NYCEM 1st -Year Technical Report
May 6, 1999:

Site Conditions Effecting Earthquake Loss Estimates for New York City

by

Klaus H. Jacob
Lamont-Doherty Earth Observatory
of Columbia University,

PO Box 1000;
Palisades NY 10964 - 8000
Phone: 914-365-8440;
Fax: 914-365-8150;
email: jacob@ldeo.columbia.edu

 

Table of Contents

Executive Summary
Technical Report
  1. Project Objectives
  2. Issues
  3. Technical Tasks
    3.1 Pilot Bedrock Map
    3.2 Geotechnical Boring Data
    3.3 Blowcount-Velocity-Depth Relation
    3.4 Average Shear Wave Velocity vs. Depth
    3.5 NEHRP Site Classes
    3.6 Results: Site Classes and Census Tracts
    3.7 Site Class Distribution
    3.8 Site Classes: V vs. Blowcounts
    3.9 Site and Building Resonance Periods
    3.10 Scenario Earthquakes and Loss Estimates
  4. Insights and Recommendations
    4.1 Availability of Borings
    4.2 Calibration Data
    4.3 Filling in Data Gaps
    4.4 Verifying Site Resonances
    4.5 Assumptions about Rock Velocities
    4.6 Interpretations of NEHRP Site Class Definitions
    4.7 Scope of Future Efforts
    4.8 HAZUS, Mitigation Strategies, and Beyond
References
Acknowledgments
Appendix A: NEHRP Definitions of Site Classes/Amplification Factors
List of Figures
List of Tables

 

List of Figures

Figures will open in a new window
Figure 1: Trial version of a bedrock elevation map of Manhattan
Figure 2: Shear velocity versus Standard Penetration Test blowcount and depth
Figure 3: Shear wave velocity as a function of depth
Figure 4: NEHRP site classes from based on V₁₀₀ from NYC DDC boring files
Figure 5: NEHRP site classes for borings to bedrock with no soil information
Figure 6: NEHRP site classes in lower Manhattan for all borings
Figure 7: NEHRP site classes assigned to individual census tracts
Figure 8: Ground motion for a scenario earthquake with magnitude Mw=6.0
Figure 9: Ground motion S_a(0.3s) for M=6 scenario for default site class D
Figure 10: Ground motion S_a(0.3s) for M=6 scenario with variable site classes
Figure 11: Building-Related Economic Losses S of 59th Street for M=6
Figure 12: Motions, $-Losses and Loss Ratio vs. Mw for default soils D
Figure 13: Motions, $-Losses and Loss Ratio vs. Mw for variable site classes
Figure 14: Ratio of HAZUS results for modified vs. default site class D

 

List of Tables

Table 1: Assigned velocities for materials differing from regular soils
Table 2: Definition of NEHRP Site Classes by V₁₀₀ and N₁₀₀

Executive Summary

Project Objectives: The New York City Area Consortium for Earthquake Loss Mitigation (NYCEM) focuses on obtaining quantitative loss estimates for earthquakes in the NYC metropolitan region. Such loss estimates are expected to be useful tools for developing effective earthquake risk mitigation strategies. This first-year pilot study is geographically limited to the southern half of Manhattan, south of Central Park (from 59th Street to Battery Park). For this pilot study we use geotechnical data consisting of standard penetration test (SPT) blow counts and standard soil descriptions from construction-related soil borings in New York City to assess the effect of near-surface geology on seismic ground-motion site-response (microzonation). In addition, information on depth to bedrock from older borings which do not contain information on the type of penetrated soils are also used for microzoning the shaking effects. The general objectives of seismic microzonation are generally to quantify urban seismic hazard and the risk (loss potential) by accounting for the local variations in shaking levels due to near-surface geological differences. These differences can influence the amplitudes and the spectral content of ground motions and, thereby, the shaking levels of buildings and lifelines, and hence control the expected losses. The specific goals for this pilot study are narrower than what is generally attempted during microzonation. We limit this pilot study to determining the site classes A through E as defined in the 1997-Edition of the NEHRP Provisions for Seismic Regulations for Buildings. Knowledge of these site classes is required as input into the HAZUS earthquake loss estimation algorithm. Spatial resolution for HAZUS requires that a single site category be assigned to each census tract (typically a few street blocks in dimension).

Data and Analysis Method: The analog information of a total of 150 geotechnical borings in lower Manhattan (below 59th Street) was entered into digital spread sheets. Data included casing and SPT blow counts and standardized descriptions of the stratigraphic materials encountered as a function of depth. We use a locally derived relation Vs(ft/s)=(220+3N)D^0.3 (Equ. 1) to translate the measured SPT blow counts into shear wave velocity versus depth profiles. Equation (1) is based on the few locally available calibration borings relating SPT counts N to shear wave velocity Vs and depth D(ft). The Vs (shear wave velocity) depth profiles were then used to determine for each boring site the NEHRP site classes. At more than 200 older boring sites where only depth to bedrock but no soil properties are known, we proceeded differently. First we derived an average relation of shear wave velocity vs. depth for lower Manhattan from a subset of borings using the N to Vs conversion described in Equation (1). Only about 50 of the 150 available geotechnical soil borings, i.e. those which reach bedrock were used for this purpose. For these borings, the nominal average relation for shear velocity vs. depth for soils in southern Manhattan was found to be: Vs(ft/s) = 435.11+11.08D(ft) (Equ. 2), whereby any figures after the decimal period are insignificant. In fact, the scatter is very large (about a factor 2 up and down from the mean) since this average relation does not attempt to differentiate between the different soil materials and densities encountered. Then, at the more than 200 bedrock boring sites without soil information, the average relation (2) was used as nominal velocity input for the soil profiles down to bedrock or 100 feet, which ever depth is less. For sites with soil profiles less than 100 ft thick we used an average Vs for rock of 5,000 ft/s for the remaining depth interval in rock down to 100 ft below grade.

