Integrating Geodesy and Geophysics to monitor and model the Dutch Subsurface

Gas production-induced surface subsidence is a significant problem in the province of Groningen, affecting the environment, infrastructure and water management. Also, there is correlation between surface subsidence and induced seismicity (Bourne et al., 2015). Both subsidence and seismicity are likely linked by their common origin of reservoir compaction. Adaptation to this primary subsidence, by changing and compartmentalizing the ground water table, causes a secondary subsidence signal, strongly dependent on soils and the shallow subsurface. Continuous, frequent and accurate estimates and predictions of subsidence over the entire Groningen field are therefore highly relevant.

The principal innovative aspect of this collaborative research proposal is that we will jointly use InSAR observations and process-driven models for the deep and shallow subsurface to significantly improve the accuracy of geodetic subsidence estimates and to better constrain the model parameters. The relevance of improving model parameters is that they control physical processes in the subsurface, e.g., rates of reservoir stress relaxation and fault creep, hence establishing an observationally constrained basis for forecasting hazards from various gas production scenarios. Similarly, model parameters of processes in the shallow subsurface are relevant since they are the main driver for strain on buildings and infrastructure. Recently developed data assimilation methods will be used to integrate the geodetic and geophysical observations and models. This way, subsidence is not only an undesired consequence of subsurface exploitation, but also an important source of information to understand it. 

Scientific Project description

Intensive use of the subsurface is increasingly difficult in developed societies. While the natural resources are still necessary, society is critical on the associated hindrance. The Groningen area is currently undoubtedly one of the most significant examples. Hindrance is caused by induced seismicity and surface subsidence. While these effects have long been a nuisance with limited consequences, recently the significance of these effects, particularly the seismicity, has reached unprecedented levels, in terms of risk, societal unrest, governmental and industrial accountability, and the national economy [FD, 2018]. The governmental decision to cease hydrocarbon extraction from the Groningen gas field [Reuters, 2018] is the latest development in this context.

Several questions arise in connection to this problem. Could the problems of induced seismicity have been avoided using a different production strategy? Will a better understanding of the geophysical context help in improving current and future policy decisions? Are there sources of information that are readily available, but have not been used yet? Could we do better? Here we argue that the answer to all of the questions above should be ‘yes’.

Our overall aim of this project is to significantly improve our ability to predict the effects of interaction with the subsurface, in order to avoid or mitigate undesired effects. This is necessary both in a phasing-out scenario of gas extraction, as well as for operational or planned production. We achieve this goal by proposing a paradigmatic change in the use of geodetic data in conjunction with geophysical data and models.

The current paradigm in this respect is to (i) strictly separate geodetic and geophysical data processing and modeling, and (ii) to either “confront the geophysical model results with the geodetic data” or to “use the geodetic results as input for the geophysical models”. In the former, geophysical model results are supposed to be created from geophysical data only, used in a forward model to predict geometric changes at the earth’s surface, which is subsequently compared with geodetic observables of these geometric changes. This way, the geodetic data serve as arbiter: i.e., to decide whether the model predictions were adequate. In the latter, the geodetic results of geometric changes at the earth’s surface are ingested as input data for the geophysical model, and can consequently not be used as arbiter anymore. In both ways, there is a one-way interaction between geodesy and geophysics: only the direction is different. Proponents of this paradigm argue that this way, one avoids an incestuous relation where geophysical model results could influence the geodetic data processing and (un)consciously steer the results to a desired outcome.

There are important flaws to this paradigm. First, there is a difference between geodetic measurements and geodetic results, i.e., geodetic measurements need to be processed to obtain results (estimates), which requires a model as well. The parameterization of this model is a matter of choice, expertise, and the optimal use of a priori (contextual) information. By excluding a potential source of information, i.e. geophysical information, the geodetic model will be sub-optimal, yielding sub-optimal estimates. Second, the one-way interaction prohibits iterative improvements. For example, it may be that the geodetic data are accurate, but not representative for the signal of interest. Vice versa, the geophysical models may completely miss a significant driving mechanism, which could have been identified by improved geodetic data analysis. Third, this paradigm stems from times when (geodetic) data were expensive and non-abundant; this has dramatically changed nowadays.

