But What Does Ground Shaking Intensity REALLY Mean?
Based on ABAG "On Shaky Ground" Reports
Shaking Intensity and Building Damage
Shaking Intensity and Landsliding
Shaking Intensity and Liquefaction
Correlation of Shaking Intensity With Other Measures of Shaking Severity
Introduction

"Official" Modified Mercalli Intensity Descriptions Can Be Confusing

This web site provides maps showing modeled shaking intensity for expected future earthquakes using the modified Mercalli intensity (MMI) scale. The full description of each intensity level is provided in the description of the MMI scale. However, the"official" descriptions of MMI level (Ref. 2) were written approximately 40 years ago and are often difficult to interpret, vague and archaic.

Current Research Provides Examples of MMI Impacts

We can now provide more "quantitative" descriptions of the impacts of shaking on buildings, probabilities of ground failure (including liquefaction and landsliding), and conversions among intensity scales and to other measures of shaking strength than were provided by the "official" descriptions. These data are based on research by ABAG and others in the past few years, and are provided below.

Shaking Intensity and Building Damage
The Question

How does ground shaking intensity relate to damage to various types of building construction?

What We Know

The likelihood of building damage is radically different for different types of buildings. After the Northridge earthquake, the Superior Apartments (shown below) were heavily damaged. However, a group of single family homes behind the apartments experienced little damage. These apartments were constructed to comply with modern building codes.

The damage to buildings can be depicted using two separate measures of damage:

  1. The percentage of buildings of a particular construction type (defined by use, construction materials, height and age) "red-tagged" by the local government building inspector as "unsafe for human occupancy," that is, uninhabitable, or
  2. The average dollar loss (expressed as a percentage of the replacement value) for each construction type.

Based on information compiled by ABAG for residential construction (Ref. 3) and by EQE and OES for commercial construction (Ref. 4), it is relatively easy to generate a table of percent of housing units and commercial buildings typically "red tagged" for several construction types, as shown in the following table.

What We Don't Know

Although a table of average dollar loss by construction type might, arguably, be more useful than the habitability information provided here, it is our judgement that information is insufficient to create such a table at this time. Data on the value of buildings "at risk" in past earthquakes and reliable damage data are scarce. In addition, there is no reliable data on the habitability of tilt-up concrete buildings (separate from other types of concrete buildings), or on wood-frame commercial buildings (separate from residential buildings). Information on these two types of buildings is therefore not included in this table.


Example of Damage to Post-1940s Multifamily Residential (Superior Apartments, Northridge)
Source: Jeanne Perkins, ABAG
Example of Damage to Mobile Home
Source: Karl Steinbrugge
Example of Damage to Unreinforced Masonry Cafe with Residential Units Above
Source: Henry Degenkolb
Example of Damage to Concrete Building
Source: Northridge Earthquake Collection, Earthquake Engineering Research Center, University of California, Berkeley

TABLE 1: Percent of Dwelling Units (for Residential) and Buildings (for Commercial) Red Tagged as Uninhabitable by Construction Type and MMI Intensity
  INTENSITY
RESIDENTIAL TYPE 1 V VI VII VIII IX X
Mobile Homes almost 0 0 0.87 40 90 100
Unreinforced Masonry almost 0 0.05 2.9 45 70 80
Non-Wood, 4-7 Stories, <1940 almost 0 0.30 8.0 45 70 80
Wood-Frame, 4-7 Stories, <1940, Multi-family almost 0 1.4 2.5 45 70 80
Wood-Frame, 1-3 Stories, <1940, Multi-family almost 0 0.05 0.53 11 44 64
Wood-Frame, 1-3 Stories, >1939, Multi-family 2 almost 0 0.01 0.04 6.5 15 25
Wood-Frame, 1-3 Stories, <1940, Single Family 3 almost 0 0.04 0.12 1.8 8.4 12
Wood-Frame, 1-3 Stories, >1939, Single Family almost 0 0 0.02 0.18 0.69 1.8
COMMERCIAL OR INDUSTRIAL ..
Unreinforced Masonry 4
almost 0
1.0 8.0 45 70 80
Micellaneous Concrete 5
almost 0
0 1.0 20 33 40
Note 1. The relationship between intensity and construction type for residential buildings are taken from Ref. 3, pg. 68.
Note 2. These percentages include a mixture of buildings with anf without full or partial parking underneath the structure. Data for buildings with and without parkingare not directly available. However, the values for multi-family buildings without parking are probably closer to those for >1939 wood-frame single family homes, and those for buildings with parking could easily be double the percentages listed here.
Note 3. Homes built prior to 1940 were not bolted to their foundations. However, these percentages include an unknown mixture of homes that have, and have not, been retrofitted by adding thesse bolts and installing plywood sheathing on the inside of the crawl space.
Note 4. Note that the percentages of commercial unreinforced masonry buildings red tagged are higher than those for the residential unreinforced masonry because these buildings typically have fewer room partitions.
Note 5. Taken from Ref. (Table 4-3), except for MMI VII (which was revised downward from 8% to 1% based on lack of damage in the Loma Prieta earthquake). These percentages apply to "general" concrete buildings.
Shaking Intensity and Landsliding
The Question Can ground shaking intensity be correlated to earthquake-triggered landsliding?
What Is the Hazard? Landslides are often triggered by the shaking of earthquakes. These ground failures are of two principal types (Ref. 5): ¨ disrupted slides, falls and flows - landslides with highly jumbled materials that start on steep slopes and move at relatively high speeds, such as soil or rock slides, rock falls and avalanches, and debris flows; and ¨ coherent slides - blocks of unjumbled materials that move on a discrete slide surface, such as slumps, block slides and earth flows.

