DIFFERENCES BETWEEN MEASUREMENT SITE CHARACTERISTICS AND prognosis by

In an attempt to map site response characteristics across Israel, a pilot project was carried out by the Seismology Division of the Geophysical Institute of Israel along a ~10 Km wide strip between Ashqelon and Haifa. Ambient seismic noise measurements were made at 191 sites in different lithological conditions within this area. Integration of these data with available geological and geophysical information enabled the evaluation of the characteristic site response function at each of these sites.

In this report we analyze the correlation between the surface geology (lithology of the investigated area) and the main features of the evaluated site response functions namely the resonance frequency of the surface layers and their associated level of amplification. In this study area, it has been observed that calcareous sandstone does not show significant amplification effects. The loose sediments of Sand, Alluvium and Hamra units yield amplification factors of 2-3 in the frequency range 1.2-3.5 Hz. In the Carmel coast, the complex of calcareous sandstone and loose sediments, with a total thickness of 15-30 m, that covers the Judea Group carbonates, may yield amplification factor up to 8 at frequencies ranging from 2 to 6 Hz. A statistical analysis of the 182 amplification values acquired in this study indicates that the loose sediments in the Coastal Plain may be considered as one unit with a median amplification of 3, differing from both the calcareous sandstones (Kurkar) (median amplification of 2) and the thin loose sediments in the Carmel coast overlying the Judea Group, with appreciably higher values (median amplification of 7).

The near-surface geological conditions have strong influence on the damage distribution of many destructive earthquakes (e.g. Singh et al., 1988 and Hough et al., 1990). Nearly all recent destructive earthquakes such as in Spitak, Armenia 1988, Northridge, California 1994, Kobe, Japan 1995, Armenia, Columbia 1999, Kocaeli, Turkey 1999 and many more have provided additional evidence of the dramatic importance of site effects (Borcherd et al., 1989; Spudich et al., 1996; Bouchon and Barker et al., 1996). It is evident that in order to assess the seismic hazard and the earthquake risk in an area, or for preparation of earthquake scenarios, the local site effects should be taken into consideration.

The traditional approach for considering site effects in the process of risk and damage assessments is based on corrections of intensity or the design spectral response accelerations as a function of the local lithology (see for example Singh et al., 1988; Kocaeli, Turkey, Earthquake of August 17, 1999. Reconnaissance Report ,2000). Building regulations such as those recommended in FEMA 303 (1998) and in the International Building Code (2000), are different for different soil conditions. Boore et al. (1996), among others, refer to the average velocity of shear waves in the upper 30 meters as a controlling parameter of the site effect term in the attenuation functions for predicting spectral accelerations and peak ground accelerations.

Lacking detailed information about the subsurface stratigraphy and the geo-technical properties of the soil layers in the subsurface, except in few locations, we attempt to correlate site response functions at limited number of sites with the local surface geology. Hence, the main objective of this work is to study the correlation between the surface geology, grouped into few units, with the characteristic features of the local site response, namely, the resonance frequency and the associated amplification, of the soil layers. This experiment comprises the assessed response functions of 182 sites located along the coast of Israel (Zaslavsky et al., 2002).

GEOLOGICAL BACKGROUND

The investigated area stretches along the Coastal Plain, 10 km wide and 140 km long, from Asqelon to Haifa. In this area, three units were recognized in the seismic amplification hazard maps prepared by Rosensaft et al. (1999), based on surface and near-surface geology. These units represent two classes of amplification potential, namely “Intermediate” and “High”. In these maps, “Low” potential for amplification was assigned to hard rocks, mainly of the Judea Group, that are not exposed in our area of study. In this study we use the following annotation for the three lithological units:

Unit 1 is identical with the “Kurkar” mapping unit of Sneh and Rosensaft (1994), comprising Quaternary marine and eolian calcareous sandstone with intercalations of loam. It is not synonymous with the “Kurkar Group” but is a part of it, and to a large extent represents the Kurkar coastal ridges and other kurkar outcrops in the coastal plain. This unit covers about 5% of the area investigated.

Unit 2 is identical to the “sand dunes” mapping unit of Sneh and Rosensaft (1994), comprising recent loose sands of Quaternary age, 5-20m thick. These sediments are exposed in a strip along the coastline, reaching at some places a distance of 5km from the coast, covering about 20% of this area. This unit overlies either the “Kurkar” mapping unit or Holocene alluvium.

