YFYanming Feng

Yanming Feng

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Yanming Feng has a total of 103 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    TED: a global temperature-driven thermoelastic displacement dataset for GNSS reference stations (2000–2023)

    by Ran Lu, Zhao Li, Lingfeng Ye, Peng Yuan, Yanming Feng

    Nonlinear variations in GNSS station height time series include contributions from non-tidal mass loading and temperature-driven thermoelastic displacement (TED). However, a globally consistent station-level TED product has not been widely available. Here we present a global dataset of vertical TED for approximately 14 000 GNSS stations over 2000–2023, generated using a full-spectrum layered finite-element model. The model is driven by hourly ERA5 soil-temperature forcing and parameterized with SoilGrids-derived depth-dependent subsurface properties, allowing within-day temperature variability to be incorporated into daily TED time series that retain seasonal and nonseasonal components. For the 2612 stations with documented foundation-depth information, site-information-constrained solutions agree closely with SoilGrids-based solutions, with annual-amplitude differences mostly about 0.01–0.03 mm and annual-phase differences generally within 1–3°. Sensitivity tests using ± 10 % perturbations in the linear thermal expansion coefficient, thermal diffusivity, and Young's modulus indicate that the amplitude and phase of the annual TED component are stable under reasonable parameter uncertainty. Globally, annual TED amplitudes are mostly within about 0.2–1.2 mm and locally exceed 1.5 mm, with larger signals concentrated in arid inland and continental regions. The percentile-based peak-to-peak variability of the full daily TED series commonly reaches 2–4 mm and locally approaches 5–6 mm. When applied alone to GNSS height time series, TED produces positive WRMS reductions at about 79 % of the assessed stations, with most improvements within 3 %–5 % and the largest reductions reaching about 10 %. The dataset is publicly available for direct use in GNSS height time series correction and related geophysical applications: https://doi.org/10.5281/zenodo.21069958 (Lu et al., 2026a).


  • Open Access

    TED: A global temperature-driven thermoelastic displacement dataset for GNSS reference stations (2000–2023)

    by Ran Lu, Zhao Li, Lingfeng Ye, Peng Yuan, Yanming Feng

    The nonlinear signals in global GNSS station height time series reflects both non-tidal mass loading (atmospheric, oceanic, and hydrological) and temperature-driven thermoelastic deformation (TED). However, a globally consistent and reproducible TED data product has long been lacking. Here we present a global dataset of vertical TED for ~15,000 GNSS stations spanning 2000–2023, generated using a full-spectrum, layered finite-element model. The model is driven by hourly ERA5 soil-temperature profiles and parameterized with depth-dependent thermophysical properties from the SoilGrids dataset, enabling consistent quantification of TED from semi-diurnal/diurnal variability through seasonal to interannual timescales. Compared with an identical homogeneous-medium benchmark, subsurface stratification typically changes annual amplitudes by ~0.3 mm and shifts the timing of the annual maximum by ~1 month, yielding regionally coherent and smoothly varying spatial patterns. At stations with independent site characterization, the site-constrained solutions agree closely with SoilGrids-based solutions, with annual-amplitude differences of 0.01–0.03 mm and annual-phase differences mostly within 1–3°. Sensitivity tests using ±10% perturbations in thermal expansion, thermal diffusivity, and Young’s modulus indicate that annual-cycle amplitude and phase are robust. Globally, annual TED amplitudes are typically 1–2 mm, exceed 2–3 mm at some stations, and reach peak-to-peak values up to ~5 mm, with the largest signals concentrated in arid inland and continental climate regions. When TED corrections are applied together with non-tidal mass-loading corrections, the residual vertical dispersion decreases at most stations, with vertical scatter reduced by up to ~70 % at selected sites. The dataset is publicly available for direct use in GNSS coordinate time series correction and related geophysical applications: https://doi.org/10.5281/zenodo.18256342 (Lu et al., 2026).


