Abstract Tropospheric delay modeling accuracy is a critical factor affecting the performance of Precise Point Positioning Real-Time Kinematic (PPP-RTK). However, reliable tropospheric delay estimation remains challenging in regions characterized by sparse reference stations and complex terrain. To address this issue, this study proposes a novel tropospheric delay modeling method that integrates tropospheric delays derived from the Pangu-Weather Model with Global Navigation Satellite System (GNSS) observations to enhance PPP-RTK positioning performance. The proposed method first estimates tropospheric delays from both techniques at different locations and subsequently models the inter-technique bias. The method is validated using a network of 274 GNSS stations across Europe. Results from a one-month experiment demonstrate that the Pangu-Weather Model achieves zenith wet delay (ZWD) estimation accuracy at the level of approximately 2~3cm. In regions with sparse reference stations and complex terrain, the proposed method improves ZWD accuracy, with root mean square (RMS) reductions of 36.1%, 54.7%, and 68.3% compared with three conventional methods. In PPP-RTK positioning, the proposed method enhances convergence speed, accuracy, and stability. The average convergence time is reduced by 9.9%, 12.7%, and 11.3%, while the three-dimensional positioning accuracy is improved by up to 12.6%, 27.8%, and 22.8%. The proportion of vertical positioning errors within 5cm increases from 95.6% to 99.4%. Furthermore, large-scale experiments demonstrate consistent improvements across multiple error intervals, with the proportion of errors within 1cm increased by up to 3.2%, 3.8%, and 16.5% in the east, north, and up components, respectively. These results demonstrate the effectiveness, robustness, and applicability of the proposed method, particularly in challenging environments with sparse reference stations and complex terrain.

Abstract As global climate change intensifies, hurricane‐induced storm surges are becoming more frequent and severe. While Global Navigation Satellite System‐Interferometric Reflectometry (GNSS‐IR) is widely used to monitor sea level variations, its capability to detect rapid and extreme events remains limited. We propose a short‐time feature extraction GNSS‐IR strategy constrained by astronomical tidal models. By analyzing the continuity and stability of spectral reflections, the method identifies coherent signals from transient sea level changes and effectively addresses the typical 10–20 min temporal bias introduced by the static‐surface assumption. Validation results show that the method achieves a long‐term monitoring accuracy of 4.6 cm over 1 year, and maintains a stable accuracy of approximately 10 cm during storm surges. It also achieves 4.0 cm accuracy over 12‐hr period and enables short‐term sea level prediction with an accuracy of 7 cm. These findings highlight the potential of near‐shore GNSS‐IR to strengthen tide gauge networks and marine assessments.

Storm surges occurring over short periods can cause abrupt changes in ocean mass near the coast, leading to transient crustal subsidence known as Non-Tidal Ocean Loading (NTOL). Existing NTOL products provided by various institutions are typically derived from oceanic gravity or fluid dynamic models combined with Earth’s elastic response theory. However, these model-based products lack validation against in-situ observations, raising concerns about their reliability under extreme conditions. In contrast, geodetic Global Navigation Satellite Systems (GNSS) observations offer high accuracy, high spatiotemporal resolution, and near real-time availability. In this study, we analyze sub-daily GNSS vertical displacements from 12 stations in Hong Kong during a storm surge in October 2021 and compare them with NTOL predictions from three global models (GFZ, IMLS, and GGFC) to assess their consistency in a complex coastal environment. Results show that the IMLS and GGFC NTOL products are more reliable than the GFZ product for detecting sub-daily loading deformations in a small region. The GFZ product suffers from spurious NTOL displacement signals (max ≈ 15 mm) during the peak storm surge period. Among the correlation coefficients between 3-h GNSS vertical displacements and the three NTOL products, the IMLS product achieves the highest correlation (max = 0.67), followed by the GGFC product, while the GFZ product exhibits the lowest correlation. Extending the GNSS solution time span (from 3-h to 6-h or 9-h) improves its correlation with NTOL. However, since storm surge effects typically last only tens of minutes to a few hours, a 6-h or 9-h GNSS solution may not capture the full details of the storm surge-induced deformation. Our work demonstrates that sub-daily GNSS observations can detect the transient NTOL-induced subsidence caused by storm surges, thereby validating the accuracy and applicability of these NTOL models and filling the gap left by the lack of observational support in previous model-based studies.

