A N OVERVIEW OF THE TECHNIQUE FOR SEISMICITY MICROZONATION MAPPING OF THE U LAN ‐ U DE CITY TERRITORY

: For purposes of seismicity microzonation of the Ulan‐Ude city territory, engineering geophysical studies are conducted to reveal which types of rocks and soils are dominant in the study area and to classify them by site‐specific velocities of P‐ and S‐waves and amplitude‐frequency characteristics. The article describes a technique for establi‐ shing the baseline seismic signal corresponding to parameters of relatively strong earthquakes in potential earth‐ quake foci (PEF) zones. It is shown that the established baseline signal is applicable. Presented are results of theoreti‐ cal calculations based on seismicity‐soil models providing reference parameters of bedrock, medium and water‐ saturated soils (soil categories 1, 2 and 3, respectively). Seismic impacts are assessed for the zone with the baseline seismic intensity of 8 points, as per MSK‐64 seismic in‐ tensity scale. The reference model is used to identify zones with seismic intensity from 7 to 9 points in the city territo‐ ry, and it is established that such zones differ in thickness of water‐saturated and non‐water‐saturated soil layers. As a result, a schematic map showing the main parameters of seismic impacts is constructed in the first approxima‐ tion. The obtained data are useful for the development of recommendations concerning further engineering seismo‐ logical studies and activities for the appropriate revision and upgrading of the seismic microzonation technique in order to complete seismic microzonation of the Ulan‐Ude city territory.


INTRODUCTION
The city of Ulan-Ude is located in the seismically active region of Russia. According to [The Map..., 1999], the city's territory with medium soil conditions is characterized by seismic intensity of 8, 8 and 9 points (as per MSK-64 seismic intensity scale) and three levels of seismic hazard with 10 % (A), 5 % (B), and 1 % (C) probabilities of exceedance in 50 years. Analyses of historical macroseismic data [Solonenko, Treskov, 1960] show that the maximum seismic impact of the strongest earthquakes in Ulan-Ude did not exceed 7 points even with account of two major events which took place in the Southern and Central Baikal regions -the Tsagan (12 January 1862; M=7.5) and Middle Baikal earthquakes (29 August 1959;M=6.8).
Under the code of practice in the construction industry in the Russian Federation, the baseline seismic intensity was assessed according to the RF construction standards and rules specified in SNiP II-А.12-69* dated 01 July 1970, and for the Ulan-Ude territory with medium geological conditions, it was estimated at 7 points. Later on, SNiP II-А.12-69* was replaced by SNiP II-7-81 (in force since 01 January 1982), and the baseline seismic intensity for Ulan-Ude is now estimated at 8 points [The USSR Seismic Zonation Map, 1984] which assumes the recurrence of a major seismic event every 1000 years, according to Map (B) under SNiP II-7-81* (updated revision) [SNiP…, 2011].
It is envisaged by the current construction regulations and standards that optimal locations must be selected for construction projects with account of seismic resistance calculations, which necessitates quantification of the main parameters of seismic impacts that may be imposed to foundations of building and facilities. In this regard, a seismicity microzonation (SMZ) map needs to be constructed for the Ulan-Ude city territory in scales 1:25000 and 1:5000 with account of new assumptions. The required mapping should be preceded by stages when a general seismic zonation map and a detailed seismic zonation map of the territory are constructed in larger scales. Therefore, to achieve the objective of seismic microzonation mapping of the Ulan-Ude city, it is required to update the general seismicity zonation data and assess levels of seismic hazard for new construction project areas in the city. These tasks can be fulfilled by combining geotechnical, instrumentation and computational methods. For the purpose of seismicity microzonation, seismic intensity is estimated in points as per SNIP II-7-81* or determined as seismic loads shown by estimated or real accelerograms, i.e. curves showing how vibrations of soil layers are accelerated during strong earthquakes. To assess potential seismic hazard, it is needed to take into account the intensity and other parameters of elastic vibrations under the base structures of buildings and facilities and consider the manifestations of inelastic strain and residual deformation of soil layers. Ranges of elastic vibrations of the soil layers are recordable by direct instrumental observations conducted in the study area.
