**3. The earthquake of July 1930**

The 23 July 1930 earthquake happened at 00:08 GMT with the greatest intensity of X MCS (Mercalli-Cancani-Sieberg scale) and a magnitude MS=6.7, followed by many aftershocks, some of which with intensity of VII MCS. The epicentre of the main shock (figure 2a) was located at 41°05'N and 15°37'E in Irpinia (Freeman, 1930). The earthquake affected a very wide area, 36,000 km2, comprising the regions of Campania, Puglia and Basilicata. The main shock was particularly destructive, resulting in 1425 fatalities, about 10,000 injured and more than 100,000 homeless people, 22 villages destroyed and about 40,000 dwellings damaged (Spadea et al., 1985). The epicentral zone of IMCS=X is elliptical and extends over an area of 180 km2 with the major axis of 34 km parallel to the Apennine trend (WNW-ESE). The area of the greatest effects, primary (surface faulting) and secondary effects (slope movements, ground cracks, hydrological anomalies) (IMCS≥ VIII) is also elliptical, extending over about 6000 km2.

Many foreshocks and aftershocks accompanied the main event. At least two foreshocks preceded it at 23:30 on 22 July and at 00:30. The aftershocks with destructive effects occurred until 1931, also with intensity IMCS>VI (Spadea et al., 1985). Surface wave magnitudes (MS) in the range 6.2-6.7 have been estimated for the 23 July 1930 earthquake. Whereas Westaway (1992) determined a seismic moment of Mo=3.2x1025 dyne-cm, Jimenez et al. (1989), on the basis of seismograms recorded at Jena (Germany), calculated Mo=2x1025 dyne-cm. The fault plane orientation was WNW-ESE (Apennine chain trend), the fault length was 32.6 km and the depth 15 km, estimated on the basis of the equivalent ray of the major isoseismal lines (Gasperini et al., 1999).

Hydrological changes were observed in the whole macroseismic field, mostly in the far field, near the main carbonate aquifers in a widespread karstic environment (Esposito et al., 2009). They include flow increases both in springs and wells, turbid water and drying up of springs, appearance of new springs and variations in the chemical parameters of waters. From the data collected by the IHS (Annales 1925-1940) many anomalies were analysed by evaluating the shape and timing of hydrological changes in 151 sites both in the Tyrrhenian and in Adriatic watersheds (48 wells, 88 stream gauge stations, 15 springs). Spring flow at Madonna del Carmine increased from 10 l/min to 40 l/min after the earthquake, and at Monte della Guardia increased from 5 l/min to 16 l/min at the end of August (Esposito et al., 2009).

At Solfatara, a volcanic crater in Pozzuoli (near Neaples), at a distance of about 100 km from the epicentre, variations in endogenous activity were observed for about 20 days after the earthquake. Majo (1931) reports a temporary decrease in fumarolic gases, diffuse H2S emanation from the soil, strong gas bubbling in a mud pool, and a notable temperature increase in monitored points. The possible influence of seismicity on gas release from depth can be demonstrated on the occasion of a small earthquake felt locally on 12 August 1930 near Pozzuoli. Table 2 reports the temperatures in °C measured in various sites inside the Solfatara crater (a) before 23 July, (b) during the period 28 July – 8 August and (c) during the period 13 – 26 August. In a nearby site, at Stufe di Nerone, a considerable increase in CO2 and temperatures was also observed.


Table 2. Temperatures in °C measured in various sites inside the Solfatara crater.

Another site with fumaroles and mud pools, Ansanto Valley, situated about 20 km from the epicenter, presented an increase in gas emission and mud boiling, together with light flashes (Alfano, 1931). This is a very sensitive site since such phenomena also occurred in this site for other earthquakes in the southern Apennines (Italiano et al., 2000).

some of which with intensity of VII MCS. The epicentre of the main shock (figure 2a) was located at 41°05'N and 15°37'E in Irpinia (Freeman, 1930). The earthquake affected a very wide area, 36,000 km2, comprising the regions of Campania, Puglia and Basilicata. The main shock was particularly destructive, resulting in 1425 fatalities, about 10,000 injured and more than 100,000 homeless people, 22 villages destroyed and about 40,000 dwellings damaged (Spadea et al., 1985). The epicentral zone of IMCS=X is elliptical and extends over an area of 180 km2 with the major axis of 34 km parallel to the Apennine trend (WNW-ESE). The area of the greatest effects, primary (surface faulting) and secondary effects (slope movements, ground cracks,