Accurate monitoring data is fundamental for structural deformation analysis, particularly those data with significant features induced by specific loads. Feature-rich data are often sparsely distributed. However, conventional AI-based data augmentation methods, which rely on distance-based weighting, consequently fail to achieve high-precision augmentation with feature data. This study proposes a modified Transformer-based data augmentation method, that employs feature-based weighting, marking the first introduction of the Transformer architecture into data augmentation for structural deformation monitoring. The proposed method incorporates an analysis of the physical mechanisms underlying engineering structural deformation and modifies Transformer architecture to enhance feature extraction capabilities through dimensional increase. The method processes feature data without distance constraints, enabling high-precision interpolation of missing values. Experimental results on field decimeter-level deformation data demonstrated a mean absolute error (MAE) of 2.63 mm and a root mean square error (RMSE) of 3.37 mm. Compared to non-dimensionality-increased Transformers, and other deep-learning approaches, the proposed method achieved the highest accuracy. This method contributes to enhancing the completeness of monitoring data with critical features in engineering structures, supporting more accurate structural analysis in engineering applications.

Constraining the physical properties of Earth's deep interior, particularly the viscosity of the solid inner core and the density structure of large low-velocity provinces (LLVPs), remains a major challenge in geophysics. Here we develop a unified dynamical framework that combines mantle-inner core gravitational coupling (MICG) with torsional oscillations in the fluid outer core and show that their interaction can produce a distinct and testable geodetic signature. Guided by this prediction, we analyze satellite gravity observations together with independent corrections for surface mass variability. We identify a robust approximately 6-year signal in the Stokes coefficient Delta S22, while no corresponding stationary signal is detected in Delta C22. A signal with the same periodicity is independently detected in length-of-day variations (Delta LOD), and the two signals exhibit a near anti-phase relationship. Interpreting this coupled signature within the proposed framework allows us to constrain the inner core viscosity to approximately 4.6 (+/- 1.8) x 10^16 Pa s and the equatorial relief of the inner core boundary to a semi-axis difference of about 200 +/- 70 m. The inversion further indicates mean density anomalies of +5.5 (+/- 0.6) per mil at the base of LLVPs. These results indicate that satellite gravimetry provides a direct observational window into deep-Earth dynamics and the physical properties of Earth's deep interior.

Ambiguity resolution (AR) is fundamental to achieving high-precision real-time kinematic (RTK) positioning, yet its reliability remains compromised when conventional AR validation procedures fail. To address this limitation, this study proposes a robust ambiguity resolution framework that integrates partial ambiguity resolution (PAR) and best integer equivariant estimation (BIE), supported by a novel modified carrier-phase-based receiver autonomous integrity monitoring (CRAIM) strategy. The framework operates through a hierarchical three-tier fallback mechanism: Tier 1 employs PAR with modified CRAIM-based fault detection and exclusion (PAR-FDE) to deliver fixed solutions after conventional PAR failure; upon validation failure, Tier 2 activates BIE with modified CRAIM-based fault detection (BIE-FD) to provide robust alternatives; Tier 3 reverts to conventional BIE solutions only when both preceding stages are unsuccessful. The PAR-FDE component identifies and excludes incorrect integer ambiguities, whereas the BIE-FD component detects and isolates erroneous ambiguity candidates. Experimental validation using two distinct datasets demonstrates the framework’s superior performance in fault detection and exclusion during ambiguity resolution, concomitantly enhancing positioning accuracy, solution continuity, and availability. For Dataset No.1, the positioning accuracy of the framework achieves improvements of 89.08

Precise track measurement of the geometric state of high-speed railways is a prerequisite for their smooth and safe operation. Current track inspection trolleys, which integrate only an inertial navigation system (INS) and a total station (TS), rely entirely on the track control network (CPIII) deployed along the track when calibrating their absolute location to avoid INS errors. Due to the high dependency on the surrounding CPIII points, this method faces severe challenges in terms of operational efficiency and cost control. To address this issue, this study utilizes the fast and precise positioning capability of the Chinese Beidou System (BDS) and proposes a novel adaptive location calibration approach using tightly integrated BDS/TS data. Using the Kalman filtering framework, this approach integrates BDS observations with the TS distance measurements in the observation domain, and the number of CPIII points to be observed is adaptively reduced according to the surrounding environments. Thus, the absolute location of track inspection trolleys can be quickly and accurately calibrated without INS data, greatly reducing dependency on CPIII points. Experiments were conducted under two typical scenarios: open-sky and blocked BDS signals. The results demonstrate that, under open-sky scenarios, the adopted BDS-only solution achieves positioning errors of less than 1.0 cm in the north, east, and up directions within 5 min, completely getting rid of the reliance on the control network, while in obstructed scenarios, where the BDS-only solution fails to converge at the 1 cm level within 5 min, the tightly integrated BDS/TS approach, combined with CPIII data, enables fast convergence in the northward and eastward, with positioning errors of less than 1 cm. The proposed approach provides a novel location calibration scheme in the track geometric states measurement under different environments, effectively reducing the dependence of track measurement operations on CPIII points and significantly enhancing measurement efficiency and flexibility.