Our objective is a paradigmatic change, in which we establish an integrated continuous iterative optimization procedure, using data assimilation, between geophysical modeling (of different strata and mechanisms) and geodetic data processing. We will apply this mainly for the Groningen situation, as further described in five work packages below. The likelihood of success for this study is high, due to a number of factors. Both temporally as well as spatially, the production of the Groningen gas field shows (and will show) considerable variability, which will become well observable by new geodetic techniques, most notably InSAR and GNSS. Moreover, in contrast to earlier studies, we specifically model the interdependence of various driving mechanisms, such as (i) pressure change in the reservoir, (ii) aquifer effects aside and below the reservoir, (iii) creep of the cap rock (salt), (iv) heterogeneity of the overburden, (v) effective ground water level change due to subsidence, and (vi) consequent volume change in shallow peat and clay layers, (vii) deliberate ground water level changes, and (viii) geotechnical dynamics of structures in and above the surface, see Figure 1. Finally, new geodetic data (InSAR) have become available with a daily interval and unprecedented precision.

Figure 1 Primary and secondary ground motion (due to gas production) on an area with organic soils (brown) and a ground water table. (A) start situation in which a deep driving mechanism is causing subsidence. (B) Water tables would effectively rise, in this sketch to form a lake. (C) Groundwater pumping would yield compaction and oxidation (green arrows) and uplift due to upward pressure (red arrow). (D) At the surface, the primary subsidence bowl widens and veers up in the center. In case of compartmentalization (E) areas with an increased vadose zone subside and vice versa. (F) shows stronger spatial gradients at the surface.

The quantification of the various processes is only possible with prior information contained in the models and the observations. Data assimilation methods provide a means to combine the observations with the models in such a way that we find the most likely solution given the prior assumptions. Reservoir and overburden models and mechanics are available, but need to be iteratively adapted to ingest the high-precision observations. Long-term subsidence due to consolidation and creep processes in the shallow subsurface are reasonably well covered by geophysical models, but these need further validation and hence optimization of the individual modelled processes acting with different rates leading to changes in surface elevation. Processes acting on the short term, e.g. shrinkage and swelling due to seasonal fluctuations in the groundwater table and heave due to subsidence-related relative groundwater level rise, are not incorporated in the models yet. This could effectively be done by making use of the new high spatio-temporal resolution of geodetic data.

Our program consists of 5 work packages, related to 4 PhD projects, all focusing on detecting and modeling the temporal variability of the various driving mechanisms, at fine (sub-km) spatial scales.

  • WP 1 Improvement of geodetic models and parameter estimation: InSAR data
  • WP 2 Improvement of geophysical models at sub-Holocene depths: reservoir and overburden
  • WP 3 Improvement of geophysical models at Holocene and Pleistocene depths: soils, hydrology and water management
  • WP 4 Data assimilation: analysis of model-data differences to help quantify model uncertainties
  • WP 5 (cross cutting WP1-4): Advancement of numerical data processing techniques

The permanent scientific staff of SUBSIDENCE consists of the following people:

  • Prof. Ramon Hanssen, TU Delft (PI, WP1 and WP5 lead)
  • Dr. Rob Govers, Utrecht University (Co-PI and WP 2 lead)
  • Dr. Esther Stouthamer, Utrecht University (Co-PI and WP 3 lead)
  • Femke Vossepoel, TU Delft (Co-PI and WP 4 lead)

In addition the project team will encompass four PhD candidates (in WP 1, 2, 3, and 4), and a Computational Data Analytics expert. The PhD candidates of WP1 and WP4, as well as the Computational Data Analytics expert will be primarily stationed in Delft, while the PhD candidates of WP2 and WP3 will be primarily stationed in Utrecht. The candidates are expected to work as a team, and will work one day per week at the other university. The PhD openings within SUBSIDENCE of Utrecht University website will be jointly made available via the website of Utrecht University (, but applications can also be directed to the PI of the project. 