Many foreshocks and aftershocks accompanied the main event. At least two foreshocks preceded it at 23:30 on 22 July and at 00:30. The aftershocks with destructive effects occurred until 1931, also with intensity IMCS>VI (Spadea et al., 1985). Surface wave magnitudes (MS) in the range 6.2-6.7 have been estimated for the 23 July 1930 earthquake. Whereas Westaway (1992) determined a seismic moment of Mo=3.2x1025 dyne-cm, Jimenez et al. (1989), on the basis of seismograms recorded at Jena (Germany), calculated Mo=2x1025 dyne-cm. The fault plane orientation was WNW-ESE (Apennine chain trend), the fault length was 32.6 km and the depth 15 km, estimated on the basis of the equivalent ray of the major isoseismal lines

Hydrological changes were observed in the whole macroseismic field, mostly in the far field, near the main carbonate aquifers in a widespread karstic environment (Esposito et al., 2009). They include flow increases both in springs and wells, turbid water and drying up of springs, appearance of new springs and variations in the chemical parameters of waters. From the data collected by the IHS (Annales 1925-1940) many anomalies were analysed by evaluating the shape and timing of hydrological changes in 151 sites both in the Tyrrhenian and in Adriatic watersheds (48 wells, 88 stream gauge stations, 15 springs). Spring flow at Madonna del Carmine increased from 10 l/min to 40 l/min after the earthquake, and at Monte della Guardia increased from 5 l/min to 16 l/min at the end of August (Esposito et al., 2009). At Solfatara, a volcanic crater in Pozzuoli (near Neaples), at a distance of about 100 km from the epicentre, variations in endogenous activity were observed for about 20 days after the earthquake. Majo (1931) reports a temporary decrease in fumarolic gases, diffuse H2S emanation from the soil, strong gas bubbling in a mud pool, and a notable temperature increase in monitored points. The possible influence of seismicity on gas release from depth can be demonstrated on the occasion of a small earthquake felt locally on 12 August 1930 near Pozzuoli. Table 2 reports the temperatures in °C measured in various sites inside the Solfatara crater (a) before 23 July, (b) during the period 28 July – 8 August and (c) during the period 13 – 26 August. In a nearby site, at Stufe di Nerone, a considerable increase in CO2

Site Town a b c

Fangaia (mud pool) Pozzuoli 99.5 104.5 99.4 Bocca Grande (main Fumarole) Pozzuoli 162.5 163.8 162.0 Pietra Spaccata Pozzuoli 98.0 101.5 98.2 Stufe di Nerone Bacoli 92.0 98.0 93.0

Another site with fumaroles and mud pools, Ansanto Valley, situated about 20 km from the epicenter, presented an increase in gas emission and mud boiling, together with light flashes (Alfano, 1931). This is a very sensitive site since such phenomena also occurred in this site

Table 2. Temperatures in °C measured in various sites inside the Solfatara crater.

for other earthquakes in the southern Apennines (Italiano et al., 2000).

hydrological anomalies) (IMCS≥ VIII) is also elliptical, extending over about 6000 km2.

(Gasperini et al., 1999).

and temperatures was also observed.

Near Venosa (the Vulture volcanic complex, not far from the epicentral zone) an increase in soil temperature was measured; Oddone (1932) imputed it to chemical reaction produced by a water table uplift in layers with Fe and S.

Effects on fumarole activity at distances of about 100 km from the epicenter seems hard to explain, but correlations between seismicity and volcanic phenomena are well known (Wakita et al., 1985; Hill et al., 2002; Husen et al., 2004).

Hill et al. (2002) pointed out that earthquakes and volcanoes are linked through plate tectonics and large earthquakes are capable of triggering eruptions within a matter of minutes or days at nearby volcanoes. In USA, a series of earthquakes as large as M=6.3 on 25-28 May 1980, caused turbidity and temporary increases in the discharge of hot springs in the Long Valley caldera of east-central California. These earthquakes had other obvious effects on the hydrothermal system, including emptying and refilling of boiling pools and temporary increases in fumarolic activity (Sorey and Clark, 1981). In central Japan anomalies in gas compositions were observed at fumaroles (at an epicentral distance of 9 km) and three mineral springs (at epicentral distances of 50, 71 and 95 km) about 1-3 months prior to an inland earthquake of M=6.8 on 14 September 1984 (Sugisaki and Sugiura, 1986).

Husen et al. (2004) report changes in geyser eruption behavior in Yellowstone National Park at very large distances (more than 3000 km from the epicenter) for Denali fault earthquake (Alaska), M=7.9. They interpreted these changes as being induced by dynamic stresses associated with the arrival of large-amplitude surface waves. They reported also an increase of seismic activity in Yellowstone Park and suggest that this seismicity were triggered by the redistribution of hydrothermal fluids and locally increased pore pressure.