Comprehensive studies can provide source data for dividing the study area into zones which seismic intensity may differ by ±1-2 points, and forecasts for each zone can be adjusted with regard to site-specific tectonic, geological and geomorphological conditions. Calculations of incremental points against the baseline seismic intensity are significantly influenced by data on groundwater levels and lithological compositions of rocks and soils. Such calculations are also impacted by significant variations in the intensity of the seismic field due to heterogeneities in the bedrock to a depth comparable to the wavelength (up to 1 km). Should any sudden change take place in geological conditions while new construction activities are performed, the relevant seismic microzonation data should be revised and updated accordingly.
In this article, we present results of the initial stage of engineering seismological studies in the Ulan-Ude city territory and consider possibilities of zonation by the main parameters of seismic impacts of potential strong earthquakes in order to identify potential seismic hazard areas in compliance with the current regulatory requirements concerning urban construction. A technique for construction of a new seismicity microzonation map of the Ulan-Ude city territory is justified.

APPLICATION OF INDIRECT SEISMICITY MICROZONATION METHODS
Generally, seismic hazard assessment is based on results of the acoustic (seismic) impedance method, data from catalogues of recorded earthquakes and microseisms, and data obtained by computational methods. Herein we briefly describe our technique of measurements, present estimations of seismic parameters and describe the rocks and soils that dominate in the study area.
The seismic impedance method [Guidelines…, 1985[Guidelines…, , 1986Medvedev, 1962;RSN 60-86, 1986;Pavlov, 1984]. Incremental points are calculated from the equation published in [Medvedev, 1962]: where I is estimated value of incremental points; ρ э V э and ρ i V i is seismic impedance of the reference soil and the studied soil for P-/S-waves, Vp/Vs; h is groundwater level; coefficient R=1 is accepted for areas with dominant sandy and clayey soils, and R=0.5 for areas with dominant gravel-pebble and coarsely clastic rocks. If the groundwater level is at a depth below 10m from the ground surface, the correction coefficient is close to zero. In order to calculate the seismic hazard in points and then to estimate it in terms of maximum acceleration, the following data are needed: rock and soil composition, velocity of seismic wave propagation in rocks and soils, thickness and composition of unconsolidated soil layers, and bulk weight of the reference rocks and soils and the studied rocks and soils [Guidelines..., 2004;Pavlov..., 1988]. Therefore, the top section of the profile of rocks and soils to the bedrock needs to be characterised to correctly select locations of measurements and then to properly analyse the measurement results. A general description of the top section is presented herein at a level sufficient to support the first stage of our studies aimed at seismic microzonation mapping of the Ulan-Ude city territory. In the regional Quaternary deposits, facies are diverse, and compositions of rocks and soils are variable. On the left-bank floodplain terrace of the Selenga River ( Fig. 1), powdery fine-grained alluvial sands are alternating with small lenses of sandy loam and clay. The sand beds are 1.0 to 5.0 m thick and underlain by gravel. Groundwater occurs at depths ranging from 1 to 3 m. The left bank of the Uda River is composed of eolian fine-grained sand, and the sand beds vary in thickness from 10 to 15 m along the river and 50 to 80 m closer to the slope. Groundwater occurs at depths of 5-10 m and 50-60 m. Bedrocks are represented by the Jurassic-Cretaceous sandstone, argillite and granitic rocks.
The right bank of the Selenga River comprises a thick bed of conglomerates with sandstone interlayers that are overlain by either gravelly soil or fine-grained sands (1.5-3.0 m and 10-15 m thick beds, respectively). The groundwater table is deep-seated.
In terms of geomorphology, the terrain of the Ulan-Ude area is significantly rough. In the north, spurs of the Ulan Burgasy ridge are low, and hills are cut by ravines and gullies and located almost perpendicular to the valleys of the Uda and Selenga Rivers. In the south, spurs of the Tsagan-Daban ridge come to the Uda River valley.
Several terraces are recognized in the valleys of the Uda and Selenga Rivers: Terrace 1 is 2 to 4 m high (Ulan-Ude downtown), Terrace 2 is 10 to 20 m high (the Soviet and Oktyabrsky districts of the city), and Terrace 3 is 40 to 50 m high (the Zheleznodorozhny district and a part of the Oktyabrsky district).