Results: Since in Manhattan south of 59th Street (as opposed to "upper" Manhattan) bedrock is rarely near the surface, the majority of sites turned out to be D or C sites (in NEHRP site classification terminology). This reflects the fact that most sediments in lower Manhattan are quite deep and are soft because they are predominantly of post-glacial Holocene and Recent estuarine origin (organic clays, silts and fine sands). Often they directly overlie very hard meta-sedimentary to crystalline rocks, with medium-hard sediments rarely present. No sites belonged to the stiffest site class A (very hard rock), but a fair number to B (firm rock), and several to class E with very deep soft soils. When using the newly derived soil types rather than the D-type default values, and when averaged over all of Manhattan south of 59th Street, the total loss estimates obtained from HAZUS for buildings were reduced by factors varying from 0.7, to 0.72 to 0.8 for scenario events with magnitudes Mw = 5.0, 6.0, and 7.0, respectively, at a common epicentral distance of 20km. This trend of diminished loss reduction with increased magnitude reflects the non-linear behavior of soil amplification with increasing ground motion amplitudes that is built into the NEHRP’97 soil amplification factors for ground motions.

Discussion: Problems to determine site classes with sufficient spatial resolution are caused by uneven distribution of borings both in depth and laterally. Not every census tract has a representative boring; conversely, some census tracts were too large and multiple borings indicated that assignment of a single site class to the tract is an oversimplification and in conflict with the fact that sectors of some census tracts should be assigned to different site classes. To satisfy immediate needs for site class input into the HAZUS loss estimation program, we used for the time being visual interpolation and extrapolation to fill in data gaps, and smoothed and averaged site classes where the information was too fine-grained for a single tract. Future microzonation efforts and maps may not be bound by the coarse spatial resolution imposed by the patterns of census tracts that HAZUS currently uses, but instead they may pursue a more geological approach to use the geotechnical information at the best spatial resolution extractable from the available data. The pilot study also looked into the options and difficulties for producing (for the first time in NYC) digital maps of bedrock (subsurface) elevation contours and of depth to bedrock below grade. It became clear that sufficient original (analog) information appears to be available, albeit conflicting for some areas when using different sources. A methodology for smoothing the sparse discrete digital information on bedrock topography and depth was successfully tested on a subset of the available data. It quickly became apparent, however, that the task to produce useful, optimally resolved comprehensive digital bedrock maps is too labor intensive to be consummated within the scope of a pilot study. We infer from this pilot study that -given sufficient resources to perform a truly comprehensive study- the geotechnical data available in New York City are suitable and of sufficient quality and quantity to be readily translated into a useful microzonation product for a quantitative assessment of seismic site response. A compilation of the 3-D near- and subsurface geology of New York City which is riddled with underground structures and life lines has obviously additional, potentially much broader application potentials. These may include urban and land use planning, early planning stages of major engineering projects, and maintenance/operation of existing underground lifeline structures. In this sense, many organizational entities in New York City, including utilities, planners, operators and decision makers, are natural stakeholders in a thorough inventory and assessment of the 3-D geotechnical properties beneath the largest and densest concentration of people and structures in the US. Concerns for reducing future seismic losses in this metropolis will therefore contribute tools, methods and products that almost certainly will allow to better manage this mega-city’s future and sustainability, and do so in a much broader context than just earthquakes.

Key Words: Seismic Site Response, Microzonation, Geotechnical Borings, Standard Penetration Test (SPT), NEHRP Site Classes, Bedrock Subsurface Topography, Urban Hazard, Risk, Mitigation, Land Use, Earthquake Loss Estimation, HAZUS, New York City, Manhattan.

Technical Report

1. Project Objectives:

A pilot study has been performed in which geotechnical data (standard penetration test -SPT- blow counts, and standard soil descriptions) from existing, construction-related soil borings in the Borough of Manhattan, New York City, are used and analyzed for their implications for seismic ground-motion site-response. The goal is to better quantify urban seismic hazard and risk (loss potential) by accounting for the variation in shaking levels due to near-surface geological differences. While the ultimate goal is to capture the site response in the spectral domain and relate it to the spectral (modal) response of structures due to building height and construction type, the goals for this pilot study were for the most part much simpler: determination of the site classes A through E as defined in the 1997-Edition of the NEHRP Provisions for Seismic Regulations for Buildings (FEMA, 1998a &b). These site classes are needed for use in the HAZUS loss estimation algorithm (NIBS, 1997). The spatial resolution of site class zonation required for HAZUS loss estimates is based on census tracts, of which there are about 300 for all of Manhattan. Tract size varies, but on average has (in NYC) linear dimensions on the order of 1 to 3 street blocks in length forming generally quadrilaterals or prismatic tract areas. The pilot study was limited to the southern half of Manhattan, south of Central Park (from 59th Street to Battery Park at the southern tip of Manhattan).

2. Issues:

The following difficulties had to be overcome:

1. The NEHRP provisions require either SPT blow counts and/or shear wave velocities to depth of 100 feet (30 m) below grade; but not in all census tracts do borings exist that reach to depths of 100 feet. In these cases one must infer the velocities between the bottom of the boreholes and the 100 feet depth contour from regional average soil and rock properties vs. depth.
2. Borings are not available for all census tracts, hence site class must be inferred or interpolated from information of sites located nearest the tracts that lack borings.
3. Soil conditions from multiple borings in a single census tract can in some areas vary laterally so strongly that it is sometimes difficult to assign a single, dominant site class to a census tract; i.e. the granularity of the census tracts is at some locations -but not everywhere in the study region- too course with respect to the spatial variations in geologic site conditions. 4) Only few shear wave velocity measurements are available locally to calibrate the relation between SPT blow counts, shear wave velocity and depth.

