The Gutenberg-Richter law is a fundamental empirical law in seismology describing earthquake frequency-magnitude distributions, with one of its key parameters, the so-called b-value, quantifying the relative frequency of small versus large events. While the b-value is commonly interpreted as reflecting crustal heterogeneity and regional stress conditions, its underlying physical origin remains poorly understood, particularly the relative roles of geometrical versus mechanical controls. Here, we develop analytical and numerical models to elucidate the origin of the b-value in three-dimensional fault networks subject to mainshock-aftershock sequences. We demonstrate that the b-value emerges from the power-law scaling of fault rupture area together with the scaling of slip magnitude. Our results reveal a two-branch frequency-magnitude distribution, with the regime transition governed by fault criticality and fracture energy dissipation, while the transition magnitude reflects the finite population of faults triggered during the sequence. Our findings provide a physically grounded interpretation of earthquake b-values, establishing a link between fault mechanics and earthquake statistics.

The Swedish Transport Administration (STA) currently monitors the railway between Kiruna and the Swedish-Norwegian border with Distributed Acoustic Sensing (DAS), a distance of approximately 130 km. In collaboration with STA and Luleå University of Technology, the Swedish National Seismic Network (SNSN) has established data transmission on a request basis from the interrogator. As the railway crosses the Pärvie fault, the largest known, and still very active, glacially triggered fault, we hope to significantly improve detection and analysis of small earthquakes on that section of the fault. In this presentation we will show how we define low noise sections of the cable, using local and teleseismic events, and then use these as individual seismic stations. Over the 130 km, as the railway winds its way across the mountains, the cable generally runs in directions from N-S via NW-SE to W-E, providing many possible incidence directions. We discuss the technicalities of the data sharing, the existing metadata problems, how the DAS data is analyzed and incorporated into the routine processing at SNSN.

Climate change has led to more frequent and widespread droughts motivating robust monitoring of groundwater resources. Ambient seismic noise interferometry allows to derive relative seismic velocity changes (Δv/v) over time and space in the subsurface. Δv/v correlates well with groundwater fluctuations. Traditional datasets used to monitor groundwater changes, such as groundwater level data from wells and GRACE satellite gravimetric data, are either spatially sparse or limited in spatial resolution. Seismic velocity changes offer an additional, high-resolution measure of groundwater changes. Here, we aim to enhance groundwater monitoring in central Scandinavia, which experienced severe droughts in 2018 and 2022, and increase understanding on how groundwater levels decrease during droughts and recharge during periods of higher precipitations. One challenge of the ambient seismic noise interferometry method is the assumption of uniform noise sources, which rarely applies to seismic stations in Norway and Sweden. In this study, we test several denoising and spatial inversion robustness methods, including denoising autoencoders, convolutional neural networks, and variational inference. Through the integration of seismic and hydrological data, complex signal enhancement, and probabilistic inversion, we develop a robust method for monitoring groundwater in areas with heterogeneous station spacing and non-uniform noise sources.

The Hälsingland earthquake cluster, on the east coast of central Sweden, represents a puzzling case of intraplate seismicity in a tectonically stable continental region. The cluster measures approximately 100 km in length and extends in a near-linear trend from inland in the southwest into the Baltic Sea in the northeast, oriented approximately 35 degrees to the coastline. Unlike many of the earthquake clusters that occur in Sweden, the cause of the Hälsingland seismicity is not well understood, as it has not been possible to associate the cluster with any distinct geological feature, such as old deformation zones or a younger glacially triggered fault. Between September 2021 and September 2025, a temporary network consisting of thirteen broadband seismic stations was deployed in the Hälsingland region in an effort to establish better understanding of the drivers behind the Hälsingland seismicity. During this period, 873 earthquakes were detected and manually analyzed in the region, with local magnitudes ranging from -1.0 to 2.3. Using travel-time data from local quarry blasting, we derived a new, regional seismic velocity model and relocated all the earthquakes in the new model. The earthquake depths range from near-surface down to 39 km, with approximately 80% occurring at depths between 5 and 20 km. As part of this project, a previously unknown glacially triggered fault (GTF) system, the Mörtsjö fault system, was identified in the Hälsingland region, approximately 25 km north of the Bollnäs fault, the southernmost confirmed GTF in Sweden. Both the Mörtsjö and Bollnäs GTFs are small and located outside the most seismically active part of the Hälsingland region. However, relative earthquake relocations reveal multiple events which may be generated by movement on the faults. Waveform cross-correlation analysis shows moderate correlation between most earthquake pairs in the Hälsingland cluster but also identifies multiple families of closely spaced, highly correlating earthquakes, including a single family consisting of more than 30 events. The spread of the earthquake focal mechanisms does not clearly indicate a dominant fault orientation. While strike-slip motion dominates, multiple examples of both reverse and normal motion also occur, often in close proximity to each other. Inverting the focal mechanisms for the earthquake-generating stress field indicates a strike-slip stress state with a NW-SE direction of maximum horizontal stress. The inversion also suggests mostly E-W striking fault planes, suggesting that the faults rupturing in the Hälsingland earthquakes are not oriented in agreement with the general lineament of the cluster. We find that most of the Hälsingland seismicity does not occur on a well defined fault but rather in an active zone which extends to large depth but is only vaguely associated with changes in large scale geological features such as magnetic properties and Moho thickness.

