1. INTRODUCTION

1It is known that Albania is one of the most active seismic countries in the Mediterranean region. During the last thirty years alone our country was hit many times by damaging earthquakes: 16 of M=5.5 (Io=VII), 3 of M=6.0-6.5 (Io=VIII), and 2 of M=6.5-7.0 (Io=IX). The strongest one was the earthquake of April 15, 1979 (M=6.8), the epicentre of which was situated between Albania and Montenegro. The consequences of the recent earthquakes and the earlier ones point out a great difference in damage type for various soil conditions. Recent investigations evidence that it is indispensable to distinguish the damage due to strong ground motion (i.e. to building vibrations) from the one due to ground failure. The experience of April 15, 1979 earthquake shows the importance of focal mechanism to damage type.

2So, the good comprehension of earthquake consequences is of primary importance in seismic hazard assessment at local level and its implementation for aseismic design criteria.

3Recently the investigations for seismic hazard assessment at local level (seismic microzoning studies) have been accomplished in many sites in Albania. Their outputs, in some cases, were compared with the observed consequences of past and recent earthquakes (in Shkodra town) or with consequences of very recent earthquakes, as the one of January 9, 1988 that hit Tirana city. They show the same trends.

4Many of these outputs are already included in the new aseismic regulations of Albania.

5We are of the opinion that works of engineering seismology for the seismic hazard assessment of a site at local level are of importance to seismic risk reduction in seismic-prone areas.

6On the other hand, a lot of problems concerning the methodologies used, output parameters and their reliability, and how to map them for physical and urban planning at local or project level, emerge.

2. CASE STUDIES ON THE INFLUENCE OF SOIL CONDITIONS ON DAMAGE TYPE IN A SITE

7The effects of local soil conditions are clearly demonstrated in many cases during past strong earthquakes and recent ones¹. These effects were taken into consideration in the seismic regionalization of Albania at national level (on scale 1:500 000).² Very significant are the effects observed during four earthquakes:

a. The earthquake of June 1, 1905 (M=6.6, Io=IX), Shkodra (Kociaj & Sulstarova 1980)

8A lot of soil instabilities such as cracks, uneven ground settlements, subsidence of river banks, due to liquefaction; rock-fall events due to topographic effects; and fault rupturing on surface (10 km long and vertical displacements on both sides of about 1 m) were observed in the epicentral zone during this earthquake.

b. The earthquake of November 30, 1967 (M=6.6, Io=IX), Diber (Sulstarova & Kociaj 1980)

9One of the most important surface phenomena observed during this earthquake was a fault rupturing 10 km long and a vertical displacement up to 0.5m, accompanied with rock-falls in a wide area.

c. The earthquake of April 15, 1979 (M=6.8, Io=IX+), Albania-Montenegro Border

10During this earthquake the influence of soil conditions was directly correlated with the structure damage in Albania and Montenegro, and was demonstrated in the distribution of observed seismic intensities in Albania (Shkodra-Lezha region)³ and in Montenegro.⁴

11In the epicentral zone the damage level was influenced by several types of soil instabilities such as: rock-fall events (in carbonate rocks, fault zones and steep rock slopes), landslides (of thin alluvium and delluvium layers from unstable slopes with triggering of active landslides and landslides of flysch sediments overlying limestones in steep slopes), liquefaction phenomena followed by cracks along the Adriatic seaside and on both sides of Buna river.

d. The earthquake of January 9, 1988 (M=5.4, Io=VII). Tirana (Kociaj & Pitarka 1989)

12This earthquake occurred just when the microzoning study of Tirana city was accomplished. The Tirana case study was a natural experiment to check up the results and methodologies used for seismic hazard assessment of a site at local level for moderate earthquakes.

3. THE SEISMIC HAZARD ASSESSMENT OF A SITE AT LOCAL LEVEL

13For seismic hazard assessment at local level detailed information is required. Investigations and analyses of seismic hazard at national, regional and local levels should be carried out considering the repeat damage due to building vibrations or ground failure. As a matter of fact, the methodologies used for seismic hazard assessment of a site at local level should start from the national and regional level.

