Scientific Quarterly Journal of Geosciences

Scientific Quarterly Journal of Geosciences

Physics-informed neural networks for GPS velocity field interpolation in the Alborz tectonic region

Document Type : Original Research Paper

Authors
Asghar Rastbood 1
Nima Jahanghiri 2
Ebrahim Aghazadeh Chakherlou 1
1 Department of Surveying Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran
2 Department of Surveying Engineering, Ahar Azad University, Ahar, Iran
10.22071/gsj.2026.553765.2232
Abstract
The interpolation of GPS velocity vectors into continuous fields represents a significant challenge in geodesy and geophysics. Conventional methods, such as the elastic Green's function proposed by Sandwell and Wessel (2016), face limitations in modeling complex phenomena and accounting for data uncertainties. This paper investigates the application of Physics-Informed Neural Networks as a powerful alternative to these classical methods. In this study, a PINN model was implemented that directly incorporates the governing equations of elasticity into the Neural Network's loss function. The model was trained on GPS data from 89 stations across the Alborz Tectonic Region in the oblique collision zone of the Arabia-Eurasia tectonic plates. Results demonstrate that the proposed model can successfully reconstruct the velocity field with acceptable accuracy, achieving RMSE values of approximately 0.68 mm/yr and 0.99 mm/yr for the east and north components, respectively. The method offers several advantages, including high flexibility in modeling complex physics, the capability to integrate diverse data types, and automatic consideration of observational uncertainties. Although the computational time of this approach is longer compared to classical methods, its inherent ability to overcome the limitations of traditional techniques makes it a promising candidate for the next generation of geodynamic data processing tools.
Keywords
Physics-Informed Neural Networks
Interpolation
GPS velocity vector
Elastic Green's functions
Machine learning
Subjects
Anitescu, C., İsmail Ateş, B., and Rabczuk, T., 2023. Physics-Informed Neural Networks: Theory and Applications. In: Machine learning in modeling and simulation, Chapter 5, pp. 179-218.
Boresi, A. P., and Schmidt, R. J., 2002. Advanced mechanics of materials. John Wiley & Sons.
Cuomo, S., Di Cola, V. S., Giampaolo, F., Rozza, G., Raissi, M., and Piccialli, F., 2022. Scientific machine learning through physics–informed neural networks: Where we are and what’s next. Journal of Scientific Computing, 92(3), 88.
Dahal, A., and Lombardo, L., 2025. Towards physics-informed neural networks for landslide prediction. Engineering Geology, 344, 107852.
Ghods A., Shabanian E., Bergman E., Faridi M., Donner S., Mortezanejad G., and Aziz-Zanjani A., 2015. The Varzaghan–Ahar, Iran, Earthquake Doublet (Mw 6.4, 6.2): implications for the geodynamics of northwest Iran. Geophys. J. Int., 203, 522–540.
Hessami, K., Jamali, F., and Tabassi, H., 2003. Major Active Faults of Iran (map), Ministry of Science, Research and Technology, International Institute of Earthquake Engineering and Seismology.
Hornik, K., Stinchcombe, M., and White, H., 1989. Multilayer feedforward networks are universal approximators. Neural networks, 2(5), 359-366.
Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., and Yang, L., 2021. Physics-informed machine learning. Nature Reviews Physics, 3(6), 422-440.
Kay, S. M., 1993. Fundamentals of statistical signal processing: estimation theory. Prentice-Hall, Inc.
Khorrami F., Vernant P., Masson F., Nilfouroushan F., Mousavi Z., Nankali H., Saadat S. A., Walpersdorf A., Hosseini S., Tavakoli P., Aghamohammadi A., and Alijanzade M., 2019. An up-to-date crustal deformation map of Iran using integrated campaign-mode and permanent GPS velocities. Geophys. J. Int., 217, 832–843.