Seismic wave velocities were measured by a LAKKOLIT digital 24-channel engineering seismic station made in Russia. The refraction method was applied as described in [Seismic Surveying, 1981]. Measurements were carried out in separate sounding sessions, and reverse and catch-up time-distance plots (46, 92 and 150 m) were provided. Geophones were spaced by 2, 4 and 6 m (in the downtown, the distance was 12 m). Seismic waves were generated by shocks. Recording was done under observation schemes ZZ and YY corresponding to vertically oriented geophones and horizontal shocks perpendicular to the profile, with receivers oriented in the same direction. The selected measurement technique made it possible to obtain average values of seismic wave velocities for the top zone of the profile to depths from 10 to 30 m. It should be noted that the detection of 'useful' waves was challenging due to considerable background noise, and notwithstanding the accumulation of shocks, the detection of transverse waves was supported by data on surface waves.
In the city territory, velocities of P-and S-waves were measured at 37 locations assumed to cover all of the areas distinguished by the available geotechnical data. In the 'reference' bedrocks, Vp and Vs were measured in the city territory and in the vicinity of the city (measurements were taken in quarries and on sites where the bedrocks occur at shallow depths). At 23 locations (Nos. 38 to 60), special measurements were taken in order to design seismicity-soil models corresponding to zones in the city which may be subject to the highest and lowest seismic hazard. Such models also provided information complementing to the measurement statistics. Reflection seismic data processing was performed by the RadExPro software.
Examples of the recorded seismograms are given in Fig. 2. Time-distance plots of P-and S-waves and seismic wave velocity profiles for sites that meet the specified seismicity-soil conditions are given in Fig. 3. The seismograms, plots and profiles give evidence that it is challenging to select 'useful' waves when seismic measurements are taken in urban areas, even if special attention is given to registration timelines, the amount of accumulated excitations and their intensity. Data from all the seismic measurement locations (see Fig. 1) were consolidated, and histograms were constructed to show the distribution of wave velocities and reveal most probable values (Fig. 4). However, the available histograms are limited in number, and additional measurements are required for each type of soil.
In general, it is evidenced by the seismic velocities recorded in the top zone of the profile near the city of Ulan-Ude that the seismic velocity values differ dramatically in ground conditions of three types -rocky, water-saturated soil and unconsolidated non-water-saturated soil. The bedrocks are represented mainly by conglomerate, sandstone, argillite and granitic rocks. Velocities of P-and S-waves in these rocks are low in the top zone of the profile (to depth from 3 to 5m). In the uppermost zone, the P-wave velocity range from 1400 to 2000 m/sec. In less fractured rocks, Vp values range from 1500 to 3500 m/sec (Fig. 4, a) and Vs values range from 1000 to 2100 m/sec with increasing depth (Table  1). According to measurement in 50 bedrock samples 4-5 -unconsolidated none-water-saturated soil; 20-21 -water-saturated gravel-pebble sediments; 8-9 -strongly and weakly fractured rocky soils; 24-25 -unconsolidated, broken rocks and relatively intact bedrocks. Top and bottom numbers show velocities of P-and S-waves, respectively.
For water-saturated gravel-pebble and sandy soils, the typical velocities of P-waves range from 1650 to 2000 m/sec (Fig. 4,c), and the P/S-wave velocity ratio varies from 3.0 to 4.5 (Table 1). Measurements in soils of the same type but not water-saturated show P-wave velocities from 400-500 to 800 m/sec and S-wave velocities from 180 to 420 m/sec (note: the layer of seasonal freezing was excluded from the calculations) (Fig. 4, d). The available data on physical pro-perties of soils which are required for further calculations are summarized in Table 1.
Data on soil composition, velocities of seismic wave propagation in soils of the specified types, thickness of unconsolidated sediments in the upper segment of the profile, and bulk weight of the reference and studied soils were collected as required for seismic hazard assessment, construction of the set of seismic models and application of the selected computational methods (see Section 2). Equation (1) was used to estimate values of incremental seismic intensity for each observation location. The average velocity value estimated for the top 10-meter thick zone was taken into account. Calculation results are given in Table 2. Рис. 4. Гистограммы распределения скоростей продольных волн в грунтах района г. Улан-Удэ: a -для скальных грунтов; b -в образцах скальных пород; c -для обводненных грунтов; d -для воздушно-сухих грунтов.
