the deformation field perform similarly in drastically reducing the amplitude of the aliasing effect. The application of these methods to reprocessed GPS terrestrial frames permits an assessment of the level of agreement between GPS and our loading model, which is found to be about 1.5 mm WRMS in height and 0.8 mm WRMS in the horizontal at the annual frequency. Aliased loading signals are not the main source of discrepancies between loading displacement models and GPS position time series.     

Quality assessment of GPS reprocessed terrestrial reference frame

The International GNSS Service (IGS) contributes to the construction of the International Terrestrial Reference Frame (ITRF) by submitting time series of station positions and Earth Rotation Parameters (ERP). For the first time, its submission to the ITRF2008 construction is based on a combination of entirely reprocessed GPS solutions delivered by 11 Analysis Centers (ACs). We analyze the IGS submission and four of the individual AC contributions in terms of the GNSS frame origin and scale, station position repeatability and time series seasonal variations. We show here that the GPS Terrestrial Reference Frame (TRF) origin is consistent with Satellite laser Ranging (SLR) at the centimeter level with a drift lower than 1 mm/year. Although the scale drift compared to Very Long baseline Interferometry (VLBI) and SLR mean scale is smaller than 0.4 mm/year, we think that it would be premature to use that information in the ITRF scale definition due to its strong dependence on the GPS satellite and ground antenna phase center variations. The new position time series also show a better repeatability compared to past IGS combined products and their annual variations are shown to be more consistent with loading models. The comparison of GPS station positions and velocities to those of VLBI via local ties in co-located sites demonstrates that the IGS reprocessed solution submitted to the ITRF2008 is more reliable and precise than any of the past submissions. However, we show that some of the remaining inconsistencies between GPS and VLBI positioning may be caused by uncalibrated GNSS radomes.     

Annual variation detected by GPS,GRACE and loading models

Most GPS coordinate time series, surface displacements derived from the Gravity Recovery and Climate Experiment (GRACE), and loading models display significant annual signals at many regions. This paper compares the annual signals of the GPS position time series from the Crustal Dynamics Data Information System (CDDIS), estimates of loading from GRACE monthly gravity field models calculated by three processing centers (Center of Spatial Research, CSR; Jet Propulsion Laboratory, JPL; GeoForschungsZentrum, GFZ) and three geophysical fluids models (National Center for Environmental Prediction, NCEP; Estimating the Circulation and Climate of the Ocean, ECCO; Global Land Data Assimilation System, GLDAS) for 270 globally distributed stations for the period 2003-2011. The results show that annual variations derived from the level-2 products from the three GRACE product centers are very similar. The absolute difference in annual amplitude between any two centers is never larger than 1.25 mm in the vertical and 0.11 mm in horizontal displacement. The mean phase differences of the GRACE results are less than ten days for all three components. When we correct the GPS vertical coordinate time series using the GRACE annual amplitudes using the products from three GRACE analysis centers, we find that we are able to reduce the GPS annual signal in the vertical at about 80% stations and the average reduction is about 47%. In the north and the east, the annual amplitude is reduced on 77% and 72% of the stations with the average reduction 32% and 33%. We also compare the annual surface displacement signal derived from two environmental models; the two models use the same atmospheric and non-tidal ocean loading and differ only in the continental water storage model that we use, either NCEP or GLDAS. We find that the model containing the GLDAS continental water storage is able to better reduce the annual signal in the GPS coordinate time series.     

An estimate of the influence of loading effects on tectonic velocities in the Pyrenees