Unit 3 comprises the “hamra” and the “alluvium” mapping units of Sneh and Rosensaft (1994). The ‘hamra” mapping unit is made of ferruginous red clayey sand and loam and it is exposed mainly between the Kurkar ridges. The “alluvium” mapping unit consists of clay and silt in stream floodplains. These sediments usually overlie the Kurkar Group. Unit 3 spreads over about 80% of the investigated area.

These four mapping units belong to the Kurkar Group, which consists mainly of alternating calcareous sandstone and loam. The thickness of Kurkar Group varies from 180-200m near the shoreline to 100-120m at the distance of 5-7 km from the coastline. The Kurkar Group is underlain by clays of the Yafo Fm. of Pliocene age, but between Binyamina and Haifa (Carmel coast), the Kurkar Group directly overlies hard carbonates of the Judea Group. Here it is only 50m thick near the coastline, wedging out eastwards at the foot of the Carmel, where the Judea Group outcrops.

SITE RESPONSE FUNCTIONS

A detailed description of the work carried out to evaluate the site response functions in the investigated area is presented by Zaslavsky et al. (2002). In accordance with the geological map we planned 22 lines of measurement from Ashquelon to Haifa so that 18 of them would be perpendicular the coastline and intersect the three different geological units, and four lines (19-22) along the coastline (see Figures 1). The distance between the lines is approximately 5km. Altogether, site response investigations were carried out at 182 points (stations) along these lines. Most of the stations are located close to boreholes. The number of stations, distributed along each line and the distance between them, depended on the spatial distributions of the geological units and the availability of borehole data.

Figure 1. Geological map and location of recording sites: a – the central coastal plain; b - Carmel coast.

At each of the investigated site we recorded the seismic ambient noise using 1 Hz seismometers. The technique of Nakamura (1989) was applied to detect the fundamental resonance frequency of the soil layer at the site and its associated level of ground motion amplification. Nakamura (1989, 2000) hypothesized that site response could be estimated by simply evaluating the spectral ratio of the horizontal versus vertical components of noise observed at the site. The theory and results obtained by implementing the Nakamura technique (Ochamachi et al., 1991; Lermo and Chavez-Garcia, 1994; Seeking et al., 1996; Toshinava et al., 1997; Chavez-Garcia and Cueuca, 1998; Zaslavski et al., 1995, 2000) support the use of ambient noise measurements to estimate the site response of surface deposits to emerging seismic waves.

Seismic noise was measured at every site for at least 2 hours to assure that we have long enough recordings that are free from transient disturbances such as a moving vehicle. From each data file we selected several segments of 30 sec for spectral calculations. This time window is proven to be sufficiently long to provide stable results. The selected time windows were Fourier transformed using cosine tapering before transformation. Following the smoothing of the spectra with a triangular moving Hanning window, the spectra of an EW and NS channel at a station were divided by spectra of the vertical channel in order to obtain spectral ratios (Nakamura estimate). The spectral ratios, obtained from different time windows, are averaged. From the average spectral ratio we obtain the fundamental frequency and its associated level of amplification. Examples of site response spectra for the three geological units are presented in Figures 2, 3 and 4.

Figure 2a shows the individual and average horizontal-to-vertical spectral ratio obtained from ambient noise recorded at Site 49 on line 14 and site 25 on line 3. Both sites are on geological unit 1 (the Kurkar mapping unit). We observe that for Site 49 the average spectral ratio is flat in the frequency range 0.3 to 10 Hz, suggesting no amplification. Figure 2b displays spectral ratios for Site 25 with a well-defined peak at about 1.9 Hz with an amplification factor of 2.

Figure 2: Individual and average horizontal-to-vertical spectral ratios for stations on Unit 1

Sites 65 and 20 are on loose sand (unit 2). Figure 3a displays individual and average spectral ratios at Site 65 (an estimated amplification factor up to 3 in the range 2.5 to 3.0 Hz). Figure 3b shows individual and average spectral ratio for Site 20. In this case we see a prominent peak at about 2.5 Hz, having an amplification factor up to 4.

Figure 3. Individual and average horizontal-to-vertical spectral ratios for Unit 2

Examples of the horizontal-to-vertical spectral ratios at sites measured in Unit 3 (“Hamra”) are plotted in Figure 4 a,b,c,d. This unit is characterized by a variety of both frequency and amplification factor values. In the frequency range from 1.2 Hz to 5.8 Hz we observe amplification factor values up to 9.