Probabilistic seismic hazard assessment (PSHA) is challenging in cratonic regions such as Sweden, where the characteristics of strong motion (detrimental for structures) are uncertain due to a lack of data. The current approach is therefore to use available data sets of small-to-moderate earthquakes to identify seismogenic areas and to adapt models from more seismically active regions. One issue encountered in this process is estimating the moment magnitude (Mw) of these earthquakes. In fact, evaluation of local magnitude (Ml) is preferred for magnitude

On 26 September 2022 two seismic events near the Danish island of Bornholm in the Baltic Sea were detected. The first event with a magnitude Mw 2.3 occurred at 00:03 UTC 40 km east-southeast of Bornholm. The determined location and the origin time of the event are consistent with data of the pressure decrease on one of the Nord Stream 2 pipelines. Another sequence of events occurred 17 hours later at 17:03 UTC around 60 km north-east of Bornholm with a maximum magnitude of Mw 2.7. It consists of three closely successive, but separable, single events. Using relative localisation methods and the gas pressure inside the pipeline recorded at the landing site in Germany, we can assign the epicentres of the three events to the locations of the leaks in the pipelines of Nord Stream 1 and 2. Based on comparable events in the region, which include both tectonic earthquakes and explosions, the explosive character of the investigated Nord Stream events can be verified. Infrasound signals associated with the destruction of the Nord Stream pipelines were recorded at two stations (I26DE in the Bavarian Forest and IKUDE near Kühlungsborn) in Germany. Particularly after the event sequence at 17:03 UTC, distinctive signals were registered whose characteristics indicate an explosive event with subsequent gas leakage at the surface. Our modelling of the sources shows that the measured seismic signals can sufficiently be explained by the instantaneous gas release. Synthetic seismograms for such a source and a subsurface model adapted for the study area show high consistency with the measured signals. Based on the released energy and the characteristics of the recorded waveforms, we conclude that the impulsive gas release from the burst gas pipes constitutes the dominant part of the signal source. The model places an upper limit of approximately 50 kg TNT equivalent on the yield of the chemical explosive component of the events, but we note that smaller yields may also be consistent with the data. We also carried out an analysis of the seismic signals of the event on the Balticconnector pipeline between Finland and Estonia on 8 October 2023 and found that again the instantaneous gas release can sufficiently explain the observed data. This supports a possible mechanical cause of the damage.

Reliable, cost efficient, and continuous observations of nearshore hydrodynamics are often required for the design and maintenance of coastal structures as well as to understand coastal change. In the last decades, advances in digitization and computational efficiency for signal processing have led to an increased use of marine radars as a tool for hydrographic applications, such as the retrieval of bathymetry, surface currents, winds and sea state. Many marine radar products are based on a three dimensional fast Fourier transformation (3D-FFT) of the image sequences obtained from a scanning radar. These methods have been extensively validated in deep to intermediate water depths. In the nearshore, and increasingly shallow waters, validation studies are rare and the available studies mainly focus on the retrieval of bathymetry. Validated radar measurements of spatially varying wave and current fields are not yet available. The present study is thus focussed on the assessment of the limitations of radar hydrography in a nearshore environment.