Surface displacements due to temporal changes in environmental mass redistributions are observable in the coordinate time series of many Global Navigation Satellite System (GNSS) sites. In this study, we investigated the effect of loading on estimates of tectonic velocity computed from campaign-style GNSS observations. The study region is in the Pyrenees mountain range between France and Spain (ResPyr campaigns). In this area, seismic activity is continuous and moderate and the expected amplitude of the horizontal tectonic velocity is less than 0.5 mm/yr. In order to determine the velocity, 4 sparse GNSS campaigns were carried out from 1995 to 2010. Considering this small rate of deformation, loading phenomena can contribute a non-negligible artifact to the velocity computation that could affect our geodynamical interpretation. In this investigation, we specifically considered the atmospheric, hydrological, and non-tidal ocean loading phenomena. The computed loading deformations for this region show the horizontal displacements are dominated by the non-tidal ocean loading (maximum 4 mm for the North and 3.1 mm for the East components); the main vertical contributions are due to the atmospheric and continental water storage loading (maximum 14.3 for the atmosphere and 8.1 mm for the hydrology, respectively). We have found that the dominant loading effect on the horizontal velocity is the non-tidal ocean loading (mean of 0.11 mm/yr), whereas the vertical component is dominated by the hydrological loading (mean of 0.21 mm/yr). Since the study area is in a mountainous region, we also analyzed the difference between the atmospheric and the topography dependent atmospheric loading models at our GNSS campaign sites. We did not find any significant difference between the two atmospheric loading models in terms of horizontal velocity. Finally, we performed simulations to identify the optimum timing and frequency of future GNSS campaigns in this area that would minimize the loading effects on tectonic velocity estimates.     

Vertical deformations from homogeneously processed GRACE and global GPS long-term series

Temporal variations in the geographic distribution of surface mass cause surface displacements. Surface displacements derived from GRACE gravity field coefficient time series also should be observed in GPS coordinate time series, if both time series are sufficiently free of systematic errors. A successful validation can be an important contribution to climate change research, as the biggest contributors to mass variability in the system Earth include the movement of oceanic, atmospheric, and continental water and ice. In our analysis, we find that if the signals are larger than their precision, both geodetic sensor systems see common signals for almost all the 115 stations surveyed. Almost 80% of the stations have their signal WRMS decreased, when we subtract monthly GRACE surface displacements from those observed by GPS data. Almost all other stations are on ocean islands or small peninsulas, where the physically expected loading signals are very small. For a fair comparison, the data (79 months from September 2002 to April 2009) had to be treated appropriately: the GPS data were completely reprocessed with state-of-the-art models. We used an objective cluster analysis to identify and eliminate stations, where local effects or technical artifacts dominated the signals. In addition, it was necessary for both sets of results to be expressed in equivalent reference frames, meaning that net translations between the GPS and GRACE data sets had to be treated adequately. These data sets are then compared and statistically analyzed: we determine the stability (precision) of GRACE-derived, monthly vertical deformation data to be ~1.2 mm, using the data from three GRACE processing centers. We statistically analyze the mean annual signals, computed from the GPS and GRACE series. There is a detailed discussion of the results for five overall representative stations, in order to help the reader to link the displayed criteria of similarity to real data. A series of tests were performed with the goal of explaining the remaining GPS–GRACE residuals.     

Comparative analysis of different environmental loading methods and their impacts on the GPS height time series

Three different environmental loading methods are used to estimate surface displacements and correct non-linear variations in a set of GPS weekly height time series. Loading data are provided by (1) Global Geophysical Fluid Center (GGFC), (2) Loading Model of Quasi-Observation Combination Analysis software (QLM) and (3) our own daily loading time series (we call it OMD for optimum model data). We find that OMD has the smallest scatter in height across the selected 233 globally distributed GPS reference stations, GGFC has the next smallest variability, and QLM has the largest scatter. By removing the load-induced height changes from the GPS height time series, we are able to reduce the scatter on 74, 64 and 41 % of the stations using the OMD models, the GGFC model and QLM model respectively. We demonstrate that the discrepancy between the center of earth (CE) and the center of figure (CF) reference frames can be ignored. The most important differences between the predicted models are caused by (1) differences in the hydrology data from the National Center for Atmospheric Research (NCEP) vs. those from the Global Land Data Assimilation System (GLDAS), (2) grid interpolation, and (3) whether the topographic effect is removed or not. Both QLM and GGFC are extremely convenient tools for non-specialists to use to calculate loading effects. Due to the limitation of NCEP reanalysis hydrology data compared with the GLDAS model, the GGFC dataset is much more suitable than QLM for applying environmental loading corrections to GPS height time series. However, loading results for Greenland from GGFC should be discarded since hydrology data from GLDAS in this region are not accurate. The QLM model is equivalent to OMD in Greenland and, hence, could be used as a complement to the GGFC product to model the load in this region. We find that the predicted loading from all three models cannot reduce the scatter of the height coordinate for some stations in Europe.     