Figure 4. Individual and average horizontal-to-vertical spectral ratios for sites in Unit 3

The question of whether the Nakamura technique accurately determines the amplification level is still debated (see for example Bonilla et al., 1997; Satoh et al., 2001). However, it is generally agreed that this technique is reliable in detecting the resonance frequency of the soil column and in many cases will also provide a good estimate of the amplification level. Realizing the potential deficiency of the Nakamura technique, we adopted site response functions that are derived analytically using the computer code of Joyner (1977). Assuming that the soils behave linearly over the whole range of expected acceleration levels, the computation requires a one-dimensional model of the subsurface, which includes information about the thickness, density, water content and shear-wave velocity of each of the soil layers and the underlying rock. Development of the 1D model was based on information, when available, from nearby boreholes, refraction surveys and lithological cross sections. By means of trial-and-error we choose the model parameters that fulfill the principle conditions:

1. The inferred analytical response function matches the observed average H/V spectral ratio from many ambient noise measurements. An example is shown in Fig. 5.

2. The inferred analytical response function is consistent with the models attributed to nearby measuring points to form a systematic picture of the physical characteristics of the subsurface in the area.

Under these conditions we would interpret lateral increase of the resonance frequency as indication of the thinning of the soft soil layers and enhanced amplification would be interpreted as an indication for low velocity soils such as along river trails.

Figure 5. Example of the H/V spectral ratio evaluated from ambient noise measurements and the evaluated response function of the site

The characteristics of the response functions of the analyzed sites can be quantified in terms of the maximum amplification level and its associated resonance frequencies. In tables 1, 2 and 3 we grouped the results in accordance with the amplification potential units of Rosensaft et al. (1999).

Table 1. Maximum amplification level at fundamental resonance frequencies measured in sites in Unit 1.

NN	Line	Site	Coordinates		Resonant frequency, Hz	Amplification factor
NN	Line	Site	EW	NS	Resonant frequency, Hz	Amplification factor
1	1	1	125386	157825	-	1
2	2	5	131835	162062	1.6	1.8
3	3	9	123566	148655	-	1
4	4	17	120904	147131	4.2	2.9
5	6	21	115253	134038	2.7	3
6	7	25	116580	126510	1.9	2.2
7	8	32	111635	121022	1.8	2
8	11	43	131290	168137	-	1
9	10a	49	132019	170488	-	1
10	10b	56	129899	164323	-	1
11	10c	52	130502	166576	1.8	2.8
12	10c	53	133643	165427	-	1
13	10c	57	131607	176460	-	1
14	12	61	133101	181543	2.4	2.3
15	14	67	136335	192379	-	1
16	15	70	137505	193056	-	1
17	15	71	138482	198785	-	1
18	16	74	137425	199138	-	1
19	17	76	142876	205247		1
20	17	78	141182	211415	-	1
21	19	85	142359	217373	-	1
22	20	89	143179	223997	-	2.1
23	20	90	144175	227157	-	1
24	21	94	145279	235221	-	1
25	22	102	146154	242788	-	1
26	22	108	128850	136176	-	1
27	22	112	125898	135711	-	1
28	5a	109	125722	139070	-	1
29	5a	110	122889	140517	-	1
30	16a	128	129104	161222	-	1
31	1a	129	128176	159349	4.2	2.6
32	18a	133	142156	213612	-	1
33	17a	138	136895	196171	3.7	2.6
34	8a	146	119307	123524	-	1
35	10a	48-D	130029	170824	2.1	3
36	11	143	131290	168137	-	1

Table 2. Maximum amplification level at fundamental resonance frequencies measured in sites in Unit 2