 

3. Technical Tasks:

The pilot study tackled the following tasks and achieved the following results so far:

3.1 Pilot Bedrock Map

More than 300 points of depth to bedrock have been entered into a digital data base. The points are spatially clustering because of construction-related biases. A preliminary sub-surface bedrock elevation contour map (Fig.1) has been obtained from these sparse non-uniform data by fitting a smoothed surface to the discrete bedrock depth information using a spline fitting technique from the GMT generic mapping tool box (Wessel and Smith, 1995). Depth to bedrock varies locally in the glacial, steeply carved and backfilled environment from near zero to about 240 feet (˜80m), but smoothed values typically do not exceed 170 feet. In Manhattan alone there may be, in principle, as many as 3000 points for depth to bedrock available in various analog data and historic map files, of which, however, only a small portion contains useful information on the soils overlying the bedrock. The inferred bedrock elevations were contoured by hand in the 1930’s as part of a labor-intensive public works project (e.g. Fluhr and Murphy, 1944). Baskerville (1990, 1992, 1994) used only a subset of those data for contouring the elevation of the top of basement rocks. In fact we find some discrepancies of his mapped values with some of the boring data. The various data sources need to be checked and recreated from dispersed and diverse paper and map files before they can be fully utilized in future digital refinements of, and follow-ups to this pilot study. The analog-to-digital transfer of data is labor intensive. But the technique employed by us appears promising and effective and we believe that with sufficient resources a fairly accurate bedrock elevation contour map in digital form can be assembled in a reasonably short time.

3.2 Geotechnical Boring Data and Data Analysis.

Most of the data used in this pilot study were selected from analog boring files archived at the NYC Department for Design and Construction (DDC). Our selection of borings attempted to maximize the spatial coverage using the smallest number of borings, yet resolving the spatial variability of soil and rock conditions to the extent possible for a pilot study. Hence some judgment had to be exercised on what constituted a most effective density of borings. The raw analog data entries from the selected boring logs were entered at the NYC DDC offices in Queens, NY, into Microsoft EXCEL spread sheets using a laptop. The files were transferred to a stationary PC workstation at LDEO for analysis and evaluation. Original blow counts were corrected for hammer weight, hammer drop distance, and casing size by the following relations (Budhu et al., 1989): First the hammer ratio Rs is computed by

Rs = (Do3- Di3) / 144 WH

(1)

where D_o and D_i are the outer and inner diameter of the sampling tube in inches, respectively; W is the hammer weight in lb.; and H is the hammer fall distance in inches. The field values of blow counts N are then modified to corrected blow counts N_e by the relation:

Ne = N / (4050 Rs5/7 )

(2)

 

3.3 Blowcount - Average Shear Wave Velocity - Depth Relation for Soils in NYC.

In the published literature there are many relations between blow counts and shear wave velocities. Some are material-dependent (clays, silts, sands), while others depend on depth or effective overburden pressure. After trying initially some of these published relations we learned that most did not match well against the local data available for calibration. There are only few in-situ measurements available for NYC where both shear wave velocity and blow counts had been independently determined as a function of depth. We used these measurements for calibrating average relations of Vs versus depth D and corrected SPT blow count N_e. We were able to obtain 6 such calibration profiles by courtesy of various engineering contractors. Using this sparse data set we found an approximate relation (Fig. 2):

Vs (ft/s) = (200+3 Ne) D0.3

(3)

with V_s in ft/s and D in ft. Note that this relation averages over all types of soils, from fills, organic and clean clays, to silts and sands and thus depends very much on the local average stratification and variability between these different materials as a function of depth. Hence this highly region-specific relation should not be readily transported to other regions unless they have glacial / post glacial / estuarine / urban sedimentation history similar to that of the lower Manhattan / New York City geologic setting. Equation (3) was used to translate the individual SPT N_e-values into shear wave velocities as a function of depth, in about 150 borings that were available for the pilot study. Most of the borings are sampled about every 5 feet in depth.

3.4 Average Relation for Shear Wave Velocity versus Depth.

Using the SPT-to-shear wave velocity converted data for borings in the depth range 0 to 240 feet, an average relation, for the study area, between V_s (ft/s) and depth D (ft) was established by linear regression (Fig. 3), yielding nominally

Vs (ft/s) = 435.11+ 11.08 D (ft)

(4)

It indicates at the depth of D=100 ft, which is needed for the NEHRP site class definition, an average shear wave velocity of V_s (100 ft) = 1543 ft/s (about 500m/s). This value was then used to linearly interpolate the soil velocities in shallow wells with depth D_b<100ft to the depth of 100ft using the equation

Vs (D)= Vs (Db) + {[ Vs (100 ft) - Vs (Db)] / [100ft - Db]} [D - Db]

(5)

where the velocity at the bottom of the borehole V_s (D_b) is obtained from equation (3) and the corresponding measured N_e-value. For the depth intervals where bedrock was inferred from nearby borings within the upper 100 feet (see item 3.1 above), we assumed a bedrock shear wave velocity of 5,000 ft/s (1.3km/s), consistent with NEHRP’97 hard-rock site class A which also assumes 5,000 ft/s. However, this velocity is only about 2/3 to 1/2 of what may be a typical shear wave velocity for the highly competent, glacially polished 1-Billion year old crystalline Manhattan Meta-schists and Fordham Gneisses. The lower value chosen to conform to NEHRP definition allows for fractured or altered rock at the soil-rock interface, including occasionally weathered Inwood Marbles. We will return to the implication of having chosen this rather low value compared to actual rock velocities.