T a b l e 1. Physical properties of rocks and soils Т а б л и ц а 1. Физические свойства грунтов Rocks  In our calculations, the reference is the rocky soil with Vp=2200 m/sec, Vs=1200 m/sec and ρ=2.5 g/cm 3 (average velocities in the 10-metre thick layer on sites with bedrock outcrops). For sites with the abovementioned values, the seismic hazard is assumed one point lower than the baseline level. In this case, the average soil type (i.e. non-water-saturated sand and gravel-pebble) is characterised by Vp=600 m/sec, Vs=300 m/sec and ρ=1.8 g/cm 3 . In zones with the above-described soil, the seismic hazard corresponds to the baseline seismic intensity.
Our measurements of seismic wave velocities in the city of Ulan-Ude and calculations of incremental seismic intensity (Table 2) show that relative to the selected reference soil (rocky soil -7 points), the rocky/soil foundations of buildings and facilities may be subject to an incremental seismic impact (+0.17 to +2.3 points), and the seismic hazard for the rocks and soils ranges from 7.17 to 9.3 points.

FOUNDATIONS IN THE ULAN-UDE CITY TERRITORY
To solve the problem related to seismicity microzonation mapping in compliance with the current regulations concerning urban construction, the seismic ha-zard of rocky/soil foundations should be mapped with account of the maximum seismic wave acceleration, dominant periods of strong earthquakes, resonance frequencies of unconsolidated beds and other characteristics of the seismic impacts.
To provide a basis for seismicity microzonation mapping of the Ulan-Ude city territory, quantitative data on soil movements of ground are needed. In the current stage of our studies, we analyze dynamic characteristics of perceptible earthquakes that occurred in the study region, establish seismic signals corresponding to the baseline seismic intensity, develop the seismicity-soil models, try to forecast seismic impacts with regard to different construction conditions and classify zones in the Ulan-Ude city territory by the seismic impact parameters. To achieve the objectives, modeling and computer simulation methods are applied.
The priority task is to establish the baseline seismic signal for the Ulan-Ude city territory [Dzhurik, 2014]. Determining a 'baseline' seismic impact is challenging as a reference accelerogram cannot be unambiguously selected. The unambiguity is due to the fact that an earthquake can be manifested in different ways in particular local zones, depending on characteristics of the earthquake source, seismic signal propagation track, structures and compositions of rocky/soil foundations of buildings and facilities. Besides, it is needed to take into account a number of complicating factors, such as E n d o f T a b l e 2 О к о н ч а н и е т а б л и ц ы 2 several potential earthquake foci zones (Fig. 5), physical and mechanical properties of rocks and soils, and types of displacement/movement in earthquake foci areas. Such factors predetermine whether an impulsetype seismic event may occur or an earthquake with a relatively slow increase and decrease of seismic intensity on the surface may take place. Methods for selecting the baseline accelerograms are mainly oriented at the acquisition or calculation of peak accelerations and scaling [Pavlov, 1988] in accordance with relevant seismic scales [Nazarov, Shebalin, 1975]. Due to the fact that registered strong seismic events are not numerous in potential earthquake foci zones (and also in the vast regions under review), it becomes necessary to refer to data from catalogues of strong earthquakes registered by the global seismic network or use data on small earthquakes and establish phase characteristics [RB-006-98, 1998].
In this study, we use only the earthquakes records by the regional network of seismic stations . Since the medium is considered as a formgenerating factor of a focal pulse, the phase spectrum of local earthquakes, one way or another, takes into account the earthquake excitation and scattering properties of the inhomogeneous medium. In engineering surveys for construction purposes, the earthquake resistance of buildings and facilities is typically calculated from accelerograms [Ratnikova, 1984]. It is advisable to obtain accelerograms for each PEF zone that can be described by sets of average seismic characteristics.
In view of the above, our study has two main objectives: (1) Obtain potential earthquake accelerograms for each PEF zone with the reference to the available accelerograms of earthquakes that actually took place in the studied zones (for three components, NS, EW and Z); (2) Using the properly grounded models showing seismicity of the 'reference' rocks, correlate the ob-  Table 3) of potential danger for the Ulan-Ude city territory. Circles show earthquake epicentres selected for establishing the reference signals; triangles show locations of permanent seismic stations.