The Fennoscandian earthquake catalogue (FENCAT) assembles data on the natural seismicity in Fennoscandia, Northern Europe. We present an updated and standardized version of the catalogue originally published in the early 1990s. New instrumental data are recorded by the seismic networks of Denmark, Estonia, Finland, Norway and Sweden, and analysed by the Geological Survey of Denmark and Greenland, the Geological Survey of Estonia, the University of Helsinki in Finland, the University of Bergen and the NORSAR research foundation in Norway and Uppsala University in Sweden. The updated catalogue provides the available earthquake parameters in a brief, user-friendly version: origin time, source coordinates, focal depth, macroseismic data (maximum intensity and radius of the area of perceptibility), up to three observed magnitudes, seismic moment estimate and a standardized moment-related magnitude, mW(HEL), for each event. The standardized magnitude is defined in this paper and its relation to other magnitude scales is provided. Suspected non-earthquakes (e.g. frost events, explosions, human-induced events) have been removed. The standardized event magnitudes range from mW(HEL) -1.0 to 6.2. To enable the usage of earthquake data in a large variety of seismological, geological and earthquake engineering investigations, the data are not truncated at the low-magnitude end.The updated catalogue, FENCAT (2021), contains about 23 000 earthquakes for the period 1467-2021 in an area bounded by 54-75 degrees N latitudes and 0-45 degrees E longitudes. The completeness and quality of the earthquake solutions is best within the areal coverage of the above-mentioned networks.

The Swedish National Seismic Network (SNSN) currently operates 67 permanent and 13 temporary broadband seismic stations. All stations transmit continuous realtime data to the data centre in Uppsala, and data streams of about 40 stations are automatically forwarded to subscribing institutes in the neighboring countries and to ORFEUS. In addition to the SNSN stations we receive realtime data from about 120 stations located in Norway, Finland, Denmark, Germany, Poland, the Baltic States, and Russia. SNSN processes the waveform data of this virtual network of about 200 stations using the SeisComp and Earthworm systems in parallel. Both systems are set up to be very sensitive in order to detect as small events as possible, which also increases the probability of generating spurious events. In order to screen out spurious events we generate a common bulletin which contains events that have been located by both systems independently. Our common bulletin is very reliable (no spurious events during the last 1.5 years), captures events down to about ML = 1 and contains almost all events with ML > 1.5 in Fennoscandia.All events in the common bulletin are automatically classified by an artificial neural network as earthquakes, blasts or mining-induced events. The classifier has been developed in the framework of a PhD project, and was implemented into the SNSN processing queue during 2023 (Eggertsson et al, "Earthquake or Blast? Classification of Local-Distance Seismic Events in Sweden using Fully-Connected Neural Networks", accepted GJI 2024). It has been thoroughly tested, and, comparing the automatic classification with analyst-reviewed classification, we found a 97% matchSince December 2023, SNSN provides the automatic common bulletin as a simple webpage https://www.snsn.se/combullUTC/ - mainly for the general public and for quick reference. For the seismological community, SNSN has set up an automatic real-time forwarding of complete event parameters for all events with ML >= 2 to the European-Mediterranean Seismological Centre.

We advance the use of convolutional neural networks (CNNs) for discriminating low-yield seismic events recorded at local distances by evaluating a CNN approach based on time–frequency representations (scalograms) of seismic records from earthquakes, mine blasts, and mining-related seismic events in the Kiruna mining region of northern Sweden to (1) determine if the CNN approach can outperform the P/S amplitude ratio method in classifying these source types, and (2) examine the regional transportability of a CNN model trained on data from the United States. An accuracy of 90% or greater was obtained for the CNN approach for binary source classification between the three source types (earthquakes, mine blasts, and mining-related events), an accuracy level not achieved by the P/S amplitude ratio method, illustrating superior performance of the CNN approach over the amplitude ratio approach. The CNN model trained on explosions and earthquakes in United States yields poor binary classification performance (accuracy < 90%) when applied to earthquakes and mine blasts in the Kiruna mining region, suggesting limited transportability of the U.S.-trained model. However, the poor performance may arise from differences in the blasting style between the two data sets (single-fired borehole explosions in the United States versus ripple-fired blasts into a mine shaft at the Kiruna mine) and source depths (near surface in United States vs. 800–900 m depth in the Kiruna mine), leaving open the question of whether transportability is more limited by differences in local geologic structure or in explosion source processes.