In some instances the boring logs refer to materials for which no blow counts were given for some depth intervals. If the soils followed standard descriptive terms (e.g. fine sand, ...) but lacked the blow counts, then we estimated their average velocity according to depth D from Equation (4). To find the blow count N_e that corresponds to the velocity derived from Equ. (4), one can use the inverse of Equ. (3):

Ne= { [Vs (ft/s) / D0.3 ] - 200 } / 3

(3a)

where D is measured in feet. In other cases where no blow counts were given the borings had encountered "urban materials" with little resemblance to soils. The velocities that were assigned as first-order approximations to some of these materials with missing blow counts are summarized in Table 1.

Table 1: Assigned velocities for materials differing from regular soils, and for which blow count information was not available.

Material	Vs (ft/s)
Any materials where drill sank under own weight (Blow count too low to measure)	100
Primary Silt	300
Top Soil (in upper 10 feet)	300-400
Miscellaneous Fill	400
Steel Layer with Void underneath	500 (or eliminate if not pervasive)
Brick	500
Brick and Concrete	500-700
Asphalt	1000
Rip Rap	1000
Concrete	3000
Weathered Rock	3000
Boulders	5000
Hard Rock	5000

3.5 Rules for Assignment of NEHRP Site Classes.

Averages of both the corrected blow counts and the computed shear wave velocities were calculated; these averages are called N₁₀₀ and V₁₀₀ respectively, since only the uppermost 100 feet of any borehole is used to calculate N₁₀₀ and V₁₀₀ and then is used to determine the appropriate NEHRP site class. Many boreholes were shallower than 100 feet, and for these we used the data from the entire borehole available and filled in the rest as described below. The average shear wave velocities were calculated by dividing the distance d_i between each sample (the vertical distance through which shear waves travel at that particular velocity), by the calculated local shear wave velocity, to find the time it would take a wave to pass vertically through that interval of sediment. These time intervals were then summed, and the total distance traveled by the wave was divided by the total time necessary to travel the total vertical distance, to find the average velocity. More specifically:

VH = S di / S (di / vi )	(6a)
where H = S di = 100 ft	(6b)
and hence: V100 = 100 ft / S (di / vi )	(6c)

This "slowness-averaging" method tends to skew the final average more towards the slower velocities, which occur in soft soils, than to the high velocities found in stiff soil and bedrock. The slowness-and-depth averaged velocity so obtained is thus lower than that obtained from simply summing the average velocities and dividing them by the number of sample velocities. An equivalent method of averaging was used for the corrected blowcounts. More specifically:

NH = S di / S (di / Ni )	(7a)
where H = S di = 100 ft	(7b)
and hence: N100 = 100 ft / S (di / Ni )	(7c)

Both applications of (6) and (7) follow strictly the NEHRP procedures as we understand them.

The numbers for the averages V₁₀₀ (or N₁₀₀ ) generated by these calculations in the upper 100 feet were then categorized into site classes as follows (see Table 2) :

 

Table 2: Definition of NEHRP Site Classes by Velocity V₁₀₀ and Blow Count N₁₀₀, averaged over the upper 100 feet below grade.

Site Class	Velocity Range	Blow Count Range
A:	V100 ³ 5000 ft/s	not applicable	(8a)
B:	2500 £ V100 <5000 ft/s	N100  ³ 100	(8b)
C:	1200 £ V100 <2500 ft/s	50 £ N100 < 100	(8c)
D:	600 £ V100 <1200 ft/s	15 £ N100 < 50	(8d)
E:	V100 < 600 ft/s	N100 < 15	(8e)

Site classes D and C were found to be the most common soil types in Lower Manhattan, followed by E and B. No A’s were encountered. The N₁₀₀ computations tended to indicate softer site classes than those based on the N-to V-converted V₁₀₀ computations. Consequently there tended to be more Type E soils according to the N₁₀₀ calculations compared to those obtained from the V₁₀₀ calculations. Obviously this has to do with the fact that we used different, locally derived relations between blow counts and velocities than those that underlie the NEHRP site class definitions. Other reasons are discussed later.

For assigning the NEHRP site classes we adopted the following terminologies and conventions:

h_b = length of borehole
h_s = length of soil column or depth to bedrock measured from grade level

Different combinations of boring, soil and rock geometries required us to distinguish the following cases in obtaining the velocities in the bored soils and rocks, and/or unpenetrated soils and rocks down to at least 100 ft:

Case 1: Borehole reaches into rock, i.e. h_b> h_s

(1a): h_s > 100 ft Use soil velocities from borehole data down to 100 ft.
(1b): h_s < 100 ft Assume V_r = 5000 ft/s for portions within rock,

i.e. for depth interval h_s< x<100 ft ; otherwise use soil data from boring above rock

Case 2: Borehole does not reach into rock, i.e. boring penetrates soils only, or h_b< h_s

(2a) h_b>100ft Use soil velocities from borehole data down to 100 ft.
(2b) h_b<100ft; h_s>100ft Use soil velocities from borehole down to h_b and then for all depths x for which h_b < x < 100 ft use the following equation:

Vs (x) = Vs(hb)+ { [Vs (x=100) - Vs(hb)] / [100 ft - hb] }* [x-hb]

(9)

where Vs(x=100) is derived from the following average relation:

Vs = 435.11 + 11.08x

(10) = (4)

which yields at x=100ft:   V_s (100 ft) = 1543.11 ft/s

Note: Eqn [10 or 4] were obtained from regression of borings for which h_b > 100 ft. Boring velocities were obtained from the equation

V_s = (200 + 3N)D^0.3 ; (D is depth in feet, local shear velocity V_s in ft/s).