3.1. THE SEISMIC HAZARD ASSESSMENT AT NATIONAL LEVEL

14The seismic hazard assessment at national level is already performed within the framework of the seismic regionalization of Albania on a scale of 1:500 000 ²; and, regarding the maximum expected intensity, the country is divided into four zones of VI, VII, VIII degrees MSK-64 for medium soil conditions, and IX degrees for poor soil conditions in the focal hones of maximum expected earthquakes.

3.2. THE SEISMIC HAZARD ASSESSMENT AT REGIONAL LEVEL

15Taking into account the recent developments in seismic networks, strong ground motion records, neotectonics and microtectonics, a special analysis is made to determine the expected seismic hazard at the regional level (for a zone 50 km from the site). In such analysis the determination of the active faults, capable of generating strong earthquakes and their seismic potential are of primary importance in the evaluation of input motions, ground shakings and ground failure.⁵

3.2.a. Input motions

16In our country, parts of many inhabited areas are situated just on the active faults (Vlora, Durres, Shkodra, Korca etc.), but up to now only two strong motion records have been obtained on near field conditions, on bed rock: the one of April 15, 1979 on sandstones in Ulcinj (Montenegro) as a multishock record, and of January 9, 1988 on sandstones in Tirana hills as a single event record. Their maximum accelerations in vertical components were smaller than the horizontal ones.^{4 , 8} These strongmotion records can be both considered as input motions at bed rock level.

3.2.b. Ground shakings

17During April 15, 1979, many strong motions recorded in the epicentral area have been affected by the influence of the local soil conditions.⁴

18The modification of strong ground motions by local site conditions was computed using the two above records, the El Centro record and some synthetic ones (see 3.3.b).

3.2.c. Ground failure

19As mentioned, many ground failure phenomena were observed in our country during past strong earthquakes as rupturing, vertical displacements on both sides of faults, landslides, rock-fall events, uneven settlements, cracks, ground failure due to liquefaction phenomena etc. But the question emerged as how to quantify such phenomena.

20During April 15, 1979 and January 9, 1988 earthquakes surface ground displacement on both sides of faults have not been observed because the epicentre of April 15, 1979 earthquake (M=6.8) was situated in the Adriatic Sea, and the magnitude of January 9, 1988 earthquake (M=5.4) was small.

21However, from our practice, it can be seen that for earthquakes of M=6.5 the displacements on both sides of faults in surface may reach the amplitudes 0.5 m (in mountainous regions, November 30, 1987 earthquake) and up to 1.0 m (in plain areas, June 1, 1905 earthquake). Such considerations were taken into account for a fault zone of about 100 m width in near field conditions. Rock-fall events may occur in very steep slopes even for moderate earthquakes (M=5.5)

3.3. THE SEISMIC HAZARD ASSESSMENT AT LOCAL LEVEL

22For seismic hazard assessment at local level maps on scale 1:10,000 and 1:5000 are respectively used for cities and towns.^{6 , 7 , 8} As our strong motion network has only been operating since 1986, and we have not got strong motion records in different soil conditions, two different approaches taking into account assessment through seismic intensities and through spectral values have been purposely used.

3.3.1. The seismic hazard assessment through seismic intensities

23This approach is based mainly on the elastic behaviour of soils, recording similar phenomena to earthquakes, such as microtremors and microearthquakes in different soil conditions, and using shear waves velocities measurements.

3.3.1.a Microtremor records

24Using simultaneous records in moving and reference points, the maximum amplitude (Amax) and its spectral period (Ts) are found from Fourier spectra of samples of 6-10 minutes duration. From these two parameters are determined:

• The category of soils according to A max=f (Ts) plots^{9 ,}
• The increment of seismic intensity^{10 :}

25(1) dlm= 2.0 log(Amax/Ae)

26where: Amax, Ae are maximum amplitudes in micron of grounds respectively in moving and reference points.

27From the practice it was observed that:

• Amax=f(Ts) plots are very close to reality for the determination of soil categories using Fourier spectra of simultaneous records in moving and reference points. Only those records for moving points were taken into consideration for which the simultaneous records on reference point shows for a soil of the first category (for bedrocks) or of second category (for medium soil conditions), according to Amax=f(Ts)plots.
• The ground conditions of reference point are very important. If this point is situated on hard rocks they should represents outcrops. If we deal with thrusted or slided blocks of hard rocks (as in Vlora town) the increments of intensities are +0.5 degrees greater (instead of being zero) and spectral periods Ts are greater than for outcrops. In this case such reference point is better to be avoided.