Kostrov, V. V., 1974. Seismic moment and energy of earthquakes, and the seismic flow of rock, Izv. Acad. Sci. USSR. Phys. Solid Earth, 1, 23-44.
Leys, C., Ley, C., Klein, O., Bernard, P., and Licata, L., 2013. Detecting outliers: Do not use standard deviation around the mean, use absolute deviation around the median. Journal of experimental social psychology, 49(4), 764-766.
Masson, F., Chéry, J., Hatzfeld, D., Martinod, J., Vernant, P., Tavakoli, F., and Ghafory-Ashtiani, M., 2005. Seismic versus aseismic deformation in Iran inferred from earthquakes and geodetic data. Geophysical Journal International, 160(1), 217-226.
Masson, F., Lehujeur, M., Ziegler, Y., and Doubre, C., 2014. Strain rate tensor in Iran from a new GPS velocity field. Geophysical Journal International, 197(1), 10-21.
Maurer, J., and Materna, K., 2023. Quantification of geodetic strain rate uncertainties and implications for seismic hazard estimates. Geophysical Journal International, 234(3), 2128-2142.
Motavalli‐Anbaran, S. H., Zeyen, H., Brunet, M. F., and Ardestani, V. E., 2011. Crustal and lithospheric structure of the Alborz Mountains, Iran, and surrounding areas from integrated geophysical modeling. Tectonics, 30(5).
Okazaki, T., Ito, T., Hirahara, K., and Ueda, N., 2022. Physics-informed deep learning approach for modeling crustal deformation. Nature Communications, 13(1), 7092.
Okazaki, T., Fukahata, Y., and Nishimura, T., 2021. Consistent estimation of strain-rate fields from GNSS velocity data using basis function expansion with ABIC. Earth, Planets and Space, 73(1), 153.
Poulet, T., and Behnoudfar, P., 2023. Physics-informed neural network reconciles Australian displacements and tectonic stresses. Scientific Reports, 13(1), 23095.
Raissi, M., Perdikaris, P., and Karniadakis, G. E., 2019. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378, 686-707.
Rashidi, A., Kianimehr, H., Yamini-Fard, F., Tatar, M., and Zafarani, H., 2022. Present stress map and deformation distribution in the NE Lut Block, eastern Iran: Insights from seismic and geodetic strain and moment rates. Pure and Applied Geophysics, 179(5), 1887-1917.
Rastbood, A., 2023. Present-day crustal deformation in Alborz region based on GPS velocity field and spline functions. Iranian Journal of Geophysics (IJG), 17(2).
Rastbood, A., Salmanian, M., and Hossainali, M. M., 2024. Present-day crustal deformation based on an interpolated GPS velocity field in the collision zone of the Arabia-Eurasia tectonic plates. Studia Geophysica et Geodaetica, 68(3), 191-215.
Sadd, M. H., 2009. Elasticity: theory, applications, and numerics. Academic Press.
Sandwell, D. T., and Wessel, P., 2016. Interpolation of 2-D vector data using constraints from elasticity. Geophysical Research Letters, 43(10), 10,703-10,709.
Shen, Z. K., Wang, M., Zeng, Y., and Wang, F., 2015. Optimal interpolation of spatially discretized geodetic data. Bulletin of the Seismological Society of America, 105(4), 2117-2127.
Shen, Z. K., Jackson, D. D., and Ge, B. X., 1996. Crustal deformation across and beyond the Los Angeles basin from geodetic measurements. Journal of Geophysical Research: Solid Earth, 101(B12), 27957-27980.
Talebian, M., Ghorashi, M., and Nazari, H., 2013. Seismotectonic map of the Central Alborz, Research Institute for Earth Sciences, Geological Survey of Iran.
Zandt, G., and Ammon, C. J., 1995. Continental crust composition constrained by measurements of crustal Poisson's ratio. Nature, 374(6518), 152-154.
Scientific Quarterly Journal of Geosciences
Volume 36, Issue 2 - Serial Number 140
Summer 2026, Vol. 36, Issue 2, Serial No. 140
Spring 2026
Pages 149-170