(2c) h_b < 100 ft; h_s < 100 ft; for depth range in bedrock: h_s < x < 100 ft: use again V_rock = 5000 ft/s. For the fraction of the soil column h < h_b penetrated by the boring, compute velocities from the boring data. For the portion of the soil column h_b<h< h_s not penetrated by the boring use:

Vs (h) = Vs (hb)+ { [Vs (x) - Vs (hb)] / [x - hb] } * [h - hb]

(11)

where the soil velocity just above the soil-rock interface at x=h_s is to be calculated from the average relation

Vs (x) = 435.11 + 11.08x (at x=hs)

(12)=(4)

i.e., x is in this case the depth to rock if known, and less than 100 ft.  If depth to rock is not known, e.g., h_s = ? we then do not know whether case (2b) or (2c) should apply. In that case, assume h_s>100 ft (case 2b).

3.6 Results: Site Classes, Spatial Averaging per Census Tract, Smoothing Between Tracts.

Using the pertinent formulas and information from Tables 1 and 2 of Sections 3.4 and 3.5, together with the boring data, allows us to determine the site classes based on V₁₀₀ . Two categories of determination must be distinguished. The first concerns sites with borings for which blowcount information was available. These site classes are shown in Fig. 4. In contrast, Fig. 5 shows the site classes derived at the boring sites for which only depth to bedrock information was available, and hence the average velocity V₁₀₀ over the upper 100 ft had to be obtained for the soil portions using the generalized average velocity profile (4) or (10) shown earlier in Fig 3. The combined results for the site classes from both the borings with soil and rock descriptions and blow counts, and with only depth to bedrock information are shown in Fig. 6.

Obviously the results in Figures 4 through 6 apply to each borehole individually and not to any of the enclosing census tracts. These tracts are shown on the figures only as a geographic background reference for orientation. Fig. 7 shows the results from both "smoothing" of the individual borings within a given census tract to assign a single "dominant" site class in each tract; and the "interpolation" of site classes for those census tracts that did not contain any borings at all, at least at this initial stage of the pilot study. We had originally intended to develop algorithms for both the smoothing and the interpolation of site class assignment per census tract. However, since we believe that subsequent phases of the study will yield much higher density of information in individual tracts, and/or fill in the gaps for tracts that currently have not had any boring information, we temporarily resorted to site class assignment by informed, albeit subjective judgment, influenced by geological information contained in Fluhr and Murphy (1944), Baskerville (1990, 1992, 1994) and the depth to bedrock information as exemplified by Fig.1. Plans for future revisions and improvements are addressed in the Discussion Section.

 

3.7 Site Class Distributions

NEHRP site classes in southern Manhattan assigned to census tracts (after averaging within tracts and interpolation between tracts to cover gaps) yield mostly site class D tracts, with almost as many C tracts and a few B, but no A and E tracts (Fig. 7), although a few E but no A classes had been encountered at individual boring sites (Fig 6). This frequency distribution of site classes indicates that in the study area the sediments and fills, since they are mostly of Holocene and Recent origins, are quite soft, and bedrock is rarely close to the surface especially in the southern half of the study area. This picture is likely to be substantially different once the study proceeds to the northern half of Manhattan (and the Bronx) where firm rock frequently outcrops or is covered with only a thin layer of often dense glacial till or glacially overridden soils. These firmer sites are generally associated with higher topographic elevations and steeper slopes. Elevations at the highest points in northern Manhattan reach up to about 200 ft. The colloquial terms of "up-town" and "down-town" or Upper and Lower Manhattan do therefore refer also to general elevation patterns that have sedimentological and geotechnical corollaries. HAZUS default values assume site class D conditions for all census tracts (see HAZUS User’s Manual, footnote on bottom of page E-64 of its Appendix E; [NIBS, 1997]). Because of the preponderance of D sites, ground motions in the census tracts with this dominant site class will remain unchanged from the default values, while the ground motion in the tracts with the less common site classes C and B will be downgraded to lower ground motions than the HAZUS default site class D yields.

 

3.8 Comparison of NEHRP Site Classes from Blow Counts and Shear Wave Velocities.

NEHRP ‘97 allows site classes to be determined from either blow counts or shear wave velocities. Using, for the same boring, the N-value depth profile directly and, alternatively, the derived shear wave velocity depth profile to compare the resulting NEHRP site classes, we find that not in all instances do we obtain the same site class for the same site. Instead, differences of 1 site class for the same boring are not uncommon. There are two possible reasons for this divergence. First, no blow counts had been assigned to the portions of the 100ft depth profile that penetrate rock. Therefore, when only the N values of the soil section is used, the resulting site class will inevitably move towards a softer category (e.g. E instead of D). Second, the NEHRP site class definitions (Table 2) imply a relationship between N values and shear wave velocities that is slightly different from the one we adopted based on the calibration data set for NYC (Fig. 2 and equation [3]). Because of the possible differences between local and "global" relations of shear wave velocity to blow counts, and because the NEHRP site class definition (as we understand it) is implying that the rock section in the upper 100 ft should be accounted for, we prefer the V₁₀₀ over the N₁₀₀ site class definitions. To guarantee consistency between all sites, whether only soil or soil and rock is present in the upper 100 ft, we use for subsequent loss computations the site classes obtained from V₁₀₀ (see Fig. 6), and not those from N₁₀₀ (not shown in any Figure).

 

3.9 Site and Building Resonances.

We can estimate fundamental-mode soil resonances using the simple relation

T = 4H / Vh

(13)

where H is the depth to bedrock or 100 ft (which ever is less), and V_h is the slowness-averaged mean velocity for the soil column to depth H. Using the bounds set for in Table 2 and the results from Figures 5, 6 & 7, we obtain periods varying between about 0.1 and nearly 2.0 seconds depending on depth to bedrock and softness of the soil column. A few values fall both below and above this period range. The estimated range of soil periods coincides with the range of fundamental-mode periods for buildings with heights from about 1 to 20 stories, and with higher-mode periods for structures taller than 20 stories. Hence, the variable soil conditions are expected to affect a wide range of building heights that are quite common to New York City.

