The identification of individual past earthquakes and their characterization in time and space, as well as in magnitude, can be approached in many different ways with a large variety of methods and techniques, using a wide spectrum of objects and features. We revise the stratigraphic and geomorphic evidence currently used in the study of paleoseismicity, after more than three decades since the work by Allen (1975), which was arguably the first critical overview in the field of earthquake geology. Natural objects or geomarkers suitable for paleoseismic analyses are essentially preserved in the sediments, and in a broader sense, in the geologic record. Therefore, the study of these features requires the involvement of geoscientists, but very frequently it is a multidisciplinary effort. The constructed environment and heritage, which typically are the focus of archaeoseismology and macroseismology, here are left aside. The geomarkers suitable to paleoseismic assessment can be grouped based on their physical relation to the earthquake's causative fault. If directly associated with the fault surface rupture, these objects are known as direct or on-fault features (primary effects in the Environmental Seismic Intensity [ESI] 2007 scale). Conversely, those indicators not in direct contact with the fault plane are known as indirect or off-fault evidence (secondary effects in the ESI 2007 scale). This second class of evidence can be subdivided into three types or subclasses: type A, which encompasses seismically induced effects, including soft-sediment deformation (soil liquefaction, mud diapirism), mass movements (including slumps), broken (disturbed) speleothems, fallen precarious rocks, shattered basement rocks, and marks of degassing (pockmarks, mud volcanoes); type B, which consists of remobilized and redeposited sediments (turbidites, homogenites, and tsunamites) and transported rock fragments (erratic blocks); and type C, entailing regional markers of uplift or subsidence (such as reef tracts, micro-atolls, terrace risers, river channels, and in some cases progressive unconformities). The first subclass of objects (type A) is generated by seismic shaking. The second subclass (type B) relates either to water bodies set in motion by the earthquake (for the sediments and erratic blocks) or to earthquake shaking; in a general way, they all relate to wave propagation through different materials. The third subclass (type C) is mostly related to the tectonic deformation itself and can range from local (next to the causative fault) to regional scale. The natural exposure of the paleoseismic objects—which necessarily conditions the paleoseismic approach employed—is largely controlled by the geodynamic setting. For instance, oceanic subduction zones are mostly submarine, while collisional settings tend to occur in continental environments. Divergent and wrenching margins may occur anywhere, in any marine, transitional, or continental environment. Despite the fact that most past subduction earthquakes have to be assessed through indirect evidence, paleoseismic analyses of this category of events have made dramatic progress recently, owing to the increasingly catastrophic impact that they have on human society.

The Calabacillas fault is a 40-km-long, down-to-the-east normal fault that trends N-S on the western edge of the Llano de Albuquerque, in western Albuquerque, New Mexico. It is one of several east-dipping normal faults that define the western margin of the Rio Grande rift at the latitude of Albuquerque. In the past 0.5–1 m.y., since the abandonment of the Llano de Albuquerque surface by the Rio Puerco and Rio Grande, vertical displacement on the Calabacillas fault has created a 27-m-high, east-facing fault scarp on the western edge of the llano, equating to a long-term slip rate of 0.027–0.054 mm/yr. Our two trenches were located ~1 km from the south end of the fault, where a 1-km-wide graben has formed east of the main fault scarp. Trenching of the graben across the southern Calabacillas fault was 50% successful. The paleoearthquake event history on the 5.3-m-high antithetic scarp could not be reconstructed in detail because a strong carbonate soil profile had overprinted the entire 3-m-thick colluvial wedge deposit. It appears that numerous submeter displacements created this scarp, but the displacement was partitioned across several faults, so no single free face was higher than 10–20 cm. Free faces so small did not create colluvial wedges, and thus faulting did not trigger the pattern of footwall erosion and hanging-wall deposition needed to identify individual faulting events. On the 27-m-high main fault scarp, a 60-m-long trench straddled a minor slope break that overlies the main strand of the Calabacillas fault. The upper four soils exposed in the trench could be correlated across the main fault and indicated displacements of 10 cm, 30 cm, 55 cm, and 20 cm in the latest four paleoearthquakes. Six infrared-stimulated luminescence (IRSL) dates on eolian sands range from 14 ka at a depth of 0.5 m to 219 ka at a depth of 5 m. Secondary calcium carbonate has accumulated in soils here at a rate of 0.17–0.35 g/k.y. The latest four faulting events are dated at ca. 14 ka, 23 ka, 35 ka, and 55 ka. Thus, the displacement and recurrence times increase with increasing age, yielding relatively consistent slip rates of 0.011–0.028 mm/yr. There is evidence at this trench for a late Pleistocene (14 ka) small faulting/cracking event, similar in displacement and timing to the youngest warping event interpreted for the County Dump fault, which lies ~5 km to the east. The displacements measured in the main scarp trench are even smaller than those inferred on the County Dump fault, despite the length of the Calabacillas fault (40 km) being similar to that of the County Dump fault (35 km). If our trenches had been located farther north toward the center of the Calabacillas fault, the displacements may have been larger. The ages and recurrence intervals of the four events that occurred subsequent to 55 ka are similar to those seen at the County Dump site. The youngest event on the Calabacillas fault had only 5–10 cm of throw, which is considerably smaller than the 20–55 cm throws of the three previous events. This situation parallels the County Dump chronology, where the youngest warping event was abnormally small compared to earlier events.