This article presents a tightly integrated fusion algorithm that accelerates precise point positioning (PPP) convergence and enhances localization robustness in complex urban environments. The proposed method combines a filter-based visual-inertial odometry (VIO) framework with semantic constraints from a high-definition (HD) vector map, incorporating lane-line and directional arrow features extracted via a neural network. These HD map observations provide reliable lateral and heading constraints, effectively compensating for the degraded observability and drift in traditional VIO systems. In satellite-visible conditions, they also offer accurate initial estimates that significantly accelerate PPP convergence. The system is evaluated in diverse real-world scenarios-including open roads, urban canyons, tunnels, and soundproofed elevated highways-demonstrating submeter positioning accuracy and PPP convergence within 3-10 s, representing up to a 98.33% reduction in convergence time compared to conventional PPP. Furthermore, the approach operates using low-cost inertial and visual sensors, ensuring practical deployment in large-scale autonomous navigation systems.

The diurnal cycle of Integrated Water Vapor (IWV) plays a key role in radiation, convection, and land–atmosphere interactions, yet its global characteristics and representation in atmospheric reanalysis products remain insufficiently quantified. Using a decade of GNSS observations from more than 6,000 stations, this study characterizes the global IWV diurnal variability and assesses the performance of ERA5 in reproducing these signals. GNSS IWV exhibits a coherent diurnal cycle dominated by the S₁ (24 h) harmonic, with annual-mean amplitudes decreasing from 3 kg m⁻² in the tropics to near zero at high latitudes, a global mean of 0.40 kg m⁻², and peak times clustered around 17.4 local time; the semidiurnal S₂ (12 h) component is weaker but similarly systematic. In contrast, ERA5 exhibits sharp artificial discontinuities at the transition times (09–10 and 21–22 UTC) of its 12-hour assimilation windows, affecting 54

Monitoring lake and reservoir water levels is critical for water resource management and flood risk mitigation. We integrate Sentinel-3A/B and ICESat-2 altimetry to reconstruct monthly water levels (2018-2023) for 7,433 lakes and reservoirs (>5 km(2)) across Asia and estimate their storage variations. Reservoirs exhibit a median annual water level change of 0.36 m/yr, far exceeding the 0.05 m/yr observed for lakes, highlighting their dominant role in surface water dynamics. Eight Asian basins flood events reveal that insufficient self-regulation capacity of lakes is the primary flood trigger, while large reservoirs effectively mitigate flood frequency and intensity through regulation. These findings emphasize the importance of high-precision satellite altimetry in surface water assessments and the critical role of reservoirs in modulating hydrological extremes under climate change.

High-resolution multi-temporal remote sensing imagery is essential for land-cover mapping and long-term environmental monitoring, but radiometric inconsistencies often reduce the reliability of cross-date comparison and change analysis. This problem is particularly severe in karst regions, where rugged terrain, terrain-induced shadowing, seasonal phenological variation, rock – vegetation spectral coupling, and atmospheric heterogeneity make pseudo-invariant feature (PIF) extraction difficult and weaken conventional relative radiometric normalization (RRN) methods. To address these challenges, we propose a multi-stage cascaded RRN framework, termed GCCM, which integrates Gaussian mixture model (GMM)-based probabilistic screening, dual-branch convolutional neural network (CNN)-based deep feature learning, and canonical correlation analysis (CCA) – multivariate alteration detection (MAD)-based statistical refinement. GCCM first screens candidate invariant regions, then learns robust spectral – spatial representations for reference and target images, and finally refines invariant-region selection in the learned feature space. Experiments were conducted on multi-temporal GF-1 and Sentinel-2 image pairs over a karst area in northern Guangxi, China, under large-temporal-span and pronounced seasonal-variation conditions. The results show that GCCM generally outperforms SAD-ED and IR-MAD in radiometric normalization accuracy and PIF quality, although improvements vary among datasets and sensors. In the challenging GF-1 large-temporal-span experiments, GCCM achieved a mean R2 of 0.800 and a mean RMSE of 0.007. In the seasonal-variation and Sentinel-2 experiments, most band-wise R2 values exceeded 0.900 with low RMSE values. Ablation analysis further showed that the complete GCCM achieved a mean R2 of 0.9370 and an RMSE of 0.0126 on four representative datasets. These results indicate that GCCM provides a robust and terrain-adaptive solution for RRN in complex karst environments.