NN	Line	Site	Coordinates		Resonant frequency, Hz	Amplification factor
NN	Line	Site	EW	NS	Resonant frequency, Hz	Amplification factor
1	1	2	128820	156486	1.7	2.2
2	2	4	127508	163294	3.7	4.3
3	3	8	121801	148864	-	1
4	4	15	123779	141962	-	1
5	4	16	122277	144229	2	3.3
6	5	18	117858	137746	-	1
7	6	20	115141	133644	2.7	4.8
8	6	22	116876	133414	1.7	2.1
9	6	147	118091	132946	1.4	3.3
10	7	24	112370	128330	2.6	4.3
11	8	28	109940	123448	3.1	2
12	8	30	111168	123481	1.9	2
13	9	33	108179	121049	3	2
14	9	35	108239	120367	3	3.1
15	9	37	107669	118987	-	1
16	11a	D1	129640	169500	3.5	2.8
17	11a	D2	129280	168150	2.5	5
18	11a	45	131163	172887	3	3.7
19	13	64	136584	188116	1.6	3
20	14	65-B	137546	188091	1.7	2
21	14	65-C	137851	188070	2	2
22	16	75	139061	205949	3.1	3
23	18	79	143287	211190	1	2.4
24	18	81	141247	215802	2.8	2.5
25	20	89-2	142510	223740	3	4.2
26	20	93	143524	228981	2.6	7
27	21	97	144915	235126	2.9	5.5
28	22	103	146195	245663	-	1
29	5a	111	120650	140378	2.8	5.3
30	23	120	128985	152178	3.5	2.1
31	3a	119	127396	152655	-	1
32	13a	125	139126	201418	-	1
33	16a	126	140326	200886	2.1	3.1
34	1a	131	140961	213757	6.8	2.8
35	18a	132	142695	213165	1.6	5
36	18a	135	140335	209208	5.3	3.4
37	17a	136	142463	208697	2.3	2.2
38	15a	139	137911	196400	3	3.5
39	15a	141	114315	131070	3.7	2.6
40	7a	142	115343	130550	6.7	2.9
41	7a	144	112955	126537	2.2	3.9
42	14	65-1	137127	187319	1.6	2.2

Table 3. Maximum amplification level at fundamental resonance frequencies measured in sites in Unit 3

NN	Line	Site	Coordinates		Resonant	Amplification
NN	Line	Site	EW	NS	frequency, Hz	factor
1	1	3	133348	155362	1.2	2.2
2	2	6	133935	162148	1.4	2.3
3	3	10	125780	148126	2.3	3.7
4	3	11	126620	147841	2.8	4.6
5	4	14	124795	140965	-	1
6	5	19	120994	136049	1.6	2.5
7	6	23	120240	132129	_	1
8	7	26	115290	126780	1.8	2.4
9	7	27	118870	126465	2.7	4
10	9	36	110387	119762	_	1
11	11	44	134978	167161	-	1
12	10	45-A	131000	172811	3.5	3
13	10	45-B	131287	172954	3.2	2.6
14	10	46-A	132419	172568	1.9	1.8
15	11a	46B	132675	172263	2	2.3
16	10a	48-M	131080	171218	3.8	3.1
17	10a	48B	130566	171149	4.3	4.4
18	11a	48E	130197	171240	2.4	3.8
19	10a	48G	130681	171074	2.9	4.7
20	10a	48H	130796	171200	3.4	3.5
21	11a	48P	131464	170914	3.9	5.3
22	10	46	132492	172487	-	1
23	11a	47	135484	171924	-	1
24	11a	48	130526	170961	3.5	4
25	10a	50	133752	169524	-	1
26	10b	51	130062	167045	1.4	5.7
27	10a	54	128762	165920	3	3.4
28	10b	55	129411	164944	-	1
29	12	58	132461	176784	3.4	2.9
30	12	59	134015	175480	1.7	2.7
31	12	60	136180	175146	1.6	2.3
32	13	62	134842	180567	2	1.9
33	13	63	136961	180025	1.8	3
34	14	65-2	137690	189418	1.8	2
35	14	65-3	139160	187874	2.7	3.7
36	14	65-4	140496	188501	1.7	2
37	14	65-A	138550	188430	2	2.6
38	15	69-2	142107	191215	-	1
39	15	69-A	141665	190984	1.3	2.3
40	14	66	140902	187756	1.3	2
41	15	68	139685	191433	2.2	4
42	15	69	141636	190997	1.4	2
43	16	72	139981	197520	1.5	2
44	16	73	142661	196708	1.9	3
45	17	77	143539	204450	-	1
46	18	80	146986	210794	1.8	3.3
47	19	82	142592	215766	2	7
48	19	82-C	142757	215736	1.9	6.2
49	19	82-E	143657	215610	2	7.8