3.3.1.b Acoustic impedance method

28By means of microtremor measurements it was observed that the predominant periods in moving and reference points are not the same. In this case starting from the relation¹⁰:

29(2) a = c 2^I

30where: a - acceleration, I - seismic intensity, c - constant it can be shown that the seismic intensity increment should be:

31(3) dir = 1/2 log 2 log VoPo/ViPi + 1/log 2 To/Ti + espL (-0.04hw²) for 0.1 < T < 0.5

32where:

33Pi Vi and Po Vo are acoustic impedance for moving and reference points respectively,

34L- reducing factor equal to 1 for sands and clays and 0.5 for gravels, hw - underground water level in m.

35For 0.5 < Tp < 1.5 the second term of the formula (3) is zero.

36As it is known, this method is valid mainly for a thin layer over a half space (bedrock). So for thick alluvium deposits (typical for Adriatic coastal regions in Albania) it is not quite correct to use this method valid only for the upper part of soil profile (10m). By the way it was observed that in weathered rock, 3-5 m thick, the increment of intensities is one degree higher than in bedrocks, i.e. weathered rocks can be considered as medium soil conditions. This method can be used as a supplement to engineering-geological data, suitable to map ancient river beds.⁸

37By this method all soils can be classified in four types that coincides well with engineering-geological data.

38Highest dynamic response (increments from +2 up to +3 degrees) is observed in the sites of poor soil conditions (considered as soils of III - IV categories) which coincides with ancient swamps (in Vlora and Durres) or sometimes with historical ruins of archaeological sites of large thickness (as in Durres).

3.3.1.c Microzoning map through seismic intensities

39Based on instrumental methods, outputs and engineering-geology data, the "complex intensity" is determined as:

40(4) Ik = Ie + Ii/n

41where: Ii - Increments of intensities for different methods

42n - number of methods used

43Ie - Intensity observed in reference point.

44The determination of "complex intensity" depends not only on increments but especially on the intensity observed in the reference point. This intensity was always taken to be one degree less than the one for medium soil conditions according to engineering-geological criteria. The mapping of "complex intensity" is carried out with accuracy of +-0.2 degrees.

45The seismic hazard assessment of a site through seismic intensities includes the determination of:

• Soil categories with respective spectral periods
• Complex intensities (MSK-64 scale)
• Areas of possible soil instabilities.

3.3.2. The seismic hazard assessment through spectral parameters

3.3.2.a Formulation and parametrization of geotechnical models

46The formulation of geotechnical models or soil profiles up to bedrock depths was based on engineering-geology zoning on scale 1:10 000. For this purpose, for each site, complex investigations were carried out including: geophysical prospecting (seismic and electric), geomorphological and geological investigations and boreholes to check the accuracy of the underground topography of bedrocks and overlying spectra.

47For the parametrization of geotechnical models different kinds of logs (electric, acoustic, down hole and crosshole methods) laboratory and insitu determination of physical-mechanical properties of soils were carried out.

3.3.2.b Computing procedures

48Different computing procedures were used for different models:

1. For a one-dimensional model: wave propagation method (Milutinovic, 1982; Shuteriqi, 1982) and Tomson-Haskell method (Boncheva, Kostov, 1987).
2. For a two-dimensional model: finite element method⁸ The integral equation (Shuteriqi 1989) and finite difference method (Zahradnik 1982) are going to be applied.

3.3.2.c Outputs

49Outputs of seismic hazard assessment for a site by analytical methods are presented through computed accelerograms and their spectra 3 m from the surface (mainly for one-dimensional models and for some soil profiles).