 

3.10 Scenario Earthquakes and Effects of Soil Categories on Ground Motions and Loss Estimates.

We have placed three scenario earthquakes of magnitudes Mw=5.0, 6.0 and 7.0 near the location (red asterisk in Fig. 8) of a historic Mblg=5.2 earthquake that occurred in 1884, close to the mouth of the NY Harbor. For the scenario loss estimations we chose for the PESH (Potential Earth Science Hazards) module of HAZUS the following input options:

Latitude: 40.5600^oN
Longitude: 74.0000^oW
Depth: 10km
Aver.Distance: ˜ 20km (from center of study region in S. Manhattan)
Attenuation Model: P‘97 East Coast

The location and short period response spectral accelerations (S_a at 0.3 sec) for the M=6.0 event and for default site conditions (site class D) are shown in Fig. 8 for the NYC metropolitan area including the study region of Manhattan. The short period S_a(0.3s) values for the study area are shown in greater detail in Fig. 9 for the default site class D, and in Fig. 10 for the site classes for each census tract as depicted in Fig. 7 for the study region of Manhattan south of 59th Street. Because these site classes are either B, C, or D, the ground motions in Fig. 10 are either equal to or lower than those for the default site class D ground motions depicted in Figs. 8 and 9. The exact ratios of the down grading of motion levels can be inferred from the site amplification factors shown in the NEHRP 1997 Tables for short-period site coefficients F_a reproduced in Appendix A. Similar down grading occurs for the mid-period response spectral acceleration S_a(1.0s) using the mid-period site amplification factors F_v also shown in Appendix A. For brevity we do not show a Figure with the mid-period ground motions.

Fig. 11 shows a typical HAZUS result output sheet for direct economic losses related to buildings in the study area, but only for the Mw=6.0 scenario event. The entire building inventory for the study area is valued (under building default assumptions that are modified in the related Princeton Report, see Nordenson et al., 1999) at about US$ 160 Billion, and the total loss computed for the assumption of uniform default soils of site class D is given as approximately US$ 11.3 Billion.

Fig. 12 depicts a variety of summary results of both the PESH and loss computations for the three magnitudes (Mw = 5, 6, and 7) at the same location as defined above. In particular, Fig. 12 shows under the assumption of default soil class D the direct or derived results from the HAZUS computations: Minimum and maximum values of ground motions across the study area for S_a(0.3s) and S_a(1.0s) measured in units of g; direct economic building-related losses (DEBL) in billions of US$; and the loss ratio as defined and shown in Fig. 11 (i.e. ratio of building related losses to building capital stock assets). Fig. 13 shows the same type of results, but now under the modified assumptions of site classes B, C, and D as shown in Fig. 7 and for the related ground motions shown in Fig. 10. Finally, the ratios of the HAZUS results (for both ground motions and losses) are shown in Fig. 14 for all the parameters that were previously depicted in Figures 11 and 12. Here ratios are defined as results for modified soils divided by results for default soils. Note that a ratio of 1 applies for maximum values of ground motions in the study area because they stay of course the same (i.e. for those census tracts where modified soils [S] and default site classes [D] are both the same). In contrast, the ratios for the minimum ground motions for S_a(0.3s) and S_a(1.0s) in the study area show values of [S]/[D] as low as 0.65 and 0.40, respectively for the Mw=5.0 event; and about 0.81 and 0.53, respectively, for the Mw=7 event. These ratios directly reflect the soil amplification factors F_a and F_v, respectively, appropriate for the ground motion levels Sa listed in the Tables of Appendix A. It is not surprising that the loss ratios (for actual soil to default soil losses) fall somewhere between the ratios for minimum and maximum ground motions encountered in the study area, since the losses are integrated over the entire study area, inside which only some but not all ground motions were modified when actual site classes were accounted for. It is also not surprising that the loss ratios approach increasingly 1 with increasing magnitude, i.e. the loss ratios are about 0.71, 0.74 and 0.83 for the magnitudes Mw = 5, 6, and 7, respectively. This is so because the site amplification factors F_a and F_v decrease with increasing ground motion input (see Appendix A), which in turn increases with increasing magnitude at any given epicentral distance. In other words: because of nonlinearity of soil amplification, the higher the magnitude, the less important become the spatial variations in soil conditions in the study area (at least this is the case from the vantage point of shaking damage alone, i.e. not considering soil liquefaction effects which we have not dealt with here).

 

4. Insights Gained from Pilot Study and Recommendations for the Future.

4.1 Availability of Borings. Geotechnical boring data in analog hard-copy form are readily accessible for most public construction projects, whether completed or currently underway. But in order to achieve more uniform spatial coverage and higher resolution it will be important to include data from private sector construction projects that are sometimes harder to access; additional data from public work projects other than those compiled by NYC DDC have not yet been tapped. Close cooperation with local geotechnical firms and NY City and State agencies, and Public Authorities should allow to advance the quality of future microzonation efforts for better loss estimation and other applications.

4.2 Calibration Data. While the translation of SPT blow-counts into shear wave velocity Vs is not an exact science we find that if attention is paid to locally based calibration data (see individual data points in Fig. 2), the SPT blow count data are useful low-cost substitutes for more accurate but quite expensive shear wave velocity measurements (whether downhole, crosshole, or from surface wave dispersion along surface profiles, or by other means). It would be, however, desirable to have additional calibration borings plus shear wave velocity measurements performed at carefully selected sites. This may especially apply when the pilot study is extended to areas covered with till where currently little prior information is available. We expect to cooperate with various agencies, and in particular with the NY State Geological Survey to obtain additional data sets that will help to better calibrate the blow-count to shear velocity conversion relations.