Project organisation

The integrated approach of SUBSIDENCE is founded on the involvement of staff of various fields of expertise, employed Delft University of Technology and Utrecht University. The collaborative work is critical for the success of this project and will form the central piece in the network of institutions involved.

To reach the research objectives, five people will be recruited: two PhD students at Delft University of Technology, two PhD students at Utrecht University and a Computational Data Analytics expert (could be postdoc, engineer, or technician) at Delft University of Technology. Strong interaction between all researchers is important for the success of the project. There will be a work space for the group both in Delft as well as in Utrecht, and the Delft PhD students are expected to spend one day per week at their Utrecht counterparts, and vice versa.

Researchers involved:

Through regular visits of staff to various knowledge institutions, the research team will be able to work in seamless collaboration, leading to inspired partnership in follow-up projects.

Literature references

Bakr, M., 2015. Influence of Groundwater Management on Land Subsidence in Deltas. Water Resources Management 29 (5), pp. 1541–1555.

Bourne, S. J., Oates, S. J., Elk, J., & Doornhof, D., 2015. A seismological model for earthquakes induced by fluid extraction from a subsurface reservoir, 1–25.

Den Haan, E.J., 1994. Vertical compression of soils. Delft University of Technology, PhD thesis, Delft, The Netherlands.

Den Haan, E.J., 2008 (in Dutch). De intrinsieke tijd in het Isotachenmodel. Geotechniek, January, pp. 32-36.

Dost, B., Edwards, B., & Bommer, J. J., 2018. The Relationship between M and ML: A Review and Application to Induced Seismicity in the Groningen Gas Field, The Netherlands. Seismological Research Letters, 1–13. 

Erkens, G., van der Meulen, M.J., Middelkoop, H., 2016. Double trouble: subsidence and CO2 respiration due to 1,000 years of Dutch coastal peatlands cultivationHydrogeology Journal, 24 (3), pp. 551-568.

FD, 2018.  Gas extraction has led to earthquakes. This has caused damage and major societal unrest. English translation of an article published today by the Dutch Financial Times, Financieele Dagblad, 2 February 2018

Glegola, M.A., P.G. Ditmar, R.G. Hanea, F.C. Vossepoel, R.J. Arts, and R. Klees, 2011. Gravimetric monitoring of water influx into a gas reservoir: A numerical study based on the ensemble Kalman filter, SPE Journal, doi:10.2118/149578-PA

Hanssen, R.F., 2001. Radar interferometry: data interpretation and error analysis (Vol. 2). Springer Science & Business Media.

Hanssen, R. F. 2017. Device with integrated double SAR corner-reflectors, GNSS, leveling, Triangulation and photogrammetric benchmarks. Netherlands patent application 2019103.21 June 2017.

Kooi, H., P. Johnston, K. Lambeck, C. Smither and R. Molendijk, 1998. Geological causes of recent (~100 yr) vertical land movement in the Netherlands, Tectonophysics, 299, pp. 297-316.

Kooi, H., Bakr, M., De Lange, G., Den Haan, E., Erkens, G., 2017. A user guide to SUB-CR a modflow land subsidence and aquifer system compaction package that includes creep. Deltares-rapport 11202275-008, 57 p.

Kroon, I.C., Nguyen, BL., Fokker, P.A. et al., 2009. Disentangling Shallow and Deep Processes Causing Surface Movement

Math Geosci 41: 571.

