Natural Hazards and Earth System Sciences Analysis the 1978 – 2008 crustal and sub-crustal earthquake catalog of Vrancea region

The crustal and sub-crustal seismicity that occurred in the Vrancea area from 1978 to 2008 is investigated. Due to quarry blast contamination, the analysis of the crustal seismicity was restricted to events that occurred between 20 km and 60 km, considering sub-crustal events as all those that occurred at depths larger than 60 km. The completeness magnitude was estimated between 2.3 and 2.5 for the crustal seismicity and between 2.9 and 3.1 for the sub-crustal one. The Gutenberg-Richter b-values show clear differences between crustal and sub-crustal seismicity, with values around 0.9 and 1.3–1.4, respectively. The analysis performed in this study represents a preliminary issue for the seismic hazard assessment of surrounding regions like Moldova, for which the earthquakes generated in Vrancea represent an important threat.


Introduction
Seismic hazard assessment depends mainly on earthquake catalogs.
In this paper, we analyse the characteristics of the earthquake catalog ROMPLUS, but restricted to the focal zone Vrancea.This zone generates earthquakes that represent a major threat to the Republic of Moldova and neighbouring territories.

Study area
The seismotectonics of the South Eastern Carpathian Mountain arch and surrounding areas are reduced to three major tectonic units: East European Platform, Moesian and Intra-Alpine Plates.Their contact, near the SE Carpathian Orogen, defines the seismic active zone of Vrancea (Cornea and Lazarescu, 1982;Sandulescu, 1984;Radulian et al., 2004).According to the instrumental observations, two categories of the Carpathian seismicity are identified: crustal and subcrustal (Radu and Polonic, 1982).The crustal seismicity of the Vrancea area is due to a series of active faults that have caused a high degree of crust fracturing (Oncescu and Trifu, 1987), while sub-crustal earthquakes occur on a slab, partially or completely detached from the continental crust in a weak sinking process (Wenzel et al., 1999;Radulian et al., 1999).The Vrancea seismic zone is characterized by maximal seismic activity at intermediate depths, with epicenters distributed within an ellipsoidal domain 60×30 km 2 , rotated by azimuth θ ≈ 45 • with respect to the major axis (Radu and Polonic, 1982;Constantinescu and Enescu, 1984).The recent relocation of several hypocenters of seismic events has increased the accuracy of the ellipsoidal domain to 90 × 20 km 2 (Radulian et al., 2007;Hurukawa et al., 2008).According to instrumental macroseismic data, the released seismic energy in the Carpathian area originated from the Vrancea seismic area accounts for about 93 %, of which 95 % generated at intermediate depths (Radu and Polonic, 1982;Bala et al., 2003).According to the ROMPLUS catalog, the maximum observed magnitude in the region was M w = 7.9, corresponding to the 26 October 1802 earthquake.Several values for the maximum probable magnitude were obtained from the magnitude-foci fault geometry relationships: M w ≈ 8.1 (Dubina and Lungu, 2003), M w = 7.9-8.0(Alcaz, 2007) and M w = 8.0 (Wenzel et al., 1999).The mean recurrence interval T as a function of the Gutenberg-Richter magnitude M is T ≈ 80 yr, for M ≈ 7.5, T ≈ 30 yr, for M ≈ 7.0, T ≈ 10 yr, for M ≈ 6.5 (Radu and Polonic, 1982;Alcaz, 2007).

