IAP-24-080
Machine learning for GNSS time series analysis: benchmarking and application in geosciences
Using Global Navigation Satellite System (GNSS) data to measure bedrock is important in many areas of geoscience:
• calibrating tide gauges for land motion
• understanding glacial isostatic adjustment and
• elastic rebound as ice load changes,
• hydrologic loading
• tectonics
• volcanology
Machine learning (ML) is at the forefront of time series analysis for GNSS data, particularly in identifying error sources. This project will benchmark ML techniques for time series analysis and apply them in an area of geosciences in which the student finds the most interest – though with an emphasis on geographic areas with low data availability, where ML is currently underutilised.
Methodology
Dr Petrie has substantial expertise in precise long-term GNSS analysis, particularly on the Antarctic continent. Prof Clarke will contribute expertise in sidereal filtering and in tectonic applications of GNSS if that is the student’s application area of interest. Prof Ray is an expert in statistics, with a particular interest in the technique of functional data analysis applied to
environmental datasets as well as machine learning.
Rates obtained from GNSS time series may be seriously affected by unattributed/unmodelled effects on the GNSS. Effects may include offsets, ionospheric and tropospheric effects, hydrologic effects and non-line-of-sight and multipath errors. The student will investigate and benchmark ML techniques for identifying one or more of these effects, and then apply this to improve GNSS analysis procedure and/or time series analysis in a geographic area where ML is underutilised. Depending on the student’s interest and background, the project may lean more towards innovation in GNSS processing, developments aimed more at a particular application area, or innovation in ML/statistical techniques
Python will be the primary language for ML and statistical processing. GNSS processing will be performed using one of GIPSY-X, GAMIT, or GROOPS scientific packages.
Project Timeline
Year 1
Review existing literature, produce benchmarking data, learn to perform machine learning in Python, run initial tests on benchmarking data using ML techniques.
Year 2
Write paper on benchmarking results, present results at UK conference. Select GNSS dataset(s) of interest, learn to process GNSS data in scientific GNSS software, produce time series in GNSS software, develop ML methodology further.
Year 3
Apply developed ML techniques to specific application area dataset. Present at International conference.
Year 3.5
Write paper on results, complete thesis.
Training
& Skills
During the course of the project, the student will gain an in-depth understanding of GNSS data and state of the art precise GNSS processing, the relevant GNSS application area, time series/noise analysis, and machine learning. The student will make several visits to Newcastle during the course of the PhD. Technical/paper writing skills and presentation skills will also be developed. Conference attendance is likely to be European Geophysical Union (EGU) and a second national/international conference.
References & further reading
Mohanty, A. and Gao, G., 2024. A survey of machine learning techniques for improving Global Navigation Satellite Systems. EURASIP Journal on Advances in Signal Processing, 2024(1), p.73.
Gao, W., Li, Z., Chen, Q., Jiang, W. and Feng, Y., 2022. Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches. Journal of Geodesy, 96(10), p.71.
Siemuri, A., Selvan, K., Kuusniemi, H., Valisuo, P. and Elmusrati, M.S., 2022. A systematic review of machine learning techniques for GNSS use cases. IEEE Transactions on Aerospace and Electronic Systems, 58(6), pp.5043-5077.
Crocetti, L., Schartner, M. and Soja, B., 2021. Discontinuity detection in GNSS station coordinate time series using machine learning. Remote Sensing, 13(19), p.3906.
Blewitt, G., Hammond, W. and Kreemer, C., 2018. Harnessing the GPS data explosion for interdisciplinary science. Eos, 99(2), p.e2020943118.
Williams, S.D., Bock, Y., Fang, P., Jamason, P., Nikolaidis, R.M., Prawirodirdjo, L., Miller, M. and Johnson, D.J., 2004. Error analysis of continuous GPS position time series. Journal of geophysical research.
Herring, T.A., 1999. Geodetic applications of GPS. Proceedings of the IEEE, 87(1), pp.92-110..