4.3 Filling in Data Gaps. The depth limitations of some boreholes in some areas, and the currently encountered uneven spatial distribution of the NYC DDC boring data, require that spatial data gaps must be filled in either by extrapolation or interpolation, using sometimes the locally derived generalized velocity functions that reflect overall trends with depth, but are blind to local perturbations. It will therefore be of utmost importance to fill in gaps by obtaining additional boring data, rather than by interpolation and extrapolation techniques.

4.4 Verifying Site Resonances. Within this pilot study, we had not the resources available to check whether computed fundamental-mode site resonances using Equation (13) coincide with those observable with seismic recordings of microseisms and using horizontal to vertical spectral ratios to identify the spectral response peaks ("Nakamura’s Method", see Field and Jacob, 1995). Such verifications would provide the information needed when the soil-structure resonances of individual major buildings or structures are under consideration. For the purpose of improving the overall accuracy of seismic loss computations with the HAZUS program and given that the spatial resolution need not exceed much the dimensions of census tracts, the chosen approach to convert boring data to shear wave velocities seems the most cost efficient, quantitatively acceptable while not ideal or most accurate approach.

4.5 Assumption About Rock Velocities. We have throughout this pilot study assumed that the bedrock velocity is 5,000 ft/s. As indicated earlier, while this conforms to the NEHRP definition of bedrock, it may not be a particularly appropriate assumption for NYC-type crystalline bedrocks, such as Manhattan schists. It may be advisable to determine more realistic (presumably higher) bedrock velocities and use these values when computing the V₁₀₀ average velocities in the upper 100 feet below grade, which in turn may then move in many census tracts the obtained representative site class to a firmer type (lower rank in the alphabet, e.g. C-->B).

4.6 Multiple Interpretations and Usage of NEHRP Site Definitions. We have followed in this report what we understood to be the intent of the NEHRP provisions when computing the site classes from the V₁₀₀ velocities in conjunction with Equations (6). This implied to include the rock sections in any boring down to 100 ft into the calculations of the average velocity V₁₀₀. However, an alternative interpretation of the NEHRP intent is possible, and has been applied for instance in the studies performed by the NJGS for the test region around Newark, NJ. Instead of average velocity V₁₀₀ , an average blow count N₁₀₀ (Equations 7) was used, implying that only data points in the soils above bedrock were considered. Hence the vertical sections in the upper 100 feet that contain bedrock were not accounted for. This second method yields obviously softer soil classes than including the bedrock sections as well (as we have done in this report, using velocities instead of blowcounts, and all depth intervals down to 100 ft). Using only the soil section makes physically sense, but the so obtained results and amplification factors would in reality apply largely to buildings with periods at or below that of the approximate fundamental resonance period of the soil layer. If the site class and related design spectrum so obtained were to be applied to structures with periods much longer than the fundamental soil layer period, then an overestimation of the site amplification at long periods may be occurring. This is an issue that needs to be addressed by both the community using HAZUS, but more fundamentally is an issue that needs to be addressed in a much broader context by NEHRP’s geotechnical code-writing group.

4.7 Scope of Future Efforts for Determining Site Classes in NYC. Using the experience from this pilot study and trying to estimate the resources needed to perform a useful microzonation for the five NY City boroughs proper (but excluding the surrounding Metro region) we estimate that a total number of sites needed for all of NY City would probably approach 3000 to 5000 borings. If a fully digital map of the depth to bedrock-soil interface will be included, using several thousand data points in Manhattan alone, and if a test of the spectral site response with seismic instrumental means is deemed worthwhile, considerable efforts will be required that imply effective cooperation between the academic, public and private sectors. The tools and methodologies are available; the task is feasible and executable along the procedures outlined above; however the manpower and resources that need to be mobilized will be considerable.

4.8 HAZUS Results as Useful Mitigation Strategy Development Tool. Indications from the preliminary HAZUS runs are that the loss estimations for NY City are sensitive to the geologic site classes, especially in specific portions of the city that have highly variable soil conditions. We suggest, that a successful seismic mitigation strategy for NY City needs preparation of a more detailed microzonation product coupled to a GIS-based information system for easy dissemination to the engineering community and the public in large. The utility of such a product would go far beyond making better estimates of potential future earthquake losses. A well maintained digital soil archive of NY City would serve a wide range of purposes, some beyond purely earthquake hazard mitigation applications. Such a data base combined with proper analysis tools may have value as a general urban planning resource and as a decision tool and could constitute an important component of a future effort towards a digital "Virtual City" or "Cyber City" at least, and a more sustainable, resilient and safer City in reality at best.

 

References

Baskerville, C.A., (1990), Bedrock and Engineering Geology Maps of New York County and Parts of Kings and Queens Counties. 3 sheets. USGS Open File Report 89-462.

Baskerville, C.A. (1992), Bedrock and Engineering Geology Maps of Bronx County and Parts of New York and Queens Counties, New York. USGS Miscellaneous Investigation Series MAPI-2003, Scale 1:24,000, 2 sheets, US Geological Survey, 1992.

Baskerville, C.A. (1994), Bedrock and Engineering Geology Maps of New York County and Parts of Kings and Queens Counties, New York, and Parts of Bergen and Hudson Counties, New Jersey. USGS Miscellaneous Investigation Series MAPI-2306, Scale 1:24,000, 2 sheets, US Geological Survey, 1994.

Budhu, M., R. Giese and L. Baumgrass (1989), Liquefaction Potential of Surficial Deposits in the City of Buffalo, New York. Technical Report NCEER-89-0036. NCEER at SUNY Buffalo, NY, January 17, 1989.