Tropospheric delay is a major source of error in Satellite Laser Ranging (SLR) observations. The most typical correction method uses the zenith tropospheric delay combined with mapping functions, such as FCULa. Nevertheless, these functions often fail to accurately represent actual tropospheric delays of satellite laser ranging at low elevation angles, leading to systematic errors. Ray tracing, based on the principles of geometric optics, has been extensively studied in global navigation satellite systems (GNSS) and very long baseline interferometry (VLBI), where it has been shown to achieve higher accuracy in tropospheric delay estimation. However, its application in SLR and their evaluation remain insufficient. This study aims to improve tropospheric delay modeling in SLR by comprehensively assessing the accuracy and effectiveness of ray tracing methods and mapping functions. Meanwhile, the coefficients of the recommended FCULa model is recalibrated using high-precision ray tracing results. Results show that the 3D ray tracing method with ERA5 outperforms mapping functions overall, reducing errors by about 1 mm and improving accuracy by 2%-10%. The improvement is more significant at low elevation angles, where errors decrease by 2-3 mm, corresponding to a 4%-15% enhancement. When used in parameter estimation, ray tracing reduces station coordinate dispersion by about 5.2% on average and improves external consistency with GNSS-SLR local ties by 7%-12%. After calibrating the FCULa model parameters with 3D ray tracing results, the model's accuracy at low elevation angles improves by 1%-7%. This provides a new potential approach for precise tropospheric delay estimation in SLR.

Real-time global ionosphere map (RT-GIM) products can provide global ionospheric information in real time and are widely used for space weather monitoring and global navigation satellite system (GNSS) positioning, particularly for low-cost and single-frequency GNSS applications that rely on external ionospheric corrections. At present, several International GNSS Service (IGS) analysis centres have released RT-GIM products, including the Chinese Academy of Sciences (CAS), the Centre National d’Études Spatiales (CNES), the Universitat Politècnica de Catalunya (UPC) and Wuhan University (WHU). In addition, UPC and CAS also provide combined RT-GIM products, named IGS0 and IGS1 respectively. This study evaluated the accuracy of these products in 2024 and investigated their impacts on BeiDou-3 (BDS-3) single-frequency real-time precise point positioning (RT-PPP). Using the IGS final post-processed GIM as a reference, the annual mean root mean square (RMS) values for the products from CAS, UPC, WHU and CNES were 5.68, 7.29, 5.15 and 10.09 total electron content units (TECU), respectively, while those for IGS0 and IGS1 were 5.81 and 5.33 TECU. From a spatial perspective, the combined products IGS0 and IGS1 showed the lowest overall normalized root mean square error (NRMSE) levels, with values generally below 0.4 and relatively uniform distributions. With BDS-3 observations from globally distributed IGS stations, this study investigates the impact of RT-GIM products on BeiDou-3 single-frequency RT-PPP. The experimental results demonstrate that using RT-GIM as an external ionospheric constraint can reduce the mean convergence time of single-frequency PPP by 27.3%. During geomagnetically quiet periods, using the RT-GIM constraint can reduce the mean convergence time by 56.1% and improve the mean 3D positioning accuracy by 39.0%. During disturbed periods, the corresponding improvements are 42.9% and 34.1%, respectively.

Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) has emerged as an effective non-contact technique for monitoring snow depth, a critical indicator of global climate change. However, heterogeneous topography and inconsistent surface reflectivity introduce spatial variability into reflected signals, significantly degrading retrieval accuracy. This study proposes a multi-GNSS optimization framework based on Principal Component Analysis (PCA) that isolates the dominant common-mode signal from diverse observations, thereby effectively extracting true snow depth variations. The snow depths are retrieved from continuous multi-season GNSS observations at stations P351 and P676 and validated against in-situ measurements. The results indicate that the PCA-driven method integrates the dominant common-mode information from multi-satellite observations, enabling more stable snow depth time series across GNSS constellations. At station P351, the root-mean-square error (RMSE) of the original retrievals ranges from 18.2 to 25.6 cm, whereas the proposed method reduces it to 10.8–11.8 cm. At station P676, the original performance is substantially worse, with RMSEs of 33.9–47.7 cm, while the improved method reduces the error to 12.8–13.3 cm. The PCA-driven framework also effectively captures snow-cover dynamics, achieving its highest accuracy during stable snow-covered periods, with an RMSE as low as 4 cm. In addition, DEM analysis suggests that topographic differences are associated with substantial residual post-PCA biases. This indicates that the new method can, to some extent, mitigate spatial correlation errors associated with terrain variations and scattering characteristics related to the movement of reflection points. Overall, the PCA framework can more fully leverage the observational advantages of multi-frequency and multi-system data, thereby better exploiting the monitoring potential of GNSS satellites, and providing a reliable solution for GNSS-IR snow depth monitoring in complex environments.

Accurate rainfall forecasting plays a crucial role in weather monitoring. Currently, the application of global navigation satellite system-derived precipitable water vapor (GNSS-PWV) has mainly focused on forecasting rainfall event occurrence, while neglecting the forecasting of rainfall amount. In this study, a new method based on machine learning fusion strategy and the constraint of rainfall events is proposed. The machine learning fusion strategy is used to improve the accuracy of rainfall amount forecasting by considering the difference in rainfall types and machine learning algorithms, while the rainfall event constraint strategy is used to reduce the rainfall amount forecasting error during periods without rainfall event. In the new method, the long short-term memory (LSTM) algorithm is adopted to forecast rainfall events by considering the temporal correlation of rainfall events. Then, the support vector regression (SVR), group method of data handling (GMDH), and harmony search (HS) algorithms are combined to forecast rainfall amount with the constraint of rainfall event forecast results. The SVR performs better at forecasting little and medium rainfall, while GMDH is better at forecasting heavy rainfall. HS is mainly used to optimize the parameters of the forecasting model. The datasets collected from 2019 to 2023 at Hong Kong Global Navigation Satellite System (GNSS) stations are used to evaluate the performance of the proposed method. The experimental results show that after adopting the proposed rainfall event forecasting method, the total true positive rate (TPR) is approximately 90.45%, and the prediction errors for without rainfall events are only about 4.09%. Moreover, after adopting the proposed rainfall amount forecasting method, the average root mean square error (RMSE) and mean absolute error (MAE) are approximately 0.42 and 0.13 mm, respectively. The average improvement rate of MAE is approximately 24.39% for the internal experiment and 25.27% for the external experiment compared to the single SVR model. In conclusion, the proposed method can effectively improve the accuracy of rainfall amount forecast, which can further provide support for meteorological monitoring.

Satellite Laser Ranging (SLR) is critical for establishing a high-precision terrestrial reference frame, but its accuracy is highly vulnerable to systematic errors. Among these, abnormal barometric pressure at SLR stations is a predominant yet often overlooked issue that distorts tropospheric delay corrections and introduces centimeter-to-decimeter biases in range observations. To address this, we propose a diagnosis strategy for abnormal on-site barometric pressure by a data-driven approach, specifically designed to support robust SLR tropospheric delay correction. In the diagnosis step, an imbalance-optimized CatBoost model is employed to accurately detect abnormal barometric pressure. This data-driven approach is augmented with auxiliary information derived from the Numerical Weather Model (NWM) to enhance diagnostic reliability. Based on the diagnosis results, an adaptive pressure difference model is then implemented for correction, where abnormal barometric pressure is strategically substituted with NWM-derived pressure. Validated on global SLR data from 35 stations in 2023, the proposed strategy demonstrates a high detection rate of 86.42% for abnormal barometric pressures in the testing dataset. After NWM-based correction, the RMS of posterior residuals associated with abnormal barometric pressure is reduced from 9.83 cm to 4.82 cm, corresponding to an improvement of 50.97% compared to the solution using the original barometric pressure. The strategy enables automated quality control, effectively mitigating systematic biases while preserving valuable observations, thereby significantly enhancing the precision and reliability of operational SLR solutions. (c) 2026 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