  • Improving the Performance of GNSS Ambiguity Acceptance Test Using the RBF-SVM Approach

    by Lei Wang, Fei Yang, Yanqing Hou, Feng Zhou, Yanming Feng

  • Open Access

    Assessing the robustness of machine learning strategy for GNSS/INS vehicle positioning solutions enhancement

    by Wenzong Gao, Yanming Feng

    Accurate and reliable vehicle positioning during extended Global Navigation Satellite System/Inertial Navigation System outages remains challenging due to noise, outliers, and biases in input data. This study evaluates the robustness of machine learning (ML) models under two specific conditions: (1) noise and outliers present in training target data, and (2) errors and biases within input data. Two ML methods–Long Short-Term Memory (LSTM) and Gradient Boosting Decision Tree (GBDT)–were evaluated using the proposed methodology. Results indicate that both ML methods significantly enhance horizontal positioning accuracy compared to the conventional Kalman filter (KF). Specifically, LSTM reduces horizontal RMSE by approximately 85.8% (from 39.5 to 5.6 m), while GBDT achieves a reduction of approximately 77.7% (to 8.8 m). Additionally, the robustness analysis confirms that the proposed LSTM-based strategy demonstrates substantial robustness by maintaining consistent prediction accuracy, even in the presence of significant errors and outliers in training data, as well as being minimally impacted by errors and biases introduced in input variables.


  • Using CYGNSS and L-Band Radiometer Observations to Retrieve Surface Water Fraction: A Case Study of the Catastrophic Flood of 2022 in Pakistan

    by Zhongmin Ma, Shuangcheng Zhang, Qi Liu, Yanming Feng, Qinyu Guo, Hebin Zhao, Yuxuan Feng

    Spaceborne Global Navigation Satellite System Reflectometry (GNSS-R) has shown potential for terrestrial applications. The feasibility of monitoring and mapping surface inundation using Cyclone Global Navigation Satellite Systems (CYGNSS) has been demonstrated in literature. Nevertheless, most studies have only classified the surface into two states, inundated and non-inundated, which do not meet the requirements of refined hydrological modeling. Here, we propose a new calculation flow to retrieve the surface water fraction (Wf) based on CYGNSS data by considering fractional and dynamic surface water monitoring for the first time. First, a newly proposed physics-based algorithm was utilized for coupling CYGNSS surface reflectivity (SR) and Soil Moisture Active Passive (SMAP) brightness temperature (Tb) to attenuate the effects of surface roughness and vegetation on CYGNSS observations. After removing the effects of surface roughness and vegetation, the SR was assumed to be equal to the sum of inundated surface reflectivity and the non-inundated surface reflectivity multiplied by the corresponding fraction. Thus, Wf was obtained on a per-pixel basis. Next, Wf retrieval results were validated using the case of the catastrophic floods of 2022 in Pakistan. A comparison with the Wf obtained using SMAP Tb showed high spatial and temporal similarity; nevertheless, the CYGNSS results had a higher spatiotemporal resolution. A quantitative pixel-by-pixel comparison with the Global Flood Monitoring (GFM) data derived from Sentinel-1 radar imagery, which had a finer resolution (~10m) showed that the average overall accuracy of Wf retrieval was 64.44% and the average commission error was 17.78%. Finally, we quantified the impact of land use and land cover on the retrieval results while analyzing the advantages and limitations of the method. Thus, this study provides new insights for the future use of spaceborne GNSS-R data for monitoring inland water bodies.