A marine X-band radar system, developed by Helmholtz-Zentrum Hereon (Hereon) was deployed within view of the nearshore at the US Army Engineer Research and Development Center, Field Research Facility (FRF), in Duck, North Carolina, from October 2021 to August 2022. The radar deployment was a collaboration among researchers at the FRF, Hereon, and the University of Miami and was initiated as part of the During Nearshore Event Experiment (DUNEX), a large multi-institutional field experiment funded by the US Coastal Research Program. The Hereon radar successfully collected data during the main DUNEX field campaign (approximately October 2021) and continued to collect nearly continuously until August 2022. To facilitate use of Hereon radar data, this document describes the deployment, provides background and context, and presents metadata. Within, we describe in detail the Hereon radar system, the locations of two different installations, the time periods covered, sampling modes, environmental conditions and notable events, example data products, and potential pathways for future use of the data.

ABSTRACT We investigate the utility of the P/S amplitude discriminant for small seismic events recorded at local distances on surface seismic networks using (1) mining-related events from within the Kloof gold mine in South Africa; and (2) mining-related events and earthquakes within and adjacent to the Kiruna iron ore mine in northern Sweden. For the Kloof mine, seventy-five source mechanisms characterized by moment tensor solutions obtained using high-frequency in-mine seismic data are used to evaluate three mine-related source types, isotropic (crush), compensated linear vector dipole (crush-slip), and double-couple (DC; pure slip). For the Kiruna mine region, 270 events are used to evaluate earthquake sources, chemical explosions, and mine-related seismic events (primarily isotropic). For the Kloof mine events, we find that average P/S amplitude ratios measured in the 2–6 Hz frequency band discriminate between isotropic and DC events, and if only pure-slip events with a DC component of >60% are considered, the effective frequency band can be extended from 2 to 8 Hz. For the Kiruna region events, P/S amplitude ratios effectively discriminate earthquakes from chemical explosions in the 4–6 Hz and 10–28 Hz frequency bands. Our findings further show that average P/S amplitude ratios for mine-related events and earthquakes separate at frequencies of 10 Hz and higher. A comparison of amplitude ratios for crush and pure-slip events located within a depth range of 1 km in the Kloof mine, and a comparison of amplitude ratios of shallow (<10 km depth) and deep-focus (>20 km depth) earthquakes in the Kiruna region, indicate that the P/S amplitude discriminant is not influenced significantly by source depth. These findings thus suggest that the P/S amplitude discriminant, originally developed for larger events recorded at regional and teleseismic distances, can be extended to smaller events recorded at local distances.

Sweden is a low-seismicity, stable continental region where seismic hazard assessment is non-trivial. Diffuse seismicity, low seismicity rate, few large magnitude earthquakes and little strong motion data makes it difficult to estimate recurrence parameters and determine appropriate attenuation relationships. Here we present a probabilistic seismic hazard assessment of Sweden based on a recent earthquake catalogue which includes earthquakes with magnitudes ranging from -1.4 to 5.9. The large number of events enables recurrence parameters to be calculated also for smaller source areas, in contrast to previous studies, and with less uncertainty. We use recent ground motion models developed specifically for stable continental regions, including Fennoscandia, and calculate hazard using the OpenQuake engine. The results are presented in the form of mean peak ground acceleration (PGA) maps at 475 and 2500 year return periods, hazard curves for four seismically active areas in Sweden and deaggregation for the area of highest hazard. We find the highest hazard in the northernmost part of the country, in the post-glacial fault province. This is in contrast to previous studies, which have not considered the high seismic activity on the post-glacial faults. We find relatively high hazard along the northeast coast and in southwestern Sweden, whereas the southeast and the mountain region to the northwest have low hazard. For a 475 year return period we estimate the highest PGAs to be 0.04 0.05g, in the far north, and for a 2500 year return period it is 0.1-0.15g in the same area. Significant uncertainties remain to be addressed with regards to the intraplate seismicity in Sweden and surroundings, such as the homogenization of magnitude scales, the occurrence of large events in areas with little prior seismicity and the uncertainties surrounding the potential for very large earthquakes on the post-glacial faults in northern Fennoscandia.