Analysis of the seismicity data
We analysed the crustal and sub-crustal Vrancea seismicity that occurred from 1978 to 2008. Figure 1 shows spatial distribution of the investigated seismicity.The depth distribution is shown in Fig. 2 and reveals a minimum around 60 km, which can be considered as separating the crustal and subcrustal events.Since the crustal seismicity could be generally affected by the presence of artificial events (quarry and mine blasts), removing such man-made contaminations represents an important issue to perform before seismicity analysis; in fact, artificial events could lead to a miscalculation of seismicity rates, of b-values of the Gutenberg and Richter law (Gutenberg and Richter, 1944) and erroneous micro-seismicity analysis (Giulia, 2010).Wiemer and Baer (2000) found that quarry blasts generally occur during daytime, and introduced the day-night-normalized ratio, defined as where N d and N n are the total number of events in the daytime and nighttime period, respectively; L d is the number of hours in the daytime period and L n those in the nighttime period.The evaluation of the artificial events should take into account both the magnitude, which is less than 2.5 for Europe (Giardini et al., 2004), and also the depth, because quarry and mine blasts occur within a few kilometers from the subsurface.We checked the homogeneity of the hourly distribution of the events for different values of the magnitude m T and depth d T (measured in km) minimum thresholds.Figure 3 shows the hourdepth distribution and Fig. 4 shows the hour-magnitude distribution.A slice of the 3-dimensional histogram for a specific depth/magnitude furnishes the hourly distribution of the events for that specific depth/magnitude.It is clearly visible that the hourly distribution is rather homogeneous for depths larger than 20 km and for magnitudes higher than 2.5.Anyway, the dishomogeneity in the hour-depth distribution seems larger than that in the hour-magnitude distribution.In order to disregard the possible quarry blasts, occurring in the shallower strata of the crust, we, then, investigated the crustal seismicity given by earthquakes occurring between 20 km and 60 km.
The completeness magnitude M c is defined as the lowest magnitude of the catalog at which all the events are reliably detected (Rydelek and Sacks, 1989).The estimate of M c can be performed by several methods.In this study we used three methods: 1. the maximum curvature (MAXC) method (Wiemer and Wyss, 2000), in which the completeness magnitude is defined as the point where the first derivative of the frequency magnitude curve assumes its maximum (being the maximum of the non-cumulative frequencymagnitude distribution); 2. the goodness-of-fit (GFT) method (Wiemer and Wyss, 2000), in which the observed frequency-magnitude distribution is compared with synthetic distributions and the GFT is calculated as the absolute difference of the number of earthquakes in the magnitude bins between the observed and synthetic distribution.Synthetic distributions are calculated using estimated a-and b-values of the Gutenberg-Richter law for the observed dataset increasing the cutoff magnitude; the completeness magnitude, thus, is given by that cutoff magnitude value, at which the 90 % of the observed data is modelled by a straight line; 3. the entire magnitude range (EMR) method, firstly developed by Ogata and Katsura (1993) and then modified by Woessner and Wiemer (2005).A maximum likelihood estimation is used for a model, which comprises the complete (modelled as a power-law distribution) as well as the incomplete (modelled as a normal distribution) part of the frequency-magnitude distribution.The complete part is characterized by the Gutenberg-Richter law parameters, a and b, while the incomplete part by the parameters µ, which is the magnitude at which 50 % of the earthquakes are detected, and σ that is the standard deviation describing the width of the range where earthquakes are partially detected.By using the maximum likelihood estimation method all the parameters are calculated.
All the described methods are implemented in the Matlab software ZMAP (Wiemer, 2001).Figure 5 shows the number of earthquakes versus magnitude for the crustal (Fig. 5a) and sub-crustal (Fig. 5b) seismicity.The completeness magnitude of the crustal seismicity is 2.3 (with the MAXC and GFT methods) and 2.5 (with the EMR method), while that of the sub-crustal seismicity is 2.9 (with the MAXC and GFT methods) and 3.1 (with the EMR method).The difference between the values of the completeness magnitude in both catalogs is only 0.2 and this indicates that all three methods give approximately consistent values for the completeness The estimation of the b-value of the Gutenberg-Richter law was performed by means of the maximum likelihood estimation method (Aki, 1965), b = log 10 (e)/[< M > −M c +0.05], where < M > indicates the average of the magnitudes of the sample, M c the completeness magnitude and 0.05 is the magnitude correction.Since the completeness magnitude is in the formula, the b-value depends on the method used to estimate M c .In particular for the sub-crustal catalog, we get b = 0.942±0.02(EMR) and b = 0.879±0.01(MAXC and GFT), while for the crustal catalog we obtain b = 1.30 ± 0.05 (MAXC and GFT) and b = 1.48 ± 0.08 (EMR).Also regarding the b-value, all the methods furnish consistent values.

Conclusions
The aims of the present study were (i) checking for the quarry blast contamination, (ii) estimation of the completeness magnitude, and (iii) estimation of the b-value in Vrancea region, which is one of the most seismically active zones in Eastern Europe, where generated earthquakes constitute a source of relevant damage even in aseismic surrounding areas like the Republic of Moldova.The analysis was performed on the crustal and sub-crustal seismicity, whose border can be placed at a depth of about 60 km.The analysis of the hour-depth and hour-magnitude distributions has suggested that crustal seismicity is not influenced by quarry blasts for events that occurred at depths between 20 km and 60 km.The crustal and sub-crustal catalogs appear complete for events with a magnitude larger than 2.3-2.5 and 2.9-3.1, respectively.Differences in the seismic b-values were found between crustal and sub-crustal seismicity, with values around 1.3-1.5 and 0.9, respectively, indicating differences in the state of stress and/or in material properties.Then b could be able to provide evidence of inhomogeneities in seismogenic fault areas that can be attributed to potential asperities (Sobiesiak et al., 2007).Therefore, the difference in the b-values between crustal and sub-crustal seismicity could be a clear indication of heterogeneity of either a stress field, or different material properties in the seismogenic zone for strata shallower or deeper than 60 km.The lower b-value dominating the sub-crustal seismicity is also consistent with the main ruptures that have occurred at these depths.