In our study, to justify the seismic hazard of the Ulan-Ude city territory, we analyze accelerograms of actual earthquakes (M from 3.0 to 6.3) recorded by the Ulan-Ude seismic station. For each PEF zone, an average spectrum is calculated and taken as a characteristic of the entire zone. In total, the processed database includes records of 55 earthquakes from 2001 to 2011. Data on components NS, EW, and Z are processed separately. Some of the recorded accelerograms are rejected due to various reasons, such as an insignificant signal/noise ratio. For each earthquake, amplitude and phase spectra are calculated.
It should be noted that we select accelerograms of earthquakes differing in magnitudes and thus refer to relation  M (f) showing how the acceleration spectrum changes with magnitude variations and depends on frequency. In our study, we use the equation for the Baikal rift zone which was published in [Drennov et al., 2013]: Based on the above relationship, the amplitude spectra are scaled to magnitudes of the PEF zones. Finally, earthquake accelerograms are obtained for the PEF zones characterized by their specific magnitudes (Fig. 5). This objective is met by using the inverse Fourier transform of the average acceleration spectrum for a specified PEF zone and the phase spectrum of the strongest earthquake recorded in the given PEF zone.
Based on the phase spectra of accelerations from various earthquakes, it is possible to obtain accelerograms of different durations, from accelerograms of the impulse type (when the released energy is concentrated in a small time window) to accelerograms of large time spans. In our study, phase spectra of accelerograms of the medium time span are mainly used.
The accelerograms and their spectra for all the studied PEF zones are shown in Fig. 6 and 7, and the corresponding spectral parameters are given in (Table 3).
According to Table 3, maximum and minimum acceleration rates for the rocks and soils under the Ulan-Ude seismic station can amount to 166 cm/sec 2 and 1.7 cm/sec 2 , respectively (the three components are taken into account). The maximum acceleration rates are associated with frequencies from 1.2 to 8.3 Hz, the widths of the acceleration spectra for all the PEF zones range from 0.6 to 14.4 Hz at the level of 0.7S m . Typically, for the PEF zones located closer and having larger potential magnitudes, the acceleration spectra are somewhat wider than those of the more remote PEF zone, and this expansion is due to higher frequencies. Besides, the maximum values of the spectra are widely variable (from 0.2 to 46 cm/sec) and correlate with the frequency range from 1 to 12.3 Hz.
It is revealed that ranges of maximum values of the studied parameters are widely variable for each component. Therefore, it is needed to conduct additional studies to eliminate the uncertainties. This problem can be solved by long-term recording of earthquakes on various sites in the city which have contrasting soil conditions, such as water-saturated or air-dry soils of specific compositions. In view of the above, at the current stage of our studies, we refer to relatively reliable analyses of seismic impacts by the frequency of their occurrence. Considering amplitudes, it is needed to scale the established baseline signals with regard to forecasted seismic impacts. Such objectives comply with requirement of the current construction regulations. However, a probability of establishing the maximum amplitudes can be properly justified by conducting the required comprehensive studies and consolidating the modeling and experimental data.
At the Ulan-Ude seismic station, geophones are located on soils of category 1. Such conditions may prove sufficient for seismological reconstructions; however, in order to solve problems of engineering seismology, we need quantitative data, including, in the first place, determinations of frequency response which (as a transfer function) are required to justify the baseline signals of the 'reference' rocks/medium soils represented in the seismicity-soil models. This objective can be achieved, as noted above, by direct and computational methods of seismicity microzonation [Dzhurik et al., 2012].
Thus, for further use of the accelerograms (see Fig. 6 and 7), they are assigned to the soils of category 1. For the Ulan-Ude city territory, a single baseline signal needs to be established. A mandatory condition is that it should take into account specific features of the spectral compositions of vibrations for each selected PEF zone (see Fig. 5). The vibration spectra are normalized and then averaged. A phase response of one of the earthquakes recorded is estimated, and normalized accelerograms are calculated by the inverse Fourier transform for the three components (Fig. 8, a). In its turn (Fig. 8, b), the amplitude spectrum of this signal reflects all the specific frequency characteristics of the accelerograms predicted for the PEF zones of the highest hazard (Nos. 3, 4 and 5). Its level exceeding 0.7 Smax is in the frequency range from 1.2 to 5.0 Hz. The main peaks of the spectra occur at frequencies from 1.6 to 2.2 Hz (Table 4).