50Parameters used are ratios of maximum values on moving(v) and reference(o) points for:

• acceleration (amaxv/amaxo)
• response spectra (Sav/Sao,Svv/Svo)
• Fourier spectra (FSv/FSo)
• Transfer functions amplitudes(Av/Ao)

51Based on Fourier spectra the soils can be classified in three groups. The highest amplification is characteristic for third category which is in good coincidence with the categories determined by microtremors. Using the ratios of maximal amplitudes of transfer functions(Av/Ao), seismic intensity increments related to reference point, can be computed taking into account that the increase of amplitudes by 2 times leads to increase of seismic intensity by 1 degree. Based on mean values of the ratios of response acceleration spectra (Sav/Sao) for one-dimensional models, all soils can be classified in four groups. For soils of first category highest response is observed in high frequency range ((for small periods Ts=0.1-.3sec)), meanwhile for soils of third category the highest response is observed for large periods (Ts>=0.5-0.7 sec).

3.3.3. The seismic hazard assessment of a site by complex data

52Taking into consideration outputs of 2.3.l.b and 2.3.2.c for final microzoning mapping following elements were taken into account (fig.1):

• Three soil categories determined from microtremor data and response acceleration spectra and their corresponding predominant periods,
• Maximum expected intensity (in MSK-64 scale) determined from "complex intensity" and ratios of maximum analytical spectral values for elastic behaviour of soils

Fig. 1. The seismic microzoning map for Vlora town

• The areas of highest amplification that coincide with soils of the fourth category (ancient swamps) where liquefaction phenomena may be developed, or with archaeological ruins of thickness more than 7 m are considered as unfavourable sites.
• Based on engineering-geological and neotectonic data the areas of possible development of surface rupturing by capable faults, landslides, rock-fall events are evaluated as dangerous sites.

4. THE IMPLEMENTATION OF MICROZONING OUTPUTS IN ASEISMIC CODE

53Recently a new aseismic code (KTP.N2.89) has been approved. Some of the most important outputs of microzoning studies are included in it as follows:

54For site selection:

1. The three types of soil categories to be determined either by
   • seismic microzoning studies
   • engineering geology data, if there is any seismic microzoning.
2. From the seismic hazard point of view sites for the construction can be divided in:
   • Favourable sites: soils of categories I and II in flat topographic conditions with underground water level below 5m from the surface.
   • Unfavourable sites: water-saturated soils of category III in flat topographic conditions, sand sediments that may be liquefied, silts, recent earthfills, archaeological ruins of great thickness; steep slopes with the slope 1:3 (including hard rocks with flat interface or tectonic cracks) where rock-fall events or rock-slides can be observed.
   • Dangerous sites: where active landslides exist or landslides can be induced by earthquakes or rupturing on surface can be observed by capable faults.

55Depending on the soil category the dynamic coefficient and seismic coefficient for aseismic design are determined.

5. CHECK UP OF SEISMIC MICROZONING STUDIES

56There are two case studies to check up the outputs of seismic microzoning studies carried out in our country

5.1. Shkodra case study

57The outputs of microzoning studies through seismic intensities are compared with consequences of two earthquakes that damaged the town during this century: the earthquake of June 1, 1905 (M=6.6, Io=IX) and the earthquake of April 15, 1979 (M=6.8, Io=IX). The epicentre of the first one was situated very close to the town of Shkodra where observed intensity reaches values from VII-IX degrees MSK-64 scale. As can be seen from the figure 2 observed intensities for different damaged objects are very close to those determined from microzoning studies.⁶ The epicentre of the second one was situated 40 km away from the town in the cross direction to geological surface structures. The observed intensity in Shkodra town for this earthquake was from VII to VIII degrees. Although observed intensities of this earthquake are 1-2 degrees less compared with those of the June 1, 1905 earthquake and with microzoning outputs, they show the same trends

5.2. Tirana case study

58It is one of more interesting cases based on the data of consequences of the 9 January 1988 earthquake that hit Tirana with seismic microzoning studies finished on the eve of this earthquake. It was shown that similar trends were observed in the distribution of observed intensities and the expected ones, although their absolute values have a difference of one degree.⁸

6. CONCLUSIONS

59The seismic hazard assessment at local level of a site by microzoning studies is an important tool in seismic areas for:

• the prediction of the earthquake risk of a site,
• the site selection in physical and urban planning,
• the reduction of the seismic risk for expected strong earthquakes.

Fig. 2 Comparison of microzoning map for Shkodra town with damage of June 1, 1905 and April 15, 1979 earthquakes