Continuation of Table 3
NN	Line	Site	Coordinates		Resonant	Amplification
NN	Line	Site	EW	NS	frequency, Hz	factor
50	19	82A	142429	215479	4	5.8
51	19	82D	143209	215540	3	4.5
52	19	82L	143377	216030	2.2	5.4
53	19	82Q	142468	215867	3.1	2.8
54	19	82R	143835	215961	2.3	7
55	19	82T	143983	215201	3	7
56	18	83-2	143717	214501	3	6.8
57	19	84-1	143903	217394	-	1
58	19	84-A	143387	216971	2.1	7.8
59	19	84-C	143451	216445	2.5	6
60	19	85-1	142422	217342	1.2	7.8
61	19	86A	143858	219155	4.4	7
62	20	87-1	142510	220584	2.7	4.5
63	20	89-1b	144651	224045	1.8	5.4
64	20	90-1b	145425	227398	2.1	6.5
65	20	92-3a	146630	232920	-	1
66	20	93-1b	143382	227733	3.2	7
67	21	95-1a	146317	235648	5.7	3.6
68	21	95-2b	146459	238058	3.1	7.2
69	18	83	144892	214868	3.9	5.3
70	19	84	143845	216369	-	1
71	20	87	144492	219904	-	1
72	20	88	144624	222151	2	7
73	20	91	145460	229101	2.1	7.8
74	20	92	145583	231336	2.1	7
75	21	95	147064	235047	5.7	2.7
76	21	96	146928	239631	3.4	5.6
77	21	98	146577	242195	4.2	6.6
78	22	99	146857	244771	4.7	7
79	3a	118	131386	151589	-	1
80	1a	130	135852	158205	1.5	2.6
81	7a	143	118423	128997	-	1
82	8a	145	117319	124502	3.2	3.5
83	3	149	130154	147638	1.4	2
84	6	151	124684	131826	_	1
85	22	100	146530	245738	5.7	6.7
86	22	101-A	146303	246745	7.7	4.1
87	22	104	146260	247122	7.7	7
88	3a	121	134683	184638	6.8	6
89	13a	122	135718	183728	-	1
90	13a	123	136338	183095	2	2.3
91	13a	124	140352	181657	1.7	2.5
92	16a	127	145658	200130	-	1
93	18a	134	144696	212548	2	2.9
94	17a	137	144531	208188	1.2	3.1
95	15a	140	140975	195737	6.1	2.9
96	3	150-A	128601	144339	1.6	2.5
97	3	152-A	127922	143214	1.8	2.4
98	11a	D3	131900	167750	1.5	3
99	11a	D4	129300	167650	3	4.5

STATISTICAL DISTRIBUTION OF SITE AMPLIFICATION LEVELS IN THE DIFFERENT UNITS.

The histograms shown in Figure 6 present the distribution of amplification factors of ground motions for the three amplification-potential units (Rosensaft et al., 1999). In Unit 1 (calcareous sandstone), all of the 36 sites have low (less than 3) to no amplification. About 75% of them have amplification less than 2, a level that can hardly be detected empirically by the methods used. In Unit 2 (loose sand) the amplifications change between 1 (no amplification) to a factor of almost 6. In this data set 60% of examined sites are associated with amplification factors less than 3 and the rest reveal amplifications factors between 3 and 6. In Unit 3 (alluvium and loam) the amplification factor histogram has a shape similar to that of unit 2, yet it includes many more sites with amplification factors of 5 and above.

Checking the locations of the sites with relatively high amplifications we note that they are concentrated along the Carmel Coast area. This zone stretches from Binyamina in the south to Haifa in the north along the coastline. It is approximately 3 km wide and 32 km long. 90% of this area is covered by the alluvium and hamra mapping units (Unit 3). Unit 1 emerges at the surface as a narrow strip, 300-500m wide, and covers 5-7% of the area. The rest of the area is covered by dunes (Unit 2) with a thickness of ~2 m.

The information about the site response characteristics of the sites located in the Carmel Coast is given in Table 4 and Figure 7. It shows that in a frequency range of 2 to 5.5 Hz, the amplification factor ranges from no amplification up to a factor of 8. The statistical distribution of the amplification levels at the sites in the rest of the coastal plain is shown in Figure 8.

Table 4. Maximum amplification level at fundamental resonance frequency by lithological units witnin the Carmel Coast