It is plausible that such effects would occur in Southern Italy which is affected by young active tectonics with frequent strong earthquakes and many volcanically active areas (Pece et al., 1999).

Many anomalous behaviours of aquifers have been noted before, during and after a seismic event: sudden increases/decreases in spring flows, changes in piezometric levels in water wells, and increases in the emanation of deep gases (Gordon, 1970; Sorey and Clark, 1981; Whitehead et al., 1984; Wakita et al., 1985; Igarashi et al., 1992; Briggs, 1994; Curry et al., 1994; Rojstaczer and Wolf, 1992, 1994; Quilty and Roeloffs, 1997; Schuster and Murphy, 1996; Italiano et al., 2000; Thorson, 2001; Montgomery and Manga, 2003; Husen et al., 2004).

Characterizing the behaviour of aquifers and detecting anomalies in the late 1930s may be easier than in subsequent years since water resources were less exploited at that time. They are: pre- and co-seismic decreases in stream flows and water levels in wells; post-seismic increases in most of the discharges; only in some cases are they pre-seismic.

Here we illustrate the features of 7 types of the hydrological changes that we consider anomalous and connected with the 1930 earthquake. The first category of anomalous behaviour consists of decreases in stream flows before the earthquake, followed by increases after the seismic event. The four figures 3a show the data collected daily at two stream gauges (located very near great springs) on the Tyrrhenian side and at two stream gauges on the Adriatic side. In the three figures 3b the water levels measured with a 3-day frequency in one well in the Adriatic watershed and one on the Tyrrhenian side are reported, as well as the flow rate of Sanità Spring at Caposele (Sele river in figure 1b). In figure 3a.1 the anomaly consists of a sharp decrease in stream flow a few days before the seismic event, even if high rainfall preceded this decrease. The increase after the seismic event seems imputable to an anomalous discharge of the tributary springs that lasted for more than 10 days. In figure 3a.2, after the decreasing summertime trend, with a minimum reached on 24 July, there is a notable post-seismic increase from 25 July to 12 August due to

Fig. 3a.1 – 3a.2 – 3a.3 – 3a.4 (see text)

Fig. 3a.1 – 3a.2 – 3a.3 – 3a.4 (see text)

contributions from numerous large springs. In figure 3a.3 the anomaly is a temporary increase of a few cm after the earthquake.

Of great interest is the post-earthquake behaviour of 3 springs that contribute to the Sarno river (figure 3a.4). The measurements carried out at San Valentino Torio, where the total contribution of the 3 springs is measured, indicate a stream flow increase with a maximum of 127 cm on 29 July (6 days after the earthquake) followed by a decrease to a minimum of 88 cm on 12 August, a minimum level never reached before.

These types of variations have been observed for many earthquakes all over the world. In the USA, Whitehead et al. (1984) observed many significant hydrologic changes after an earthquake on 28 October 1983 in Idaho (M=7.3). Discharge measured at 10 springs and 48 stream gauging stations of the Big Lost River and surrounding watersheds increased in some instances by more than 100%. The Loma Prieta earthquake (17 October 1989) with Mw=6.9 produced hydrogeological effects reported by several authors who analysed the records of many gauging stations. Briggs (1994) analysed the hydrological effects of this earthquake in Waddell Creek watershed near Santa Cruz (California) at about 38 km from the epicenter. Numerous new springs appeared, and many inactive springs resumed flow; the springs maintained an exponential recession with minimal rain interference until they ceased flowing abruptly. As a consequence, post-seismic discharge near the mouth of Waddell Creek rose to 12.5 times the pre-earthquake discharge, followed by a gradual recession which was obscured by rain runoff beginning after about 50 days. Also Curry et al. (1994) observed very significant and unexplainable increases in the San Francisco peninsula and Santa Cruz Mountains watersheds immediately after the main shock of the Loma Prieta earthquake. For the watersheds of S. Lorenzo and Pescadero, Rojstaczer and Wolf (1992, 1994) observed that stream flows increased at most gauging stations within 15 minutes after the earthquake. Groundwater levels in the upland parts of watersheds were locally lowered by as much as 21 m within weeks to months after the earthquake.

As regard the 1930 earthquake, levels in water wells exhibited a general post-seismic increase. At Petrulla (figure 3b.1) and Bucciano (figure 3b.2) June and July rainfall did not influence the summer decreasing trend, and the increase lasted throughout August; note that at the Petrulla well the increase started 3 days before the earthquake. Figure 3b.3 shows the flow rate of Caposele spring at 22.5 km from the fault. A discharge increase of 150 liters/sec (about 3%) was measured a few hours after the seismic event, compared to the measurement on 16 July 1930, a week before the earthquake (Celentani Ungaro, 1931).