Leeuwen, P.J. van, 2009. Particle filtering in geophysical systems, Mon. Wea. Rev., doi:10.1175/2009MWR2835.1

Leeuwen, P.J. van, and G. Evensen, 1996. Data assimilation and inverse methods in terms of a probabilistic formulation, Mon. Wea. Rev., doi:10.1175/1520-0493(1996)124<2898:DAAIMI>2.0.CO;2

Lele, S. P., Hsu, S.-Y., Garzon, J. L., DeDontney, N., Searles, K. H., Gist, G. A., et al., 2016. Geomechanical modeling to evaluate production-induced seismicity at Groningen field (pp. 1–18). Presented at the Abu Dhabi International Petroleum Exhibition Conference, Abu Dhabi.

Mahapatra, P.S., Samiei-Esfahany, S., van der Marel, H., & Hanssen, R.F., 2014. On the use of transponders as coherent radar targets for SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing52(3), 1869-1878.

Marketos, G., Govers, R., & Spiers, C.J., 2015. Ground motions induced by a producing hydrocarbon reservoir that is overlain by a viscoelastic rocksalt layer: a numerical model. Geophys. J. Int.203(1), 198–212./10.1093/gji/ggv294.

Marketos, G., D.B.T. Broerse, C.J. Spiers, and R. Govers, Long-term subsidence study of the Ameland gas field: time-dependence induced by rocksalt flow, Report to the Steering Committee for the long-term subsidence study in the Wadden Sea region, June 26, 2015.pdf

Minderhoud, P.S.J., Erkens, G., Pham, V.H., Bui, V.T, Erban, L., Kooi, H.,  Stouthamer, E., 2017. Impacts of 25 years of groundwater extraction on subsidence in the Mekong delta, Vietnam. Environmental Research Letters 12 (6), pp. 1-13. 

Oliver, D.S., A.C. Reynolds, and N. Liu, 2008. Inverse Theory for Petroleum Reservoir Characterization and History Matching, Cambridge University Press, ISBN: 9780521881517, 392 pp.

Peters, L., R.J. Arts, G.K. Brouwer, C.R. Geel, S. Cullick, R.J. Lorentzen, Y. Chen, N. Dunlop, F.C.Vossepoel, R. Xu, P. Sarma, A.H.H. Alhuthali, and A.C. Reynolds, 2010. Results of the Brugge benchmark study for flooding optimization and history matching, SPE Reservoir Evaluation and Engineering, doi:10.2118/119094-PA 

Reuters, 2018. Netherlands to halt gas production at Groningen by 2030, 29 March 2018

Spiers, C.J., Schutjens, P.M.T.M., Brzesowsky, R.H., Peach, C.J., Liezenberg, J.L. & Zwart, H.J., 1990. Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution, in Deformation Mechanisms, Rheology and Tectonics, Vol. 54, pp. 215–227, eds Knipe, R.J. & Rutter, E.H., Geological Society of London Special Publication.

Terzaghi, K., 1943. Theoretical soil mechanics. John Wiley and Sons, New York.

Van Asselen, S., Erkens, G., Stouthamer, E., Woolderink, H., Geeraert, R., Hefting, M., 2018. The relative contribution of peat compaction and oxidation to subsidence in built-up areas in the Rhine-Meuse delta, The Netherlands. Science of the Total Environment, in press.

Van Asselen, S., Karssenberg, D., Stouthamer, E., 2011. Contribution of peat compaction to relative sea-level rise within Holocene deltas. Geophysical research Letters 38 (24), art. no. L24401.

Van den Akker, J. J. H., van den Beuving, J., Hendriks, R.F.A., Wolleswinkel, R.J., 2007. Maaivelddaling, afbraak en CO2 emissie van Nederlandse veenweidegebieden. In: Leidraad bodembescherming #83, SDU-uitgevers, 5510-1-30.

Vossepoel, F.C., A.T. Weaver, J. Vialard, and P. Delecluse, 2004. Adjustment of near-equatorial wind stress with 4D-Var data assimilation in a model of the Pacific Ocean, Mon. Wea. Rev., 132, 2070–2083.

Vossepoel, F.C., and P.J. van Leeuwen, 2007. Parameter estimation using a particle method: Inferring mixing coefficients from sea-level observations, Mon. Wea. Rev., 135, 1006– 1020.