NN	Line	Site	Coordinates		Resonant	Amplification
NN	Line	Site	EW	NS	frequency, Hz	factor
Kurkar
1	19	085	142359	217373	-	1
2	20	089	143179	223997	-	2.1
3	20	090	144175	227157	-	1
4	21	094	145279	235221	-	1
5	22	102	146154	242788	-	1
Sand dunes
1	20	89-2	142510	223740	3	4.2
2	20	93	143524	228981	2.6	7
3	21	97	144915	235126	2.9	5.5
4	22	103	146195	245663	-	1
5	19	81	141247	215802	2.8	2.5
Alluvium and Hamra
1	19	84	143845	216369	-	1
2	19	84-1	143903	217394	-	1
3	19	082	142592	215766	2	7
4	19	82-A	142429	215479	4	5.8
5	19	82-D	143209	215540	3	4.5
6	19	82-L	143377	216030	2.2	5.4
7	19	82-Q	142468	215867	3.1	2.8
8	19	82-T	143983	215201	3	7
9	19	82-R	143835	215961	2.3	7
10	19	86-A	143858	219155	4.4	7
11	20	87	144492	219904	-	1
12	20	87-1	142510	220584	2.7	4.5
13	20	88	144624	222151	2	7
14	20	89-1b	144651	224045	1.8	5.4
15	20	90-1b	145425	227398	2.1	6.5
16	20	91	145460	229101	2.1	7.8
17	20	92	145583	231336	2.1	7
18	20	92-3a	146630	232920	-	1
19	20	93-1b	143382	227733	3.2	7
20	21	95	147064	235047	5.7	2.7
21	21	95-1a	146317	235648	5.7	3.6
22	21	95-2b	146459	238058	3.1	7.2
23	21	096	146928	239631	3.4	5.6
24	21	098	146577	242195	4.2	6.6
25	22	099	146857	244771	4.7	7
26	22	100	146530	245738	5.7	6.7
27	22	101-A	146303	246745	7.7	4.1
28	22	104	146260	247122	7.7	7
29	19	82-C	142757	215736	1.9	6.2
30	19	82-E	143657	215610	2	7.8
31	19	84-A	143387	216971	2.1	7.8
32	19	84-C	143451	216445	2.5	6
33	19	85-1	142422	217342	1.2	7.8

Figure 8. Distribution of amplification factors at sites along the coast of Israel, excluding sites in the Carmel Coast.

In an attempt to draw some general conclusions concerning the spatial distribution of the amplification values, the measuring sites were classified according to their lithological characteristics, based on the digital geological map of Israel at a scale of 1:200,000 (Rosensaft et al., 1999) As it is based on maps at a scale of 1:50,000, the location of the boundaries between units is accurate at the level of some tens of meters. For each unit, the cumulative frequency of amplification values was drawn (Fig. 9), separating the Carmel Coast sites from the other ones. This figure shows the similarity of amplification values for the loam (hamra), alluvium and loose sand lithologies, the lower values of the calcareous sandstone (kurkar) and the much higher values in the Carmel Coast (median of 7). A Mann-Whitney ranking test was applied to the data, showing that the amplification values of the calcareous sandstone (median of 2) is significantly (p<0.02) lower than that of the loam and is significantly (p<0.01) lower than that of either the alluvium or the loose sand (median is 3 for all three lithological types). Hence, the amplification values of the soft sediments can be grouped together, and a confidence interval can be calculated for each value (Figure 10). The large number of measurements (111) allows a rather small confidence interval.

Site response is a result of a number of factors that cannot be well quantified solely on the basis of the lithology of surface layers. As a very first approximation, the resonance frequency of the soil, f, and the amplification, A(f), are given by the equations:

Where H₁ is soil depth, V₁ is average shear wave velocity in the soil column and V₂ is shear wave velocity in the half-space; ρ₁ and ρ₂ are densities of materials in surface layer and half-space, respectively.

Evidently, a site classification scheme should be developed with the objective of encompassing the factors having the greatest influence on seismic site response as key parameters: shear-wave velocities and densities for soils and rocks and soil thickness. From this position we have subdivided the area investigated into two geographical zones: the first zone is Ashqelon-Binyamina zone (central region of the Coastal Plain), where the reflector (half space) is the calcareous sandstone of the Kurkar Group. The impedance contrast in this case is formed by the overlying loose sediments, in most cases less than 10m thick. The second zone is the Carmel coast, where the reflector is the hard carbonates of the Judea Group, and the upper layer, 10-50 m thick, consists of sandstone, silt and loam. This geological combination is the cause of the high amplifications observed in this zone.

As demonstrated in the histogram of unit 1 (kurkar), in 80% of the sites the amplification factors are less than 2. In the central coastal plain along the seashore, the kurkar unit is underlain by clays of the Yafo Fm., with no impedance contrast between the surficial and underlying beds. This explains the low amplification values there. In the Carmel coast, the kurkar unit overlies the hard carbonates of the Judea Group, which results in an impedance contrast and slightly higher amplification (less than factor 3).

