An advance has been made towards a method for forecasting earthquakes several months before they occur. The method relies on changes of groundwater chemistry as earthquake precursors. In a study published in 2014, we showed that changes of groundwater chemistry occurred prior to and were associated with two earthquakes of magnitude 5 and higher, which affected northern Iceland in 2012 and 2013. Here we test the hypothesis that similar or larger earthquakes could have been forecast in the following decade (i.e. 2014–2023) based on our published findings. We found that we could have forecast one of the three greater than magnitude 5 earthquakes that occurred. Noting that changes of groundwater chemistry were oscillatory, we infer expansion and contraction of the groundwater source region caused by coupled crustal dilation and fracture mineralisation associated with the stress build-up before earthquakes. We conclude by proposing how our approach could be implemented elsewhere.

Hydrochemical changes before and after earthquakes have been reported for over 50years. However, few reports provide sufficient data for an association to be verified statistically. Also, no mechanism has been proposed to explain why hydrochemical changes are observed far from earthquake foci where associated strains are small (<10(-8)). Here we address these challenges based on time series of multiple hydrochemical parameters from two sites in northern Iceland. We report hydrochemical changes before and after M >5 earthquakes in 2002, 2012, and 2013. The longevity of the time series (10 and 16years) permits statistical verification of coupling between hydrochemical changes and earthquakes. We used a Student t test to find significant hydrochemical changes and a binomial test to confirm association with earthquakes. Probable association was confirmed for preseismic changes based on five parameters (Na, Si, K, O-18, and H-2) and postseismic changes based on eight parameters (Ca, Na, Si, Cl, F, SO4, O-18, and H-2). Using concentration ratios and stable isotope values, we showed that (1) gradual preseismic changes were caused by source mixing, which resulted in a shift from equilibrium and triggered water-rock interaction; (2) postseismic changes were caused by rapid source mixing; and (3) longer-term hydrochemical changes were caused by source mixing and mineral growth. Because hydrochemical changes occur at small earthquake-related strains, we attribute source mixing and water-rock interaction to microscale fracturing. Because fracture density and size scale inversely, we infer that mixing of nearby sources and water-rock interaction are feasible responses to small earthquake-related strains. Plain Language Summary Changes in groundwater chemistry before and after earthquakes have been reported for over 50years. However, few studies have been able to prove that the earthquakes caused these changes. Also, no study has explained why these changes are often reported far from where the earthquake occurred. Here we address these challenges based on measurements of groundwater chemistry made at two sites in northern Iceland over time periods of 10 and 16years. We used statistical methods to prove that the earthquakes caused changes of ground water chemistry both before and after the earthquakes. We showed that changes of groundwater chemistry before earthquakes were caused by slow mixing between different groundwaters, which triggered reactions with the wall rock that changed groundwater chemistry, and that changes of groundwater chemistry after earthquakes were causes by rapid mixing between different groundwaters. That these changes were detected far from where the earthquakes occurred suggests that cracking of the wall rock at a very small scale was all that was needed for mixing of different groundwaters and reactions with the wall rock to occur.

Chemical analysis of groundwater samples collected from a borehole at Hafralaekur, northern Iceland, from October 2008 to June 2015 revealed (1) a long-term decrease in concentration of Si and Na and (2) an abrupt increase in concentration of Na before each of two consecutive M 5 earthquakes which occurred in 2012 and 2013, both 76km from Hafralaekur. Based on a geochemical (major elements and stable isotopes), petrological, and mineralogical study of drill cuttings taken from an adjacent borehole, we are able to show that (1) the long-term decrease in concentration of Si and Na was caused by constant volume replacement of labradorite by analcime coupled with precipitation of zeolites in vesicles and along fractures and (2) the abrupt increase of Na concentration before the first earthquake records a switchover to nonstoichiometric dissolution of analcime with preferential release of Na into groundwater. We attribute decay of the Na peaks, which followed and coincided with each earthquake to uptake of Na along fractured or porous boundaries between labradorite and analcime crystals. Possible causes of these Na peaks are an increase of reactive surface area caused by fracturing or a shift from chemical equilibrium caused by mixing between groundwater components. Both could have been triggered by preseismic dilation, which was also inferred in a previous study by Skelton et al. (2014). The mechanism behind preseismic dilation so far from the focus of an earthquake remains unknown.

Groundwater chemistry has been observed to change before earthquakes and is proposed as a precursor signal. Such changes include variations in radon count rates(1,2), concentrations of dissolved elements(3-5) and stable isotope ratios(4,5). Changes in seismicwave velocities(6), water levels in boreholes(7), micro-seismicity(8) and shear wave splitting(9) are also thought to precede earthquakes. Precursor activity has been attributed to expansion of rock volume(7,10,11). However, most studies of precursory phenomena lack sufficient data to rule out other explanations unrelated to earthquakes(12). For example, reproducibility of a precursor signal has seldom been shown and few precursors have been evaluated statistically. Here we analyse the stable isotope ratios and dissolved element concentrations of groundwater taken from a borehole in northern Iceland between 2008 and 2013. We find that the chemistry of the groundwater changed four to six months before two greater than magnitude 5 earthquakes that occurred in October 2012 and April 2013. Statistical analyses indicate that the changes in groundwater chemistry were associated with the earthquakes. We suggest that the changes were caused by crustal dilation associated with stress build-up before each earthquake, which caused different groundwater components to mix. Although the changes we detect are specific for the site in Iceland, we infer that similar processes may be active elsewhere, and that groundwater chemistry is a promising target for future studies on the predictability of earthquakes.

Based on hydrochemical monitoring, petrological observations, and geochemical modeling, we identify a mechanism and estimate a time scale for fault healing after an earthquake. Hydrochemical monitoring of groundwater samples from an aquifer, which is at an approximate depth of 1200 m, was conducted over a period of 10 years. Groundwater samples have been taken from a borehole (HU-01) that crosses the Husavik-Flatey Fault (HFF) near Husavik town, northern Iceland. After 10 weeks of sampling, on 16 September 2002, an M 5.8 earthquake occurred on the Grimsey Lineament, which is approximately parallel to the HFF. This earthquake caused rupturing of a hydrological barrier resulting in an influx of groundwater from a second aquifer, which was recorded by 15-20% concentration increases for some cations and anions. This was followed by hydrochemical recovery. Based on petrological observations of tectonically exhumed fault rocks, we conclude that hydrochemical recovery recorded fault healing by precipitation of secondary minerals along fractures. Because hydrochemical recovery accelerated with time, we conclude that the growth rate of these minerals was controlled by reaction rates at mineral-water interfaces. Geochemical modeling confirmed that the secondary minerals which formed along fractures were saturated in the sampled groundwater. Fault healing and therefore hydrochemical recovery was periodically interrupted by refracturing events. Supported by field and petrographic evidence, we conclude that these events were caused by changes of fluid pressure probably coupled with earthquakes. These events became successively smaller as groundwater flux decreased with time. Despite refracturing, hydrochemical recovery reached completion 8-10 years after the earthquake.