For further theoretical calculations considering different soil conditions presented by the seismic models, it is required to correlate the background seismic signal with a reference soil/half-space, from which we can estimate changes in the signal by near-surface inhomogeneities. In further estimations, it is also reasonable to consider inhomogeneities located at depth.
To develop models that can characterize subsurface inhomogeneities, we use data on seismic wave velocities that are generalized with regard to soil compositions and conditions (see Table 2; Fig. 3 and 4) and also refer to the available file materials. Parameters of seismicity-soil reference models Nos. 1 and 2 (Table 5) are based on the above-mentioned data and correlated with predicted seismic impacts.
A set of the well-known methods and software packages [Ratnikova, 1984;RB-006-98, 1998] is used to carry out theoretical calculations.
At the current stage of studies, for reference model No. 2 (see Table 5, Fig. 9) representing the 10-metre thick water-saturated soil of the medium composition, the acceleration rates are scaled with regard to the acceleration rates to 397 cm/sec 2 and 173 cm/sec 2 for the maximum and vertical components, respectively. This corresponds to the seismic hazard of 9 points, i.e. one point higher than the reference level for the nonwater-saturated soil. The resonant frequency amounts to 12.79 Hz; the main peak of the spectrum is at the frequency of 1.56 and 1.51 Hz; the maximum spectral density amounts to 85.3 and 51.9 cm/sec for components EW and Z, respectively (Table 6).
It is noteworthy that in further studies, special attention should be paid to the justification of the potential seismic hazard of water-saturated soils  based on records of the behaviour of such soils during earthquakes. As noted earlier, frequency characteristics need to be determined for the watersaturated soil layers varying in thickness, and such data can facilitate achieving more reliable results by the calculation methods.
In general, models Nos. 3, 4 and 6 represent the unconsolidated non-water-saturated soils varying in thickness. According to estimations by the seismic T a b l e 3. Main parameters of estimated spectra for potential earthquake foci (REF) zones Т а б л и ц а 3. Основные параметры расчетных спектров для землетрясений разных зон ВОЗ N o t e. М is magnitude; Δ is distance to epicentre; am is maximum amplitude of calculated acceleration; fam is frequency corresponding to am; Sm is maximum amplitude level of the spectrum; fSm is spectrum peak frequency; f1 and f2 are frequencies that limit acceleration spectra at levels 0.7 and 0.5 Sm; f0.7Sm and f0.5Sm are spectrum widths.
Based on the theoretical calculation results (see Section 2), experimental measurements (see Section 1) and the available geotechnical and hydrogeological data, a schematic map is compiled in the first approxima-tion ( Fig. 10) to show zones differing in the basic seismic impact parameters.
The zone with the potential maximum seismic hazard of 9 points includes floodplain areas and the first above-floodplain terrace composed by alluvium (sand, clay soil and gravel) where groundwater occurs at shallow depths, less than 5m. It is possible that the weakened near-fault northern site will be also included. This zone can be represented by seismicity-soil models No. 2 and No. 7 (see Fig. 9 and Table 6). In this zone, the maximum acceleration rates are 410 cm/sec 2 and 223 cm/sec 2 for components NS and Z, respectively.
The zone with the relatively high seismic hazard (8 and 9 points, a transition zone) includes the left-bank terrace of the Uda River which is composed of silty fine-grained sand (models No. 4 and No. 5). In this zone, the maximum acceleration rates range from 154 to 410 cm/sec 2 and 64 to 223 cm/sec 2 for components NS and Z, respectively.
The seismic hazard of 8 points (models No. 3 and No. 4) may be expected at slightly sloped terraces of the Uda and Selenga Rivers where groundwater occurs at depths from 8 to 20 m. In this zone, the maximum acceleration rates are 186 cm/sec 2 and 86 cm/sec 2 for components NS and Z, respectively.
E n d o f T a b l e 5 О к о н ч а н и е т а б л и ц ы 5 N o t e. The seismic impedance method is applied to calculate seismic hazard levels (UoM -point) with respect to the bedrocks (baseline seismicity of 7 points). Average seismic wave velocities are estimated for the 10-metre thick layer. The watercut correction is +1 point.