In sites located on loose sand, the geological situation is that of thin (less than 10m) sand layer overlying sandstone. Such conditions are not expected to yield significant amplification. At sites where the thickness of loose sand or the combined thickness of the sand and the underlying alluvium and hamra exceeds about 10 m., we expect amplification more than factor 3.

In the central region we have not observed significant difference in the distribution of amplification factor between Units 2 (sand) and 3 (alluvium and hamra). However, in the Carmel coast, these units show high amplification levels at many sites. This observation is possibly associated with the thickness of Holocene sediments that are situated along riverbeds and can reach 50 m and with the relatively high impedance contrast between the soft sediments and the underlying bedrock, which is composed of hard carbonates of the Judea Group.

Figure 9. Cumulative distribution of amplification factors in the coastal plain.

Red – Alluvium in the Carmel coast (38 sites). Other lines represent sites in the in the central coastal plain: Purple – alluvium (23 sites); Green – loose sand (42 sites); Light blue – hamra (46 sites); Blue –kurkar (23 sites).

Figure 10. Cumulative distribution of amplification factors in the coastal plain.

Owing to data limitations on soil characteristics and the soil/bedrock profile, many models assume a simplified basin shape. Generally, the real depth to bedrock and the basin profile is not well known. However, borehole and geophysical data suggest that soil-bedrock interface irregularly causes the soil depth to fluctuate dramatically at the local scale. Depending on the sub-surface site conditions, an irregular bedrock surface can concentrate seismic waves through the soil beneath the surface so that some locations behave as focal points, while others experience amplified or dampened velocities. This results in highly variable damage patterns that are very difficult to predict. Figure 11 shows observed horizontal-to-vertical spectral ratio obtained from ambient noise recorded at Sites 85-1 and 85, which are only 150m apart. These sites demonstrate the great variability in site response possible over very short distance. Spectral ratios for site 85 are flat with no amplification while average spectral ratio of Site 85-1 shows a prominent peak at 1.2 Hz with amplification factor up to 8.

Figure 11. Individual and average spectral ratios obtained from microtremor measurements at sites a – 85; and b – 85-1. (Site 85-1 is about 150 m from site 85)

1. Based on site response analysis of 191 sites located along the coast of Israel between Ashqelon and Haifa we propose the generalized site classification scheme as follows:

· Kurkar resting on clay of the Yafo Fm., or loose deposits (sand, alluvium, hamra) with thickness of a few meters overlaying kurkar will probably not yield amplification effects.

· Loose deposits lying on kurkar and having thickness from 10m to 50m yield amplification factor 2-3 in frequency range 1.2-3.5 Hz.

· The bedrock represented by either hard calcareous sandstone or limestone gives higher impedance contrast with overlying sediments and consequently amplification can increase up to factor 6 in the frequency range 1.5-2.5 Hz.

· In the Carmel coast, loose sediments and kurkar with a total thickness of 15-30 m., overlying the Judea Group carbonates can yield amplification factor up to 8 in the frequency range from 2.0 to 6.0 Hz.

2. As a first approximation, surface geology of the central coastal plain is correlated with the distribution of amplification values. The statistical analysis shows that the amplification data acquired in this study are in accord with the choice of units for the amplification potential map of Rosensaft et al. (1999): An “Intermediate” potential for the calcareous sandstone (Unit 1) and a “High” potential for the loose sand (Unit 2) as well as loam and alluvium (Unit 3). Figure 9 allows an approximate generalization of the results in this area. On the other hand, in the Carmel coast, surface geology does not properly reflect the high amplification values, because the main effect is created by the high impedance contrast between the underlying hard rocks and the thin overlying softer rocks.

3. Correlation between the lithological units and site amplification effects should be limited to the area where the study was carried out. The site response is a multi-parametric problem and therefore the surface geology alone may be insufficient for estimating the site effects on seismic hazard evaluations, especially when moving from one area to another.

4. Site effects may vary significantly over very short distances, even in cases associated with the same geological unit. Therefore, seismic microzonation based on detailed site response investigations is very important for safer design of new buildings and/or for preparing detailed earthquake scenarios for a city.

Bonilla, L. F., Steidl, J. H., Lindley, G.T., Tumarkin, A. G., and Archuleta, R.J., 1997.

Site amplification in the San Fernando Valley, California: Variability of site effect estimation using S-wave, Coda, and H/V method, Bull. Seism. Soc. Am. 87, 710-730.

