We perform a statistical sensitivity analysis on a parametric fit to vertical daily displacement time series of 244 European Permanent GNSS stations, with a focus on linear vertical land motion (VLM), i.e., station velocity. We compare two independent corrections to the raw (uncorrected) observed displacements. The first correction is physical and accounts for non-tidal atmospheric, non-tidal oceanic and hydrological loading displacements, while the second approach is an empirical correction for the common-mode errors. For the uncorrected case, we show that combining power-law and white noise stochastic models with autoregressive models yields adequate noise approximations. With this as a realistic baseline, we report improvement rates of about 14% to 24% in station velocity sensitivity, after corrections are applied. We analyze the choice of the stochastic models in detail and outline potential discrepancies between the GNSS-observed displacements and those predicted by the loading models. Furthermore, we apply restricted maximum likelihood estimation (RMLE), to remove low-frequency noise biases, which yields more reliable velocity uncertainty estimates. RMLE reveals that for a number of stations noise is best modeled by a combination of random walk, flicker noise, and white noise. The sensitivity analysis yields minimum detectable VLM parameters (linear velocities, seasonal periodic motions, and offsets), which are of interest for geophysical applications of GNSS, such as tectonic or hydrological studies.
3 April 2025 in Natural Hazards and Earth System Sciences
Snow avalanches threaten people and infrastructure in mountainous areas. For the assessment of temporal protection measures of infrastructure when the avalanche danger is high, local and up-to-date information from the release zones and the avalanche track is crucial. One main factor influencing the avalanche situation is wind-drifted snow, which causes variations in snow depth across a slope. We developed a monitoring system using low-cost lidar and optical sensors to measure snow depth variations in an avalanche release area at a high spatiotemporal scale (centimetre to low decimetre spatial resolution and hourly temporal resolution). We analyse data obtained from such a monitoring system, installed within an avalanche release area at 2200 m a.s.l. close to Davos in the Swiss Alps. The system comprises two measurement stations and has been operational since November 2023. We present the experiences and insights gained from a preliminary analysis of the data obtained so far. The temporal variations of the spatial coverage show the potential and limitations of the system under varying weather conditions. A comparison of the surface elevation models derived from the lidar data and from photogrammetric processing of UAV-based images shows a good agreement, with a mean vertical difference of 0.005 m and standard deviation of 0.15 m. An avalanche event and a period of snowfall with strong winds, chosen as case studies, show the potential and limitations of the proposed system to detect changes in the snow depth distribution on a low decimetre level or better. The results obtained so far indicate that a measurement system with a few setups in or near an avalanche slope can provide information about the small-scale snow depth distribution changes in near real time. We expect that such systems and the related data processing have the potential to support experts in their decisions on avalanche safety measures in the future.
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.
Terrestrial laser scanners (TLSs) offer a possibility for more automated and efficient deformation monitoring of civil engineering structures with higher spatial resolution than standard methods, as well as without the necessity of permanently installing the monitoring equipment. In such applications, scanners are usually placed so that the lines of sight are roughly aligned with the main directions of the expected deformations, and the deformations are estimated from point cloud differences between multiple epochs. This allows high sensitivity in the direction of the surface normal, but deformations along the surface are often undetected or hard to precisely quantify. In this work, we propose an algorithm based on the detection and matching of keypoints identified within TLS intensity images. This enables precise quantification of deformations along the scanned surfaces. We also present the application of the algorithm for monitoring a bridge pier of the Hochmoselbrücke in Germany, as a case study. Deformations up to about 4 cm due to thermal expansion and bending of the pier were successfully detected from scans taken throughout the day from a single location, up to 180 m from the monitored surfaces. The results agreed within a few millimeters to independent monitoring using state-of-the-art processing of TLS point clouds obtained from a different location and using a different type/brand of instrument. The newly proposed algorithm can either be used to complement existing TLS-based deformation analysis methods by adding sensitivity in certain directions, or it can be valuable as a standalone solution.
Global Navigation Satellite System (GNSS) long-term residual height time series exhibit signals related to environmental influences. These can partly b explained through environmental surface loads, which are described with physical models. In this work, a model is computed to connect the GNSS residuals with raw meteorological parameters. A Temporal Convolutional Network (TCN) is trained on 206 GNSS stations in central Europe, and applied to 68 test stations in the same area. The resulting Root Mean Square (RMS) error reduction is on average 0.8% lower for the TCN modeled time series, compared to using physical models for the reduction. In a further experiment, the TCN is trained on the GNSS time series after reducing those by the surface loading models. The aim is a further increase of RMS reduction, which is achieved with 2.7% on average, resulting in an overall mean reduction of 28.6%. The results suggest that with meteorological features as input data, TCN modeled reductions are able to compete with reductions derived from physical models. Trained on the residuals reduced by environmental loading models, the TCN is able to slightly increase the overall reduction of variations in the GNSS station position time series.
Snow avalanches threaten people and infrastructure in mountainous areas. For the assessment of temporal protection measures of infrastructure in dangerous situations, local and up to date information is crucial. One factor influencing the avalanche situation is wind drifted snow, which causes variations in snow depth across a slope, but this data is rarely available. We present a monitoring system using low cost LiDAR and optical sensors, which we use to monitor snow depth variations in an avalanche release area. The system is operational since November 2023, autonomously measuring and providing data from our study area close to Davos in Switzerland, with high spatiotemporal resolution. In the first three operating months we gained experiences and made a preliminary assessment of the system performance. An analysis of the changes in spatial coverage shows the limitations and potentials of the system under different weather conditions. A comparison of the surface models derived from the LiDAR data and a photogrammetric drone shows a good agreement, achieving a mean of 0.005 m and standard deviation of 0.15 m. Two case studies, including an avalanche event and a period of snowfall with strong winds, show the potential of the proposed system to detect changes in the snow depth distribution on a low decimeter level, or better. In addition, we record meteorological parameters which we will use in future, together with the newly established snow depth database, for attempts to refine and further develop models for wind-induced snow redistribution. The near real time information of the snow depth distribution in avalanche prone slopes will be provided to experts, so it can aid their decisions on avalanche safety measures.