The zone with the seismic hazard of 7 points includes sites composed of rocky and semi-rocky soils, except areas of tectonic fracturing (model no. 1). In this zone, the maximum acceleration rates are 98 cm/sec 2 and 53 cm/sec 2 for components NS and Z, respectively.
It should be noted that by applying the seismic mi-crozonation method, it is possible to reveal sites with the seismic hazard from 7 to 9 points in each of the specified zones (Fig. 10). Such sites can be correlated with relevant soil models (see Table 5), and their locations can be further clarified and determined more precisely with regard to the scale of seismic microzona- Рис. 9. Акселерограммы (a) и их амплитудные спектры (b) для горизонтальной компоненты, акселерограммы (c) и их амплитудные спектры (d) для вертикальной компоненты, частотные характеристики рыхлых слоев (e).
tion of the Ulan-Ude city territory by developing more detailed models with the application of the proposed technique.

CONCLUSION
The seismic hazard zonation of the Ulan-Ude city territory is a complex problem including studies by the seismic, seismotectonic, geotechnical and seismological methods. Each of the methods solves specific research problems, and their combination provides data for achieving the major objective to construct a seismic microzonation map of the territory.
At the current stage of studies, the indirect instrumental methods of seismic microzonation are used, and the types of rocks and soils prevailing in the studied territory are determined and classified by the propagation patterns of P-and S-waves. Using the acoustic impedance method, we estimate the incremental seismic intensity values for water-saturated and non-watersaturated sandy gravel-pebble sediments. The calculations are performed against parameters of the selected reference soil, i.e. the rocky soil with average seismic wave velocities in the upper 10-metre thick layer.
At the initial stage of forecasting how strong earthquakes can impact the rocks and soils under buildings and facilities in the Ulan-Ude city territory, we refer to the main parameters of significant ground movements that occurred in the Baikal rift zone in the past ten years. The established baseline seismic signal takes into account, in the first approximation, the main parameters of the potential earthquake occurrence zones and the previously established empirical relationships showing how the main dynamic characteristics of soil acceleration can vary depending on seismic event magnitudes and distances from earthquake foci. Based on such data, accelerograms can be forecasted for different epicentral distances and magnitudes and used for more reliable determinations of baseline seismic signals for the Ulan-Ude city territory with reference to the frequency characteristics.
It is shown that the established baseline signal is applicable, and the theoretical calculations are conducted on the basis of the seismicity-soil models characterizing bedrocks, medium soils and water-saturated soils (soil categories 1, 2 and 3, respectively). Based on the calculation results and the available geotechnical and hydrogeological data, and taking into account the soil and hydrogeological conditions for construction (the baseline seismic intensity of 8 points), a schematic map of seismic hazard is constructed for the Ulan-Ude territory in the scale sufficient for construction purposes. It shows that the seismic hazard is variable from 7 to 9 points through the studied territory. The map in the current format was used when we developed a detailed programme of studies aimed at seismic microzonation.
Obviously, the mapped data will be revised and updated in a more detail. Anyway, the obtained results can be useful today for planning possible construction sites in the Ulan-Ude city territory.
The technique of seismic microzonation mapping should be based on detailed measurements, and parameters for mapping the impact of seismic events should be determined at the precision level no less than that specified in requirements to engineering and design of earthquake-resistant buildings and facilities. It is recommended to apply GIS technologies and conduct more detailed engineering and seismic measurements in order to consolidate a database for construction of more detailed maps and schemes of the studied territory in smaller scales. In order to construct a digital map of seismic microzonation, the source materials should include topographical and special geotechnical and hydrogeological maps and schemes showing thickness of unconsolidated sediments, as well as various reference materials and data from other sources.
To complete seismic microzonation mapping, it is required to identify the potential seismic sources and characterize them at the new probabilistic level. It is thus needed to determine locations subject to deformation and active faulting, estimate the periods of earthquake recurrence, determine seismic intensity levels, and reveal probabilities of potential earthquake occurrence. Information on the potential seismic sources can facilitate forecasting of strong events for the Ulan-Ude city territory. Seismic wave propagation parameters and potential seismic impacts should be estimated for specific construction sites located in the city, and such estimations should be followed by calculations of response spectra and associated probabilities of the occurrence of strong earthquakes. Once the above-mentioned detailed data are consolidated, it will become possible to construct a map of seismic risks that can serve as a useful source of information for streamlining the regional construction policy.