Boore, D.M, Joyner, W.B., Fumal, T.E., 1997. Equations for estimating horizontal response spectra and peak acceleration from western North American earthquakes: A summary of recent work, Seism. Res. Lett. 68, 128-153.

Borcherdt, R., Glassmoyer, G., Anrews M., and Granswick, E., 1989. Effect of site conditions on groun motion and damage, Earthquake Spectra, Special Supplement, 23-42.

Bouchon, M., Barker, J.S., 1996. Seismic response of a hill: the example of Tarzana, California, Bull. Seism., Soc. Am., 86:66-72.

Chavez-Garcia, F. J. and Cuenca, J., 1998. Site effects and microzonation in Acapulco, Earthquake Spectra, Vol. 14, No. 1, 75-93.

Hough, S.E., Borcherdt, R.D., Friberg, P.A., Busby, R., Fielg E., and Jacob, K.N., 1990. The role of sediment-induced amplification in the collapse of the Nimitz freeway , Nature, 344:853-855.

Joyner, W.B., 1977. A Fortran program for calculating nonlinear seismic response, U.S. Geological Survey, Open File Report, 77-671.

Kocaeli, Turkey, Earthquake of August 17, 1999. Reconnaissance Report, Earthquake Spectra, Suppl. A to vol. 16, 2000-03.

Lermo, J. and Chavez-Garcia, F. J., 1994. Are microtremors useful in site response evaluation?, Bull. Seism. Soc. Am. 84, 1350-1364.

Nakamura, Y., 1989. A method for dynamic characteristics estimation of subsurface using microtremor on the ground surface. QR of RTRI, Vol. 30, No.1, 25-33.

Nakamura Y., 2000. Clear identification of fundamental idea of Nakamura’s technique and its applications. 12^th Wirld Conference on Earthquake Engineering, Auckland, New Zealand, 2000.

Ohmachi, T., Nakamura, Y., Toshinawa, T., 1991. Ground motion characteristics of the San Francisco bay area detected by microtremor measurements. Second International conference on recent advances in geotechnical earthquake engineering and soil dynamics, St. Louis, Missouri, 1643-1648.

Rosensaft, M., Wacks, D. and Sneh, A., 1999. Association of surficial lithological

units according to an assessment of their potential to amplify earthquake oscillations. 4 maps at a 1:200,000 scale. The Geological Survey of Israel.

Satoh, T., Kawase, H., Matsushima, S., 2001. Difference between site characteristics obtained from microtremors, S-wave, P-waves, and codas, Bull. Seism. Soc. Am. 91, 313-334.

Seekins, L. C., Wennerberg, L., Margheriti, L., and Liu, H-P., 1996. Site amplification at fife locations in San Francisco, California: a comparison of S waves, codas and microtremors, Bull. Seism. Soc. Am. 86, 627-635.

Singh, S. K., Mena E., and Castro, R., 1988. Some aspects of source characteristics of the 19 September 1985 Michoacan earthquakes and ground motion amplification in and near Mexico City from strong motion data, Bull. Seism. Soc. Am. 78, 451-477.

Sneh, A. and Rosensaft, M., 1996, Geological map (1:100,000) of the coastal plain of Israel, Netanya, Tel Aviv-Yafo, Ashqelon and Nirim sheets. Geological Survey of Israel Report, GSI/29/94.

Spudich, P., Hellweg M., and. Lee, W.H.K., 1996. Directional topographic site response at Tarzana observed in aftershocks of the 1994 Northridge California’ earthquake: Implications for mainshock motions, Bull. Seism. Soc. Am., 86:193-208.

Toshinawa, T., Taber J. J., and Berill, J. B., 1997. Distribution of ground motion intensity inferred from questionnaire survey, earthquake recordings, and microtremor measurements- a case study in Christchurch, New Zealand, during 1994 Arthurs pass earthquake. Bull. Seism. Soc. Am. 87, 356-359.

Zaslavsky, Y., Gitterman, Y. and Shapira, A., 1995. Site response estimations in Israel using weak motion measurements, The fifth International conference on Seismic Zonation, Nice, France, 1713-1722.

Zaslavsky, Y., Sapira, A., Arzi A., 2000. Amplification effects from earthquakes and ambient noise in the Dead Sea rift (Israel), Soil Dinamics and Earthquake Engineering 20, 187-207.

Zaslavsky, Y., Shapira, A., and Leonov, J., 2002. Empirical evaluation of site effects by means of H/V spectral ratios at the locations of strong motion accelerorometers in Israel, Tectonophysics (in print).