The Hubbell Spring fault system lies near the eastern margin of the Albuquerque–Belen Basin in the central Rio Grande rift, and it is one of the most active normal faults in the region. Recent mapping and geophysical studies indicate that fault geometry is more complex and longer than previously thought, with several significant, subparallel, anastomosing, west-dipping splays that form a broad zone as wide as 18 km and ~74 km long. We conducted a paleoseismic investigation of the previously untrenched central Hubbell Spring fault splay (splay L) at the Carrizo Spring site. Our study included mapping, trenching, drilling, and luminescence analyses. We found structural, stratigraphic, and pedologic evidence for the occurrence of at least four, possibly five, large earthquakes that occurred since deposition of piedmont deposits on the Llano de Manzano surface ca. 83.6 ± 6.0 ka. All of these events included warping across a broad deformation zone, whereas the three largest events also included discrete slip across four fault zones. Behavior appears noncharacteristic (i.e., highly variable slip per event), with preferred vertical displacements ranging from 0.4 to 3.7 m. The total down-to-the-west throw of piedmont deposits is 7.3 ± 1.0 m. Luminescence ages indicate that the timing of the four largest surface-deforming events on fault splay L overlaps with the timing of the four youngest faulting events from previous studies of the western Hubbell Spring fault splay (or splay J), suggesting simultaneous rupture of fault splays J and L. Displacement data and correlation of buried soils on event horizons between sites also support simultaneous rupture; however, timing constraints are on the order of thousands of years, and so triggering of events between splays cannot be precluded. The smallest warping event on fault splay L, event Y(?), does not appear to correlate to any events on splay J, suggesting that independent rupture of fault splay L also occasionally occurs. Assuming simultaneous rupture of splay J and L, the average recurrence interval over the past three complete seismic cycles is 19 (+5/−4) k.y., consistent with recurrence intervals estimated for individual cycles, which are 17 k.y., 27 k.y., and 14 k.y. We estimate an average vertical slip rate for the past four complete seismic cycles on splays J and L of ~0.2 mm/yr. In comparison, recent average late Quaternary slip rate estimates for the entire Hubbell Spring fault system range between 0.2 and 1.0 mm/yr, based on topographic profiles transecting all the splays. Slip rates for individual complete seismic cycles for splays L and J vary through time by an order of magnitude, ranging from 0.044 mm/yr to 0.46 mm/yr. This is not due to temporal clustering of earthquakes but instead is primarily due to large variations in slip per event, a finding that may have significant implications for seismic hazards elsewhere in the Rio Grande rift. Additional investigations are needed to determine the paleoseismic behavior of the many other splays of the Hubbell Spring fault system and better characterize this complex fault system for seismic hazard evaluations in the Albuquerque region.

The Pereira-Armenia region, located west of the Colombian Central Cordillera, is crosscut by the Romeral fault system, which consists of an active north-south–trending, left-lateral, strike-slip fault system with a secondary thrust component in the Eje Cafetero zone (4°N–5°N). The terrain where the Liceo Taller San Miguel high school sits—9 km south of Pereira—is draped with an ~2-m-thick layer of volcanic ash younger than 30 k.y. in age. This locality has been affected by both N40°E- and E-W–trending faults that correspond to thrust faults or folds and normal right-lateral, strike-slip faults, respectively, in the tectonic model for the zone. Two kinds of strong field evidence for the E-W faults were found at a site named Canchas: (1) the 50°N tilt of the late Quaternary interbedded sequence of volcanic ash and three paleosols, and (2) a vertical fault throw of ~1.70 m affecting the sequence (layers). A normal vertical throw of ~0.65 m at Parqueadero stands as a proof of the activity of the N40°E-trending faults. This latter faulting does not correspond with the stress tensor proposed for this region, and thus this deformation could be interpreted as being a consequence of flexural slip induced by a NE-SW–striking blind thrust, where reverse faulting along bedding at depth is seen as normal faulting at the surface. Measured offsets could have generated seismic events of at least Mw 6.6 for the NE-trending fault that affected the paleosols and volcanic ash sequence at 13,150 ± 310 14 C yr B.P., and a seismic event of Mw 6.9 for the E-W–trending fault that affected the paleosols and volcanic ash sequence at 19,710 ± 830 14 C yr B.P. These two recently identified faults are now named the Tribunas (NE-SW) and the Cestillal (E-W) faults. Up to now, the fault and its seismogenic potential determinations in this region have been based solely on morphologic evidence. The maximum seismic magnitude estimated for this region ranged from Mw 6.2 to Mw 6.6 for seismic sources 35 km away from the site. Seismic magnitudes like the one calculated in this work (Mw 6.9) were previously estimated only for source-site distances greater than 50 km. This work provides field evidence that leads to a better understanding of the seismic activity of this region in the last 30 k.y. and confirms the occurrence of local Mw >6.5 seismic events in this region. Although volcanic ash drapes and eventually hides the geomorphic evidence of active deformation, it turns out to be a perfect chronometer of a fault’s activity whenever the deformation is revealed, as in this case. After the Armenia event of 1999, it is imperative to examine the seismic hazard assessments of this region in terms of local crustal seismicity.