SUMMARY Distinguishing between different types of seismic events is a task typically performed manually by expert analysts and can thus be both time and resource expensive. Analysts at the Swedish National Seismic Network (SNSN) use four different event types in the routine analysis: natural (tectonic) earthquakes, blasts (e.g. from mines, quarries and construction) and two different types of mining-induced events associated with large, underground mines. In order to aid manual event classification and to classify automatic event definitions, we have used fully connected neural networks to implement classification models which distinguish between the four event types. For each event, we bandpass filter the waveform data in 20 narrow-frequency bands before dividing each component into four non-overlapping time windows, corresponding to the P phase, P coda, S phase and S coda. In each window, we compute the root-mean-square amplitude and the resulting array of amplitudes is then used as the neural network inputs. We compare results achieved using a station-specific approach, where individual models are trained for each seismic station, to a regional approach where a single model is trained for the whole study area. An extension of the models, which distinguishes spurious phase associations from real seismic events in automatic event definitions, has also been implemented. When applying our models to evaluation data distinguishing between earthquakes and blasts, we achieve an accuracy of about 98 per cent for automatic events and 99 per cent for manually analysed events. In areas located close to large underground mines, where all four event types are observed, the corresponding accuracy is about 90 and 96 per cent, respectively. The accuracy when distinguishing spurious events from real seismic events is about 95 per cent. We find that the majority of erroneous classifications can be traced back to uncertainties in automatic phase picks and location estimates. The models are already in use at the SNSN, both for preliminary type predictions of automatic events and for reviewing manually analysed events.

The Reykjanes Peninsula (RP) hosts several volcanic lineaments that have been periodically active over the last 4000 years. Since 2021, following a ca. 800-year quiescence, eight eruptions have occurred on the RP, with more expected in the future. To better understand the origins of this renewed volcanism and help forecast future eruptions, we examine (i) if the ongoing volcanism is fed from a single or multiple magma storage zone(s) or from several smaller reservoirs and; (ii) where the zone(s) are located (i.e. mantle or lower or upper crustal depths). Using major and trace element geochemistry, oxygen isotopes, and seismic tomography we rule out a single, RP-scale, deep-seated magma storage zone. Instead we propose the presence of a ca. 10-km-wide region of crustal-level (9-12 km) magma accumulation beneath the Fagradalsfjall volcanic lineament that fed both the 2021-23 eruptions of the Fagradalsfjall Fires and the 2023-24 eruptions of the Sundhn & uacute;kur Fires.

Climate change is increasingly predisposing polar regions to large landslides. Tsunamigenic landslides have occurred recently in Greenland ( Kalaallit Nunaat ), but none have been reported from the eastern fjords. In September 2023, we detected the start of a 9-day-long, global 10.88-millihertz (92-second) monochromatic very-long-period (VLP) seismic signal, originating from East Greenland. In this study, we demonstrate how this event started with a glacial thinning–induced rock-ice avalanche of 25 × 10 6 cubic meters plunging into Dickson Fjord, triggering a 200-meter-high tsunami. Simulations show that the tsunami stabilized into a 7-meter-high long-duration seiche with a frequency (11.45 millihertz) and slow amplitude decay that were nearly identical to the seismic signal. An oscillating, fjord-transverse single force with a maximum amplitude of 5 × 10 11 newtons reproduced the seismic amplitudes and their radiation pattern relative to the fjord, demonstrating how a seiche directly caused the 9-day-long seismic signal. Our findings highlight how climate change is causing cascading, hazardous feedbacks between the cryosphere, hydrosphere, and lithosphere.

We investigate changes in the global reported fatalities from earthquake disasters in the global Emergency Events Database (EM-DAT). Drawing parallels with the Gutenberg -Richter frequency -magnitude analysis, in terms of disaster frequency versus the number of casualties, we see a significant overlap of the curves and improving levels of completeness over six 20-year periods. This implies a decrease in underreporting with time. We find that the apparent strong upward trend in the number of (repoerted) earthquake disasters in EM-DAT is caused by a gradually improved reporting primarily of events killing fewer than 10 people. Our findings imply that the true (reported and unreported) number of earthquake disasters, according to the EMDAT definition, has been surprisingly constant over, at least, the last 100 yr. We also show that the average annual number of people killed in earthquake disasters is relatively unaffected by spurious trends in reporting and has remained remarkably constant despite population increase. This implies an impressive reduced mortality risk roughly proportional to population increase since 1900. However, there is no indication in the data that the risk of future mega-disasters is negligible, and further major reductions in vulnerability should be actively pursued.