What We Know

The California Division of Mines and Geology (CDMG) has a program to map earthquake-induced landslide hazard areas throughout California. Currently, this type of Seismic Hazard Zone Map is only available for portions of the Bay Area and several areas in Los Angeles, Ventura and Orange counties in southern California. Additional mapping is subject to the availability of state and federal funding. The program is mandated by the Seismic Hazards Mapping Act (Public Resources Code, Ch. 7.8)

 

Much effort was made to document the location, shape, and severity of the landslides triggered by the October 1989 Loma Prieta earthquake and the January 1994 Northridge earthquake. Approximately 1,500 earthquake-triggered landslides were mapped, and up to 4,000 slides may have moved, in the Loma Prieta earthquake (Ref. 6). Over 11,000 landslides occurred in the Northridge earthquake (Ref. 7). Significantly, both earthquakes occurred when the ground was exceptionally dry. Extensive research on the distribution and causes of these slides shows that failure rates can be correlated with (1) shaking severity; (2) slope steepness; (3) strength and engineering properties of geologic materials; (4) water saturation (which varies with precipitation and by season); (5) existing landslide areas; and (6) vegetative cover.

Researchers have correlated areas of known earthquake-induced landslides to Arias intensity, a measure of shaking severity defined on page 11. Areas subjected to Arias intensities of greater than about 0.54 m/sec commonly have earthquake-triggered landslides. Table 7 on page 11 shows this intensity is roughly equivalent to a modified Mercalli intensity of VII or greater. Small numbers of landslides can occur at MMI VI.

Slope length and slope aspect (that is, orientation facing north, south or somewhere in between) contribute to earthquake-induced landslide susceptibility. However, slope steepness (as expressed in percent slope) is the most critical slope factor.

The mapped geologic units in the Bay Area can be grouped according to an approximate material shear strength classification of A, B, or C, with A being those units least susceptible to sliding and C being those units most susceptible to sliding. A table correlating these geologic material units with their shear strength classifications is included in Riding Out Future Quakes (Ref. 8, Appendix C).

The final factor included in this analysis is degree of water saturation. This variable depends in large part on length of time since the last major storm and rainfall to date. Because these data cannot be known ahead of time, two tables correlating landslide susceptibility with saturation have been generated - one for dry (summer) conditions and a second for wet (winter) conditions. The intensities required for landslides tend to be lowered by approximately one intensity unit under wet conditions.

What We Don't Know





Two important factors contributing to earthquake-induced landslide susceptibility have not been incorporated into these tables.

First, existing landslides are not included because any compilation of data on their location is presently sporadic; no regional depository exists for the wealth of data collected for individual development projects.

Second, vegetative cover is not incorporated into the following tables because very little research has been conducted quantifying its effect.
TABLE 2: Earthquake-Induced Landslide Susceptibility - Dry (Summer) Conditions -
Based on Modified Mercalli Intensity (MMI), Percent Slope, and Material Type (A, B or C) 1
[Values in this table are the percentage of the land units being analyzed expected to have at least one landslide. The land analyzed are one hectare squares, or units 100 meters on each side. ]

Percent Slope 0 - 5% Slope 6 - 15% Slope 16 - 30% Slope 30+ % Slope
Material Type A B C A B C A B C A B C
MMI IX and X 0 1 2 1 2 12 5 8 18 8 18 30
MMI VIII 0 0 0 0 0 5 0 5 12 5 8 18
MMI VI 0 0 0 0 0 3 0 3 4 3 5 12
MMI VI 0 0 0 0 0 0 0 0 0 0 0 0