Morphotectonic and paleoseismic studies carried out in the surrounds of Tuluá (4°N, 76°W) provide strong supporting evidence for ongoing E-W compression in the Cauca Valley, Colombia, during the late Pleistocene and Holocene. This local tectonic regime is kinematically and mechanically connected with the ENE-striking, right-lateral, strike-slip Ibagué fault system, which crosscuts and offsets the Central Cordillera of Colombia. Morphologic, stratigraphic, kinematic, and chronologic evidence obtained on flexural scarps, which are currently shaping the foothills of the Central Cordillera, attests to the recent activity of a compressional fault system. This includes both hinterland-propagating back-thrust faults and foreland-verging thrust faults that reutilize a fold-and-thrust belt, previously considered to be active only during Tertiary times. Kinematic measurements on the back-thrust faults further support an ongoing E-W–oriented maximum horizontal stress at the latitude of Tuluá. In terms of seismic hazard for this region, these investigations provide evidence for Ms ≥7 earthquakes with recurrence in the order of 6 k.y. on this frontal thrust system. In addition, should the A.D. 1766 earthquake have not taken place on these active thrust faults, the probability of occurrence of a forthcoming event with such characteristics would be high.

Geomorphic analysis of the ~30-km-long Lake Edgar fault scarp in southwestern Tasmania suggests that three large surface-rupturing events with vertical displacements of 2.4 m to 3.1 m have occurred in late Quaternary time. Optically stimulated luminescence (OSL) age estimates from a sequence of three periglacial fluvial terraces associated with faulting constrain these events to ca. 18 ka, ca. 28 ka, and ca. 48–61 ka. A similar amount of vertical displacement during each faulting event suggests that surface-breaking earthquakes on this fault are characteristically of magnitude M W 6.8–7.0. Estimates for the average slip rate calculated over two complete seismic cycles range from 0.11 to 0.24 mm/yr, which is large for a stable continental region fault. This sequence represents the first recurrence data for surface-rupturing earthquakes on an eastern Australian Quaternary fault, and one of only a few for the entire Australian continent.

After the 9 July 1997 Ms 6.8 Cariaco earthquake in northeastern Venezuela, we undertook a multiple paleoseismic trench assessment on the newly ruptured portion of the dextral El Pilar fault. The surface rupture of that earthquake extended for 37 km from the seashore village of Villa Frontado to Río Casanay, along the onshore El Pilar fault section that runs between the gulfs of Cariaco and Paria (State of Sucre). This investigation intends to shed additional light on the past seismic history of that fault. For this, three backhoe-dug trenches were excavated between the towns of Cariaco and Río Casanay in early 1998, at the localities of Las Manoas, Carrizal de La Cruz, and Guarapiche. This effort was complemented by the evaluation of an outcrop already in existence in Terranova (7 km west of Cariaco). The three trench sites exhibit very different sedimentary settings. The Las Manoas site is an active papaya-cultivated sag pond. The Carrizal de La Cruz site is an active alluvial terrace, slightly down-faulted with a scarp facing against runoff, thus acting as a sort of shutter ridge. The third trench was cut at the foot of the northern slope of a pop-up structure, forming at a restraining overlap (as attested by 1997 Cariaco earthquake rupture mapping). All trenches were ~20 m long and 3–4 m deep. The main outcomes of this assessment are: (1) Over 10 earthquakes are common to the three trenches over a period of 5.6 k.y.; (2) the latest five events, including the 1997 event, clearly seem to recur roughly every 300 yr; (3) a minimum of 15 or 16 events can be deduced from colluvial deposits interfingered with the fault pond sequence at Las Manoas trench, averaging a repeat period of 350 yr over the longer 5.6 k.y. time span; (4) the predecessor of the 1997 event was the 1684 earthquake, for which chronicles could not provide a more precise determination previous to this study; and (5) the 1974 event might be present in the two easternmost trenches.