The eigenfrequency and quality factor of Earth’s nearly diurnal free wobble (NDFW) are critical parameters for investigating coupling mechanisms at the core-mantle boundary. However, discrepancies persist in existing studies that estimate NDFW parameters using superconducting gravimeter (SG) and very long baseline interferometry (VLBI) observations. In this work, we first conducted detailed harmonic analyses on nearly all SG records from the International Geodynamics and Earth Tide Service with durations exceeding 6 yr. After stringent statistical screening, we retained a high-quality dataset for NDFW parameter estimation, which consists of 33 SG records with low χ2 misfits. Unlike previous studies, we employed an affine-invariant ensemble Markov chain Monte Carlo (MCMC) sampling strategy within the Bayesian framework to perform stacking estimation on the selected dataset. The estimated complex eigenfrequency of the NDFW is (15.006147±0.000113)–i(4.74±0.29)×10−4 deg h−1, corresponding to a free core nutation period of 430.5±1.3 sidereal days and a quality factor of 15879±971. This result achieves higher precision than previous SGbased studies and shows good agreement with the VLBI measurements. Using these NDFW parameters, we inferred two key physical parameters of Earth’s interior. The first is the kinematic viscosity ν at the top of the outer core, estimated at (1.45±0.19)×103 cm2 s−1; the discrepancy between this result and those from high-pressure experiments warrants further investigation. The second is the compliance β, with a real part of (6.032±0.464)×10−4 and an imaginary part of (2.491±0.033)×10−5, reflecting the anelasticity of the real Earth. Collectively, these results provide new constraints for advancing research on geodynamic mechanisms in the deep Earth interior.

Abstract. The Surface Water and Ocean Topography (SWOT) mission reveals fine-scale ocean variability at wavelengths below 100 km. However, most oceanographic applications of satellite altimetry rely on gridded sea level anomaly (SLA) products derived from along-track observations, whose representation of fine-scale variability is strongly influenced by covariance assumptions. Here, we use SWOT-derived spatial and temporal decorrelation scales as observational constraints on the prior covariance structure for reconstructing fine-scale SLA variability. Autocorrelation functions derived from SWOT observations are used to quantify the global spatial and temporal decorrelation characteristics of SLA, and the resulting decorrelation scales are incorporated into an optimal interpolation framework to improve fine-scale SLA reconstruction. In two subpolar regions, the SWOT-informed mapping resolves SLA variability down to wavelengths of approximately 55 km and improves the representation of geostrophic velocity, strain, relative vorticity, Okubo–Weiss parameter, and eddy kinetic energy. In the Irminger Sea, the reconstructed fields reveal an earlier wintertime enhancement of fine-scale eddy kinetic energy that is attenuated in existing gridded products. These results demonstrate that SWOT observations can enhance SLA mapping not only by providing additional measurements, but also by constraining the covariance structure underlying the reconstruction of SLA variability.

Accurate and rapid short-term (up to 30 days in advance) polar motion (PM) predictions are critical for real-time applications like earthquake monitoring and early warning, global navigation satellite system (GNSS) meteorology, etc. Traditional prediction models, such as the least squares (LS) model, primarily rely on empirical periodic signals with constant amplitude and phase for extrapolation. However, due to complicated internal and external geophysical processes, these signals exhibit irregular variations rather than remaining constant, making it challenging for traditional methods to resolve them autonomously, especially in short-term predictions. To address this issue, we propose a method that combines the LS model with time-varying PM characteristics (TVLS) using the Prony method and the autoregressive integrated moving average (ARIMA) model, along with the effective angular momentum (EAM) data, to enhance the accuracy of short-term PM prediction. Compared with the official predictions disseminated by the International Earth Rotation and Reference Systems Service (IERS), the proposed method improves the prediction accuracy of PMX and PMY by up to 60.84% and 56.70%, respectively. Our method also outperforms the LS + AR + EAM forecast models from the Second Earth Orientation Parameters Prediction Comparison Campaign (2nd EOP PCC), ranking first for forecast horizons beyond 7 days for our predicted PMX and 12 days for PMY. The improvement can be attributed to the core feature of the TVLS model, which constructs a model for the main components of the PM periodic signal based on the Prony method, effectively capturing the non-stationary characteristics by addressing amplitude and phase variations. Therefore, we conclude that the proposed method could significantly enhance short-term PM prediction accuracy and has potential applications in the fields such as real-time satellite orbit determination, precise positioning and navigation.