TABLE 3: Earthquake-Induced Landslide Susceptibility - Wet (Winter) Conditions -
Based on Modified Mercalli Intensity (MMI), Percent Slope, and Material Type (A, B or C) 1
[Values in this table are the percentage of the land units being analyzed expected to have at least one landslide. The land analyzed are one hectare squares, or units 100 meters on each side. ]

Percent Slope 0 - 5% Slope 6 - 15% Slope 16 - 30% Slope 30+ % Slope
Material Type A B C A B C A B C A B C
MMI IX and X 1 2 12 5 8 18 8 18 30 12 24 50
MMI VIII 0 1 2 1 2 12 5 8 18 8 18 30
MMI VII 0 0 0 0 0 5 0 5 12 5 8 18
MMI VI 0 0 0 0 0 3 0 3 4 3 5 12

Note 1. A Table correlating these geological material units with their shear strength classifications is included in Riding Out Future Quakes (Ref. 8, Appendix C).

Shaking Intensity and Liquefaction
The Question Can shaking intensity be correlated to areas of liquefaction?
What Is the Hazard ? When the ground liquefies, sandy materials which are saturated with water can behave like a liquid, instead of like solid ground. In essence, the sand grains momentarily behave like a liquid.

Liquefaction is defined as "the transformation of a granular material from a solid state into a liquefied state as a consequence of increased pore-water pressure" (Ref. 9, p. 1). Engineers call this "loss of shear strength." The ground needs to be shaken strongly for liquefaction to occur, and this shaking can occur as a result of an earthquake.

Liquefaction can cause ground displacement and ground failure. In addition, it can cause lateral spreads and flows (essentially landslides on flat ground next to rivers, harbors, and drainage channels).
Potential Results of Liquefaction
What We Know

The "recipe" includes three ingredients necessary for damaging liquefaction to occur:

  • INGREDIENT 1 - The ground at the site must be "loose" - uncompacted or unconsolidated sand and silt without much clay or stuck together.
  • INGREDIENT 2 - The sand and silt must be "soggy" (water saturated) due to a high water table.
  • INGREDIENT 3 - The site must be shaken long and hard enough by the earthquake to "trigger" liquefaction.
In areas farther from the earthquake fault source, a material that has high liquefaction susceptibility may liquefy, but an adjacent material of moderate susceptibility may not. Only some materials with very high liquefaction susceptibility will liquefy when exposed to strong shaking (modified Mercalli intensity (MMI) VII), with less susceptible materials being triggered with very strong shaking (MMI VIII). (Intensity is a measure of shaking severity at a particular location.) Liquefaction in areas shaken less than MMI VII, or in areas mapped as having a low to very low liquefaction susceptibility, is a statistical possibility, but it is not likely. The following maps show liquefaction hazard in various earthquake scenarios in three simplified categories, graphically shown below.
TABLE 4: Potential Likelihood for Liquefaction Based on a Combination of Shaking Intensity and Liquefaction Susceptibility
Modified Mercalli Intensity Liquefaction Susceptibilty Liquefaction Hazard
IX and X Very High High
High High
Moderate High
VIII Very High Moderate
High Moderate
Moderate Moderate
VII Very High Moderate
High Moderately Low
Moderate Moderately Low
VI Very High Very Low
High Very Low
Moderate Very Low

Where We're Going

ABAG, William Lettis & Associates (WLA), and the U.S. Geological Survey produced revised liquefaction susceptibility and liquefaction hazard maps for the San Francisco Bay Area in 2000 and 2001. As part of that effort, additional data on the shaking required for liquefaction to take place was collected. Other researchers have conducted studies of the relationship between liquefaction and Arias intensity (see Refs. 10 and 11). Click here for more information on liquefaction.
The California Division of Mines and Geology (CDMG) has a program to map liquefaction hazard areas throughout California. Currently, this type of Seismic Hazard Zone Map is only available for parts of the Bay Area and several areas in Los Angeles, Ventura and Orange counties in southern California. Additional mapping is subject to the availability of state and federal funding. The program is mandated by the Seismic Hazards Mapping Act (Public Resources Code, Ch. 7.8)
Correlation of Shaking Intensity with Other Measures of Shaking Severity
The Question The modified Mercalli intensity scale seems so subjective. Can ABAG's intensity maps be converted to other, more quantitative, measures of shaking severity? What peak velocities or undamped velocity response spectra are roughly comparable to the shaking intensities shown on ABAG's maps?
What We Know The ABAG ground shaking intensity maps were produced using a model that predicts the decrease (attenuation) of shaking away from the fault source developed by J. Boatwright (Ref. 1). The model predicts the undamped velocity response spectra, in units of cm/sec (typical of a velocity measurement), not cm/sec2 (units of acceleration). This model therefore predicts a parameter more closely related to velocity than acceleration, and does not model intensity directly.

