Global Seismicity AtlasOʻzbekchaРусскийEspañol

Induced seismicity

Induced earthquakes are caused or triggered by human activity, such as producing gas, injecting waste water or filling a reservoir; the Groningen gas field in the Netherlands is the best-recorded European example.

Interactive diagram · Real example from the atlas →

Simple explanation

Taking gas out of a porous sandstone lowers the pressure in it. The rock compacts by a fraction of a millimetre per metre, and the faults that cross the reservoir take extra stress. When that stress is released, the fault slips: a small earthquake, usually shallow (about 3 km), so even magnitude 3 can damage houses.

The number of such earthquakes follows the production history, but with a delay and not one to one. Groningen produced gas from 1963; the first recorded earthquake was in 1991, the largest (ML 3.6) in 2012, and earthquakes continued after production stopped in 2023.

The atlas shows the record: production per month, earthquakes per month, b-values and a comparison with the published studies. It does not model the physics and does not forecast future earthquakes.

An everyday picture

Squeezing a sponge that is glued into a frame: the harder and longer you squeeze, the more the glue joints crack, and some keep cracking for a while after you let go.

Technical details

Data: KNMI catalogue (FDSN event service, local magnitude ML, event type "induced or triggered event"), NLOG monthly production of the Groningen field (10⁹ m³, from 2003), study region the NLOG field outline buffered by 2 km, which reproduces the counts of Dost et al. (2017) within 2 %.

Rates use ML ≥ 1.5 for the whole period, the conservative completeness of the original network (Dost et al. 2017; Kraaijpoel et al. 2022). Phase rates carry exact Poisson 95 % intervals. b-values use three estimators: the atlas MAXC + 0.2 with bootstrap, Aki-Utsu at a fixed ML 1.5, and b-positive (van der Elst 2021).

The correlation is a Spearman rank correlation of 22 annual values (2003–2024). Annual values are autocorrelated, so p-values are indicative only. The physical link (pressure depletion, compaction and stress on faults) is modelled by Bourne et al. (2014) and later work; the atlas does not model it.

Validation against published values (Dost et al. 2017; Kraaijpoel et al. 2022; Gulia & Wiemer 2026): the classic field-wide numbers are reproduced; one map comparison differs because the KNMI catalogue alone maps only the densest part of the field. The table is on the Case studies page.

Further reading

  1. Dost, B., Ruigrok, E. & Spetzler, J. (2017). Development of seismicity and probabilistic hazard assessment for the Groningen gas field. Netherlands Journal of Geosciences, 96(5), s235–s245. doi.org/10.1017/njg.2017.20
  2. Bourne, S. J., Oates, S. J., van Elk, J. & Doornhof, D. (2014). A seismological model for earthquakes induced by fluid extraction from a subsurface reservoir. Journal of Geophysical Research: Solid Earth, 119, 8991–9015. doi.org/10.1002/2014JB011663
  3. Kraaijpoel, D., Martins, J. E., Osinga, S., Vogelaar, B. & Breunese, J. (2022). Statistical analysis of static and dynamic predictors for seismic b-value variations in the Groningen gas field. Netherlands Journal of Geosciences, 101, e18. doi.org/10.1017/njg.2022.15
  4. Gulia, L. & Wiemer, S. (2026). Confirming b-value spatial variations in Groningen gas field using machine-learning-derived seismic data. Netherlands Journal of Geosciences, 105, e13892. doi.org/10.70712/NJG.v105.13892
  5. KNMI (1993). Netherlands Seismic and Acoustic Network. Royal Netherlands Meteorological Institute. doi.org/10.21944/e970fd34-23b9-3411-b366-e4f72877d2c5
← How well earthquakes are recordedSeismic moment budget →