  • Deep-Learning-Based Scenario Recognition With GNSS Measurements on Smartphones

    by Zhiqiang Dai, Chunlei Zhai, Fang Li, Weixiang Chen, Xiangwei Zhu, Yanming Feng

    Smartphones are in everyone’s hands for applications including navigation and localization-based services, and scenario recognition is critical for seamless indoor and outdoor navigation. How to use smartphone sensing data to recognize different scenarios is a meaningful but challenging problem. To address this issue, we propose a structured grid-based deep-learning scenario recognition technique that uses smartphone GNSS measurements (satellite position, pseudorange, Doppler shift, and C/N0). In this work, the scenarios are grouped into four categories: deep indoors, shallow indoors, semi-outdoors, and open outdoors. The proposed approach utilizes Voronoi tessellations to obtain structured-grid representations from satellite positions and performs computations using convolutional neural networks (CNNs) and convolutional long short-term memory (ConvLSTM) networks. With only spatial information being considered, the CNN model is used to extract the features of Voronoi tessellations for scenario recognition, achieving a high accuracy of 98.82%. Then, to enhance the robustness of the algorithm, the ConvLSTM network is adopted, which treats the measurements as spatiotemporal sequences, improving the accuracy to 99.92%. Compared with existing methods, the proposed algorithm is simple and efficient, using only GNSS measurements without the need for additional sensors. Furthermore, the latencies of the CNN and ConvLSTM models on a CPU are only 16.82ms and 27.94ms, respectively. Therefore, the proposed algorithm has potential for real-time applications.


  • Precise GNSS Time Synchronization With Experimental Validation in Vehicular Networks

    by Khondokar Fida Hasan, Yanming Feng, Yu-Chu Tian

    Time synchronization utilizing the Global Navigation Satellite System (GNSS) is being increasingly investigated for vehicular networks. Due to GNSS signal blockages, the availability and accuracy of GNSS timing solutions in various road settings is a recognized challenge. With the recent improvement of Multi-GNSS technology and the increased capacity of consumer-grade receivers, the application of GNSS in vehicular environments has brightened up. This paper systematically analyzes the required time synchronization of vehicular networks and presents a GNSS-based time synchronization solution. It also experimentally demonstrates the availability and capabilities of GNSS time synchronization using commercial-grade GNSS receivers and off-the-shelf communication devices. Our experiments show that the timing accuracy of an individual vehicular node can be as good as ±2 microseconds, resulting in synchronization accuracy of sub-10 microseconds among nodes. A momentary complete outage of the GNSS time solution due to signal blockage on the road adds clock error, leading to synchronization inaccuracy of up to sub-20 microseconds. This level of inaccuracy still meets the desired requirement for most applications in vehicular communication.


  • Open Access

    Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

    by Wenzong Gao, Zhao Li, Qusen Chen, Weiping Jiang, Yanming Feng

    Global navigation satellite system (GNSS) site coordinate time series provides essential data for geodynamic and geophysical studies, realisation of a regional or global geodetic reference frames, and crustal deformation research. The coordinate time series has been conventionally modelled by least squares (LS) fitting with harmonic functions, alongside many other analysis methods. As a key limitation, the traditional modelling approaches simply use the functions of time variable, despite good knowledge of various underlying physical mechanisms responsible for the site displacements. This paper examines the use of machine learning (ML) models to reflect the effects or residential effects of physical variables related to Sun and the Moon ephemerides, polar motion, temperature, atmospheric pressure, and hydrology on the site displacements. To form the ML problem, these variables are constructed as the input vector of each ML training sample, while the vertical displacement of a GNSS site is regarded as the output value. In the evaluation experiments, three ML approaches, namely the gradient boosting decision tree (GBDT) approach, long short-term memory (LSTM) approach, and support vector machine (SVM) approach, are introduced and evaluated with the time series datasets collected from 9 GNSS sites over the period of 13 years. The results indicate that all three approaches achieve similar fitting precision in the range of 3–5 mm in the vertical displacement component, which is an improvement in over 30% with respect to the traditional LS fitting precision in the range of 4–7 mm. The prediction of the vertical time series with the three ML approaches shows the precision in the range of 4–7 mm over the future 24- month period. The results also indicate the relative importance of different physical features causing the displacements of each site. Overall, ML approaches demonstrate better performance and effectiveness in modelling and prediction of GNSS time series, thus impacting maintenance of geodetic reference frames, geodynamics, geophysics, and crustal deformation analysis.