To predict intensity, we correlated the resulting model maps using both modified Mercalli intensity information and rarer San Francisco intensity information (from, largely, the 1906 San Francisco earthquake) in order to calibrate the model. We use units of intensity in the map legend because they are much easier for most people to understand. Typical intensity maps made by others use damage information and what people felt to map intensities of earthquakes which have already occurred. We have attempted to model these general effects in future earthquakes based on shaking severity information.

If, however, you want or need a quantitative measure of shaking strength, you can correlate the map legend to these other measurements using Table 5, below. This table was generated using more information than was available for On Shaky Ground in 1995 (Ref. 1, pg. A46). It is consistent with Riding Out Future Quakes published in 1997 (Ref. 8, pg. 29).
What We Don't Know Overall, however, there is a shortage of actual data from seismographs near the source faults of major earthquakes to test this theoretical model. The values need to be checked, and may need to be modified, following future major earthquakes.

The maps are intended to depict the relative severity of shaking in one area relative to other areas in the earthquakes modeled. They do not, nor can any general map created prior to an earthquake, be a substitute for evaluation of the level of shaking at a specific site made by qualified seismologists or geotechnical engineers, or assessment of the performance of a specific structure at that site by a licensed structural engineer.
Where to Go for Maps
Showing Probability of Exceedance Information
Because the shaking severity maps for individual earthquakes are based on a shaking measurement called the undamped velocity response spectra, the maps could be combined to create a map based on the probability of exceeding this level. This scheme was used to create the probabilistic shaking hazard maps developed by the U.S. Geological Survey and the California Division of Mines and Geology (see Refs. 12 and 13) for peak horizontal ground acceleration, not undamped velocity response spectra used for ABAG's maps. The correlation between undamped velocity response spectra and peak acceleration is too weak to warrant inclusion in the table below.
TABLE 5: Approximate Relationships Among Intensity Scales, Particle Velocity and Undamped Velocity Response Spectra
NOTE - These correlations apply to the ABAG maps because of the way they were generated. They do not work with other MMI maps. Therefore, this table should not be used to convert MMI or San Francisco Intensity maps generated by others to Aria intensity, undamped velocity response spectra, or peak velocity.
Undamped Velocity Reponse Spectra 1 (cm/sec)
Peak Velocity (cm/sec)
Arias Intensity 2(m/sec)
San Francisco Intensity
Modified Mercalli Intensity (as shown on ABAG maps)
Summary of Damage
Used in 1995
Shaking Severity 3 Roman Numeral

(more than shaking)
(more than shaking)

    A - Very Violent     XII
XI
450
300
204
141
96
66
45
30
21
15
9
286
191
130
90
61
42
30
19
13
10
6
48.7
21.6
10.0
4.8
2.2
1.1
0.5
0.2
0.1
0.05
0.02

B -Violent

C -Very Strong

D - Strong

E - Weak

<E - Very Weak
Extreme

Heavy

Moderate

Nonstructural

Objects Fall

Pictures Move
Very Violent

Violent

Very Strong

Strong

Moderate

Light
X

IX

VIII

VII

VI

V
Note 1. Undamped velocity response spectra is equivalent, but not identical, to average acceleration spectral level. The relationship between these quantities and the intensity values has been modified due to additional data gatheres after the Loma Prieta and Northridge earthquakes (oral communication, J. Boatwright, U.S. Geological Survey). All of the quantitative measurements of shaking strength used in this table have units of velocity, not acceleration.
Note 2. Arias intensity is an estimate of the energy delivered to structures on the earth's surface. The actual formula is provided in Ref. 10:
Equation
where is Arias intensity, g is the acceleration of gravity, and the remaining term is the integral of the square of acceleration over time.
Note 3. As can be seen from this table, the terms for shaking intensity now being used on the ABAG maps are similar, but not identical, to those used to describe San Francisco intensity (an intensity scale used following the 1906 San Francisco earthquake). These quantitaive terms do not refer to the same quantitative shaking levels, however.

ABAG, the Association of Bay Area Governments, is the regional planning and services agency for the nine-county San Francisco Bay Area.
Source - 2003 "On Shaky Ground" documentation prepared by ABAG.


jbp 10/15/03