Seasonal low-degree changes in terrestrial water mass load from global GNSS measurements

Large-scale mass redistribution in the terrestrial water storage (TWS) leads to changes in the low-degree spherical harmonic coefficients of the Earth’s surface mass density field. Studying these low-degree fluctuations is an important task that contributes to our understanding of continental hydrology. In this study, we use global GNSS measurements of vertical and horizontal crustal displacements that we correct for atmospheric and oceanic effects, and use a set of modified basis functions similar to Clarke et al. (Geophys J Int 171:1–10, 2007) to perform an inversion of the corrected measurements in order to recover changes in the coefficients of degree-0 (hydrological mass change), degree-1 (centre of mass shift) and degree-2 (flattening of the Earth) caused by variations in the TWS over the period January 2003–January 2015. We infer from the GNSS-derived degree-0 estimate an annual variation in total continental water mass with an amplitude of \((3.49 \pm 0.19) \times 10^{3}\) Gt and a phase of \(70^{\circ } \pm 3^{\circ }\) (implying a peak in early March), in excellent agreement with corresponding values derived from the Global Land Data Assimilation System (GLDAS) water storage model that amount to \((3.39 \pm 0.10) \times 10^{3}\) Gt and \(71^{\circ } \pm 2^{\circ }\), respectively. The degree-1 coefficients we recover from GNSS predict annual geocentre motion (i.e. the offset change between the centre of common mass and the centre of figure) caused by changes in TWS with amplitudes of \(0.69 \pm 0.07\) mm for GX, \(1.31 \pm 0.08\) mm for GY and \(2.60 \pm 0.13\) mm for GZ. These values agree with GLDAS and estimates obtained from the combination of GRACE and the output of an ocean model using the approach of Swenson et al. (J Geophys Res 113(B8), 2008) at the level of about 0.5, 0.3 and 0.9 mm for GX, GY and GZ, respectively. Corresponding degree-1 coefficients from SLR, however, generally show higher variability and predict larger amplitudes for GX and GZ. The results we obtain for the degree-2 coefficients from GNSS are slightly mixed, and the level of agreement with the other sources heavily depends on the individual coefficient being investigated. The best agreement is observed for \(T_{20}^C\) and \(T_{22}^S\), which contain the most prominent annual signals among the degree-2 coefficients, with amplitudes amounting to \((5.47 \pm 0.44) \times 10^{-3}\) and \((4.52 \pm 0.31) \times 10^{-3}\) m of equivalent water height (EWH), respectively, as inferred from GNSS. Corresponding agreement with values from SLR and GRACE is at the level of or better than \(0.4 \times 10^{-3}\) and \(0.9 \times 10^{-3}\) m of EWH for \(T_{20}^C\) and \(T_{22}^S\), respectively, while for both coefficients, GLDAS predicts smaller amplitudes. Somewhat lower agreement is obtained for the order-1 coefficients, \(T_{21}^C\) and \(T_{21}^S\), while our GNSS inversion seems unable to reliably recover \(T_{22}^C\). For all the coefficients we consider, the GNSS-derived estimates from the modified inversion approach are more consistent with the solutions from the other sources than corresponding estimates obtained from an unconstrained standard inversion.     

Seasonal Motion in the Annapolis, Maryland GPS Monument

The permanent GPS tracking site at Annapolis, MD shows a 7-mm seasonal signal primarily in its horizontal position. It is suggested that thermal expansion of the pier on which the antenna rests is the source of this motion. A simple numerical model of the pier reproduces the observed motion of the GPS antenna, lending credence to this hypothesis. Although adding an additional level of complexity, this motion is predictable and the site retains it s value for high precision monitoring. Although the arrangement of this GPS site it somewhat uncommon, these results emphasize the importance of the underlying antenna monumentation when measuring crustal motions. ? 1999 John Wiley & Sons, Inc.