Schuster and Murphy (1996) describe an analogous hydrogeological effect for the Draney Peak earthquake, Mw=5.9 in Idaho-Wyoming (USA), on 3 February 1994: a marked increase in groundwater flow (from 4,527 to 5,695 l/min) occurred at the spring for the Auburn Fish hatchery, 5 km NE of the epicentre.

Also for the Idaho earthquake (28 october 1983, magnitude = 7.3), Whitehead et al. (1984) analyzed water levels in 69 wells: those near the epicentre generally increased rapidly after the earthquake, by as much as 3 metres. Igarashi et al. (1992), for the 2 February 1992 Tokyo Bay earthquake (M=5.9), reported possible precursor water level changes detected by the long-term groundwater observation sites. Three observation wells, about 90-110 km away from the hypocenter, showed anomalous changes: a rise and fall in water levels of 3-10 cm which began simultaneously 1-1.5 days before the earthquake. They excluded that rainfall or pumping could produce this change. The water level fall began to recover about 6 hours before the earthquake, followed by a coseismic rise of about 20 cm.

In all figures 3 the rainfall is shown. Analysis of the yearly rainfall from 1925 to 1940 shows that 1930 had slightly less than average rainfall. Moreover, the epicentral area was less rainy than the mountainous part of the Apennines and watersheds on the Tyrrhenian side. The absence of rain on the days preceding and following the event shows that the increase in the level of the aquifer was totally due to variations in spring flow rates that flow down to the river-beds.

It is difficult to assess the anomalous variations (negative or positive). In some instances the stream flow data are sufficient to permit estimates of the total "excess" stream flow derived from a particular seismic event. Using the extensive USGS hydrological network it was estimated that the Hebgen Lake earthquake (17 August 1959; M=7.5) apparently produced about 0.3 km3 of water, the Borah Peak earthquake (28 October 1983; M=7.3) about 0.5 km3

Fig. 3b.1 – 3b.2 – 3b.3 (see text)

the mountainous part of the Apennines and watersheds on the Tyrrhenian side. The absence of rain on the days preceding and following the event shows that the increase in the level of the aquifer was totally due to variations in spring flow rates that flow down to the river-beds. It is difficult to assess the anomalous variations (negative or positive). In some instances the stream flow data are sufficient to permit estimates of the total "excess" stream flow derived from a particular seismic event. Using the extensive USGS hydrological network it was estimated that the Hebgen Lake earthquake (17 August 1959; M=7.5) apparently produced about 0.3 km3 of water, the Borah Peak earthquake (28 October 1983; M=7.3) about 0.5 km3

Fig. 3b.1 – 3b.2 – 3b.3 (see text)

of water, and the Loma Prieta earthquake (17 October 1989; Mw = 6.9) only about 0.01 km3 of water (King and Muir-Wood, 1993; Rojstaczer and Wolf, 1994).

We performed an evaluation of the stage-discharge rating curves for 11 streams for which sufficient data were available (table 3). By assuming that the daily values collected in 1930 were constant in the 24 hour time frame, we calculated the average discharge in the entire anomalous period (Qav) and, obviously, the total discharge (Qtot) in this period. This permits a rough quantification of excess discharge (about 0.035 km3 for these 11 streams) which does not appear to be correlated with the distance from the epicentre.

Gordon (1970), following the Meckering earthquake (western Australia) of 14 October 1968 (mainshock Ml=6.9), reported an increase (of about 11 cm) in water level in three boreholes 110 km west of the epicentre, which started 90 minutes prior to earthquake motion and


Table 3. Excess discharge of the 23 July 1930 earthquake. Qav and Qtot are, respectively, the average and the total discharge in the entire anomalous period.

Fig. 4. Distance to epicentre (a) and to fault segment (b) versus intensity (IMCS). A clear negative linear regression is visible.

lasted about six hours. For the 18 November 1755 Cape Anne historic earthquake in New England (USA) with an epicentral intensity MM=VIII, Thorson (2001) reported hydrological responses up to 275 km from the epicentre**,** consisting in coseismic, abrupt, long-term changes in the flow rate and chemistry of water wells from five towns in Connecticut.

The anomalies were evaluated to determine whether there were patterns of hydrologic change related to epicentral or fault distance. Figure 4 shows that: (a) most of the phenomena lie between 30-120 km from the epicentre, whereas the maximum distance was 200 km; (b) most hydrological changes occurred within 30-110 km from the fault rupture segment. The maximum distance of such variations from the fault rupture was 155 km. Note that few hydrological anomalies occurred near the fault or near the epicentre (<30 km).