FEMA (1998a). 1997 Edition: NEHRP Recommended Provisions for Seismic Regulation for New Buildings, Part 1 - Provisions. Published by the Federal Emergency Management Agency (FEMA), as FEMA # 302, Washington DC

FEMA (1998b). 1997 Edition: NEHRP Recommended Provisions for Seismic Regulation for New Buildings, Part 2 - Commentary. Published by the Federal Emergency Management Agency (FEMA), as FEMA # 303, Washington DC

Field, E.H. and Jacob, K.H., (1995), A Comparison and Test of Various Site Response Estimation Techniques Including Three that are Non Reference-Site Dependent, Bull. Seismol. Soc. Am., 85, No.4, pp. 1127-1143, 1995.

Fluhr, T., W. and J.J. Murphy (1944), Map showing 25-foot Contours of Bedrock of the Boro of Manhattan, City of New York, Map # 120-A. Dated 3/17/1944.

NIBS (1997), Earthquake Loss Estimation Technology - HAZUS, User’s Manual; prepared by the National Institute of Building Sciences (NIBS) for FEMA. Washington D.C. (1997)

Nordenson, G., G. Deodatis, M. Tantala and A. Kampf (1999). NYCEM (New York City Area Consortium for Earthquake Loss Mitigation), 1st Year Technical Report, May 1, 1998 to April 30, 1999: EARTHQUAKE LOSS ESTIMATION STUDY FOR THE NEW YORK CITY AREA. Department of Civil Engineering and Operations Research; PRINCETON UNIVERSITY; April 1999

Wessel, P. and Smith, W. H. F. (1995) New version of the Generic Mapping Tools released. Eos, Transactions, American Geophysical Union 76(33), 329.

see also: http://www.soest.hawaii.edu/wessel/gmt.html   and/or: http://imina.soest.hawaii.edu/gmt/

 

Acknowledgments:

We are grateful to the NYC DDC to allow us generous access to their well organized boring files. At various stages LDEO’s John Armbruster, Sharon Hoffmann and Maggie Brewer contributed unselfishly and enthusiastically their time, efforts and ingenuity to get the often tedious job of compiling boring records into digital formats and carrying out the various aspects of the data processing and analysis. NYSEMO contributed additional funding for internship salary, and MCEER administered the crucial FEMA funding. I would like to acknowledge especially the roles that Andrea Dargush (MCEER), Dan O’Brien (NYSEMO), Bruce Swiren (FEMA), Guy Nordenson and George Deodatis (both Princeton Univ.), Michael Greenman (NYC DDC), Larry Knafo (NYC EMO) and Mishac Yegian (NEU) played to facilitate various aspects of this study.

 

 

Appendix A:

 

Excerpts from the HAZUS User’s Manual
and from the NEHRP ‘97 Seismic Provisions
Regarding
Site Classes and Site Amplification Factors.

 

Table A.1: Classification Systems - Site Classes (from the 1997 NEHRP Provisions)

Site Class	Site Class Description	Shear Wave Velocity
		Minimum	Maximum
A	HARD ROCK Eastern United States sites only	1500
B	ROCK	760	1500
C	VERY DENSE SOIL AND SOFT ROCK Undrained shear strength Us ³ 2000 psf (us ³  100 kPa) or N ³ 50 blows/ft	360	760
D	STIFF SOILS Stiff soil with undrained shear strength 1000 psf £ Us £ 2000 psf (50 kPa £ Us £ 100 kPa) or 15 £ N £ 50 blows/ft	180	360
E	SOFT SOILS Profile with more than 10 ft (3 m) of soft clay defined as soil with plasticity index PI > 20, moisture content w > 40% and undrained shear strength Us < 1000 psf (50 kPa) (N < 15 blows/ft)		180
F	SOILS REQUIRING SITE SPECIFIC EVALUATION Soils vulnerable to potential failure or collapse under seismic loading: e.g. liquefiable soils, quick and highly sensitive clays, collapsible weakly cemented soils. Peats and/or highly organic clays (10 ft (3 m) or thicker layer) Very high plasticity clays: (25 ft (8 m) or thicker layer with plasticity index > 75) Very thick soft/medium stiff clays: (120 ft (36 m) or thicker layer)

 

Table 4.1.2.4a: Values of F_a as a Function of Site Class and Mapped Short-Period Maximum Considered Earthquake Spectral Acceleration (from the 1997 NEHRP Provisions)

Site Class	Mapped Maximum Considered Earthquake Spectral Response Acceleration at Short Periods
	Ss £ 0.25	Ss = 0.50	Ss = 0.75	Ss = 1.00	Ss ³ 1.25
A	0.8	0.8	0.8	0.8	0.8
B	1.0	1.0	1.0	1.0	1.0
C	1.2	1.2	1.1	1.0	1.0
D	1.6	1.4	1.2	1.1	1.0
E	2.5	1.7	1.2	0.9	a
F	a	a	a	a	a

NOTE:

Use straight line interpolation for intermediate values of Ss. a Site-specific geotechnical investigation and dynamic site response analyses shall be performed.

 

Table 4.1.2.4b: Values of F_v as a Function of Site Class and Mapped 1 Second Maximum Considered Earthquake Spectral Acceleration (from the 1997 NEHRP Provisions)

Site Class	Mapped Maximum Considered Earthquake Spectral Response Acceleration at 1 Second Periods
	S1 £ 0.1	S1 = 0.3	S1 = 0.4	S1 = 1.00	S1 ³ 0.5
A	0.8	0.8	0.8	0.8	0.8
B	1.0	1.0	1.0	1.0	1.0
C	1.7	1.6	1.5	1.4	1.3
D	2.4	2.0	1.8	1.6	1.5
E	3.5	3.2	2.8	2.4	a
F	a	a	a	a	a

NOTE:

Use straight line interpolation for intermediate values of S1. a Site-specific geotechnical investigation and dynamic site response analyses shall be performed.

 

Home | Background | HAZUS | Research Activities | Who We Are | Technical Documents | Contacts | Links

Last modified: 4/11/2001 10:56:57 AM
Suggestions and comments to the webmaster.