In recent years the Swedish National Seismic Network (SNSN) made an increased effort to modernize station and communication equipment, and thereby has significantly improved continuous real-time data availability and data quality. Currently, the SNSN is operating 67 permanent broadband seismic stations evenly distributed in the South, along the Eastern shore and the North of Sweden. In addition, a temporay network of 13 stations was deployed in 2021 for a 3-year period to monitor small earthquakes associated with a linear cluster of events at the Western cost of the Gulf of Bothnia. SNSN transmits continuous real-time data to networks in the neighboring countries (Norway, Denmark, Finland, Germany) and in turn receives and processes data from about 120 stations abroad.Compared to many other countries, Sweden has a relatively low seismicity. This makes it all the more important to focus on small seismic events in order to map crustal structures and processes and to provide a data basis for reasonable long-term seismic hazard assessments. Turning to small events means to deal with many events and most of them being man-made seismic events (blasts related to quarries, underground mines, road/tunnel constructions, etc). Within the last 22 years SNSN has recorded and analyzed about 170,000 seismic events out of which only 11,000 (~6.5%) were classified as natural events. Automatic event processing and event type classification are of the essence in order to cope with the amount of data and to decrease the workload of the analysts.SNSN is running four different and independent automatic processing routines in parallel: SeisComp (SC), Earthworm (EW), MSIL and a migration stack algorithm (MS). The main purpose of SC and EW is to detect and locate events in realtime. Both systems are set up to be very sensitive in order to detect as small events as possible, which on the other side also increases the probability to generate spurious events. To counterbalance that we generate a common event catalogue (i.e. events that were located both by SC as well as EW) which turnes out to be very reliable. The common event bulletin captures events as small as about ML1 and contains almost all events ML > 1.5. MSIL and MS are running in offline and delayed mode which allows the backfilling of potential data gaps, before processing. These systems are catching events down to about magnitude ML0. All events of the common bulletin and the MSIL bulletin are subject to an automatic Neural Network event typ classification into earthquakes, blasts and mining-induced events. In a final step all events classified as earthquakes, significant blasts (felt events or events of special interest) or events with unclear cassification are reviewed by SNSN analysts and are being made available on the SNSN web page. In the framework of EPOS-Sweden, SNSN will make available the waveform and metadata data of the permanent network via FDSN-services.

&lt;p&gt;Soon after midnight on 26 September 2022 the Swedish National Seismic Network, using data from Sweden, Denmark and Germany, automatically detected a seismic event in the Baltic southeast of the Danish island of Bornholm. The event was followed 17 hours later by a second, more complex, event northeast of Bornholm. The automatic locations of the events were within 6-9 km of later reported gas leaks in the Nord Stream 1 and 2 pipelines. Using recently developed, machine learning based, classifiers both events were automatically classified as explosions. Subsequent analysis of the second event revealed that it was in fact two blasts, separated by about 7 seconds. As the events occurred in the transition zone between the Fennoscandian Shield and the younger terranes of Denmark and northern Germany, 3D tomographic P- and S-velocity models were developed to improve locations and assess uncertainties, bringing the locations closer to the pipelines. Spectral analysis of the blast data show clear reverberations consistent with underwater explosions and a blast depth of approximately 75 m. The conclusion that the events are underwater blasts are further supported by data on known underwater explosions and a few earthquakes in the area. The magnitude of the first event was estimated at ML 1.9 and the combined second and third event had ML 2.3. Estimating the equivalent yield in the explosions is, however, non-trivial. Comparison to ground truth underwater explosions suggests yields of one to a few hundred kilos of equivalent TNT. The contribution to the seismic energy from suddenly outflowing methane gas is under investigation and results will be included in the presentation.&lt;/p&gt;

&lt;p&gt;Distinguishing small earthquakes from man-made blasts at construction sites, in quarries and in mines is a non-trivial task during automatic event analysis and thus typically requires manual revision. We have developed station-specific classification models capable of both accurately assigning source type to seismic events in Sweden and filtering out spurious events from an automatic event catalogue. Our method divides all three components of the seismic records for each event into four non-overlapping time windows, corresponding to P-phase, P-coda, S-phase and S-coda, and computes the Root-Mean-Square (RMS) amplitude in each window. This process is repeated for a total of twenty narrow frequency bands. The resulting array of amplitudes is passed as inputs to fully connected Artificial Neural Network classifiers which attempt to filter out spurious events before distinguishing between natural earthquakes, industrial blasts and mining-induced events. The distinction includes e.g. distinguishing mining blasts from mining induced events, shallow earthquakes from blasts and differentiating between different types of mining induced events. The classifiers are trained on labelled seismic records dating from 2010 to 2021. They are already in use at the Swedish National Seismic Network where they serve as an aid to the routine manual analysis and as a tool for directly assigning preliminary source type to events in an automatic event catalogue. Initial results are promising and suggest that the method can accurately distinguish between different types of seismic events registered in Sweden and filter out the majority of spurious events.&lt;/p&gt;