The late Quaternary sedimentary fills of several lakes of the northwestern Alps are revealed to be possible paleoseismological “archives” in a moderately active seismotectonic region. The strongest historically reported events can be correlated with specific layers having textures that result from different processes, such as: (1) mass failures of subaqueous slope deposits (especially delta foresets) evolving into hyperpycnal currents influenced by seiche effects and/or multiple reflections on lake basin slopes; (2) in situ liquefaction and flowage; and (3) microfracturing. Based on identification of the sedimentary signature of a well-documented historical earthquake, the paleoseismic interpretation can be extrapolated back to 16,000 yr B.P. with reconstruction of time series and textural identification of slope failure–related turbidites (the most frequent earthquake signature). The obtained time series are compatible with historical seismicity in terms of recurrence interval. The sedimentological approach developed for moderately seismotectonic environments appears to be valid for other large lake basins undergoing high-magnitude earthquakes.

Two earthquakes are recorded in lake sediments of a former rock-avalanche–dammed lake at the outlet of the Calchaquíes valleys, Argentina. The lake existed between 13,830 ± 790 and 4810 ± 500 a, as indicated by 10 Be exposure ages of the landslide deposits that impounded that lake and caused the dam erosion. Two reverse faults, with buckle folds in the footwall and slump folds in the hanging wall, indicate that two earthquakes took place while the lake sediments were water saturated, i.e., during the lake phase. Two folds only a few meters apart occur within the same lake sediment sequence over a distance of 1.3 km on two layers. Within the same two layers, there are mixed zones of convolute bedding extending several hundred meters toward the center of the former lake, which are interpreted to be seismites. These disturbed zones occur also in other subbasins of the former lake that were not affected by faulting and folding. One seismite horizon was AMS (accelerator mass spectrometry) 14 C dated to 7500 ± 70 cal yr B.P. by organic material. This age agrees with the 10 Be surface exposure age of 7820 ± 830 a for a cluster of four landslides 40 km NNW of the outlet of this lake, suggesting that a strong earthquake occurred at this time.

We present a new method that allows paleomagnitude and paleorupture length to be estimated quantitatively given a measurement of earthquake rupture displacement at a point along a fault. Rupture displacement typically varies along a rupture profile such that a point paleoseismic displacement measurement constrains the average displacement only to within a factor of three or so. We used previously published results summarizing rupture variability and then applied a graphical method of identifying the relative likelihoods among a suite of magnitudes, one of which must have caused the measured displacement. Results were developed for displacement observations from 1 to 6 m using a magnitude range of 6.0 ≤ M ≤ 8.0. Probabilities of rupture lengths for a given displacement were developed at the same time. Although smaller earthquakes can cause ground rupture, we show that they would not strongly influence likelihoods for 1 m and larger observed displacements. Displacements less than 1 m are also of potential interest but will require extension of the method to include the declining probability that smaller magnitudes produce ground rupture. We also consider application of length distributions inferred from a displacement measurement to correlation of rupture evidence between sites. Dating evidence alone, even when excellent, does not provide a physical basis to relate rupture at one site to rupture at another. Ruptures, however, have an expected length, and thus do provide a physical basis for correlation. We present probability of correlation curves for given rupture lengths, which may be combined with probabilities of rupture length to obtain a probability of correlation given a point displacement. Applications for quantitative probabilities of magnitude and length given a paleoseismic displacement measurement include probabilistic seismic hazard analyses, where probabilities of magnitude and length must be assigned to branches in the analysis.

Abstract A well-preserved fault scarp resulting from recurrent Quarternary normal faulting occurs at the western edge of the northern Sangre de Cristo Mountains of south-central Colorado, at the mouth of west-draining Major Creek (Fig. 1). The site is approximately 56 mi (90 km) north-northeast of Alamosa, Colorado, and may be approached from Colorado 17 between Alamosa and Poncha Springs, Colorado. Turn east off of Colorado 17, 50 mi (80 km) north of Alamosa, opposite the junction with U.S. 285, onto a dirt road that leads due east for 6 mi (10 km) across the valley floor toward Valley View Hot Springs. Instead of turning off to the hot springs, bear right and follow the road as it turns south to parallel the range front. The road will continue south past the mouth of Garner Creek [1 mi (1.7 km) south of the hot springs turnoff] to the upper part of the Major Creek alluvial fan [2 mi (3.2 km) south of the hot springs turnoff]. Approximately 0.25 mi (400 m) north of the crossing of Major Creek (Fig. 2) turn onto a dirt driveway leading due east to the head of the fan where the fault scarps occur. Entry through the locked gate will require permission from the landowner on the north side of the fanhead, Dr. Ben Eismann, Dept. of Surgery, University of Colorado Health Science Center, Denver, CO 80262. The fault scarps on the south side of the road (including the 1980 trench site, Profile 35 on Fig.