Ground Motion Amplification of Soils in the Upper Mississippi Embayment

The correlation between strong ground motions and geology was identified in the mid-1800s (Del Barrio, 1855; Mallet, 1862).   Recent studies by Borcherdt (1994) and Anderson et al. (1996) have quantified the influence of near-surface geologic deposits on site response.   Tinsley and Fumal (1985) mapped surface geology in the Los Angeles basin to identify loose soils such as Holocene alluvial deposits that are susceptible to strong ground motions.   Similarly, Yamazaki et al. (2000) developed amplification factors based on geomorphology and subsurface geology to estimate expected ground motions.   These studies suggest using geology as an initial regional classification for seismic zonation.   In this study, the use of remote sensing imagery for regional classification is evaluated.   In particular, the objective is to identify Holocene-age deposits that may be susceptible to ground motion amplification.   Site response is then determined for Holocene-age and Pleistocene-age deposits in the Mississippi Embayment based on additional subsurface information.

Holocene-age alluvial deposits in the floodplains are distinguished from loess deposits of Pleistocene/Pliocene age in the inland, terrace regions based on spectral contrast and texture.   Agbu et al. (1990) observed that spectral reflectance is related to subsurface conditions since subsurface conditions affect the properties observed at the surface.   The variation in soil type, moisture content, and geology influences the spectral reflectance and texture.   Therefore, spectral reflectance and texture are the basis for classification in this study.

• Landsat TM Images

• Study Area

• Spectral Classification

• Texture Classification

• Conclusions

• References

The Landsat Thematic mapper (TM) is a multispectral satellite measuring electromagnetic energy in seven spectral bands ranging from the visible to the thermal infrared.   Each pixel represents an area 30 m by 30 m for six of the seven bands whereas pixels in the thermal infrared band represent an area 120 m by 120 m.   An image from the Landsat TM satellite was selected to assess the feasibility of using satellite imagery for identifying regions susceptible to ground motion amplification.   In particular, imagery was analyzed to distinguish between Holocene-age and Pleistocene-age deposits.   Holocene-age deposits are susceptible to ground motion amplification due to the loose, unconsolidated state of deposition.   In the Central United States, Holocene-age deposits are found throughout the floodplains of major rivers.   Pleistocene-age deposits are located in the upland, terrace regions.   Analysis of imagery focused on distinguishing between the two geologic deposits.

The Landsat TM image was obtained from the USGS Earth Resources Observation Systems (EROS) Data Center and georeferenced to the Universal Transverse Mercator (UTM) coordinate system that is based on the North American Datum of 1927.   The image was obtained on November 22, 1986 from the Landsat TM 5 satellite launched in March 1984.   Autumnal images were selected due to the lack of vegetation cover allowing imaging of the surface geology.   A portion of the acquired image is shown in Figure 1.

The study area was selected to evaluate the use of Landsat TM imagery for regional seismic zonation and is located northeast of the NMSZ.   The study area is a subset of the area in Figure 1 and is located in the Jackson Purchase region of western Kentucky.   The study area is bounded by the Ohio River to the northwest and the Mississippi River to the southwest.   Figure 2 shows the selected study area including parts of Kentucky, Missouri, and Illinois and is composed of 1000 by 1000 pixels.

Principal component analysis (PCA) was used to create uncorrelated spectral bands.   The new spectral band with the largest variance was selected for classification and is shown in Figure 3.   This image was classified based on spectral reflectance and texture to distinguish between Holocene-age and Pleistocene-age deposits.

The first approach to classification or segmentation is based on the pixel brightness values or relative spectral reflectance of the image.   Histogram equalization was used to enhance the contrast in the image.   The image in Figure 3 was then passed through a low-pass filter to reduce the effect of cultural boundaries and agricultural features and enhance geologic features.   The image was then classified by image segmentation where low pixel values (dark pixels) were labeled Holocene-age deposits and high pixel values (white pixels) were labeled Pleistocene-age deposits.   The result of this classification is shown in Figure 4.

Texture is related to patterns in pixel brightness values.   Several approaches have been applied to quantify textural analysis including first-order and second-order statistics, directional filters, and fractal geometry.   First-order statistics include calculating the mean and standard deviation of a pixel cluster.   First-order statistics are used in this study to quantify texture and are described below.   The statistics of a 35 by 35 pixel neighborhood were compared with the mean of identified Holocene-age and Pleistocene-age regions.   The minimum Euclidean distance was used to classify pixels.   The results of the texture classification are shown in Figure 5.

The image processing techniques including principal component analysis and histogram equalization improve the spectral contrast and aid in identifying geologic features.   These features may be used to infer the age of geologic deposits based on knowledge of the geomorphology.   For example, meander scrolls due to the lateral migration of the Mississippi and Ohio Rivers are found in alluvial plains and consist of recent, Holocene age deposits.   Features such as meander scrolls may be distinguished from Pleistocene-age deposits by their darker appearance due to fine-grained deposits with a high moisture content or by texture reflected in alternating bands of dark and light pixels from the deposition of fine-grained materials in swales and coarse-grained material in ridges.

Classification based solely on pixel values may not adequately classify most areas.   Qualitative classification based on visual interpretation may be as efficient as a robust quantitative classification.   Higher-order statistical classification methods such as co-occurrence matrices, semi-variograms, directional filters, or fractal geometry may improve classification by using more robust analysis methods.   However, for any classification approach, interpretation must include an understanding of the geology and geologic processes that produce the observed features as well as ground truthing data for verification.   Incorporation of other sources such as radar or high-resolution images may also improve classification.

In areas where soil types are already known or geology maps already exist, remote sensing imagery may add little information about the general subsurface conditions.   In these cases, remote sensing may yield insights into local variations of soil types based on surface features and may be used for calibrating or ground truthing spectral analysis and classification.   For areas lacking existing information, properly interpreted and calibrated remote sensing data may be a valuable source from which to infer geologic and geotechnical conditions.   Since the geology of the Central United States is well documented and various researchers have investigated the subsurface conditions in this region, deposits in the Central United States are classified from these more traditional sources in subsequent chapters.

• Agbu, P.A., D.J. Fehrenbacher, and I.J. Jansen (1990), "Soil Property Relationships with SPOT Satellite Digital Data in East Central Illinois," Journal of Soil Science Society of America, Vol. 54, pp. 807-812.

• Anderson, J.G., Y. Lee, Y. Zeng, and S. Day (1996), "Control of Strong Motion by the Upper 30 Meters," Bulletin of the Seismological Society of America, Vol. 86, No. 6, pp. 1749-1759.

• Borcherdt, R.D. (1994), "Estimates of Site-Dependent Response Spectra for Design (Methodology and Justification)," Earthquake Spectra, Vol. 10, No. 4, pp. 617-653.

• Del Barrio (1855), Proceedings of the University of Chile, translated from old Spanish by R. Dobry.

• Mallet, R. (1862), Great Neapolitan Earthquake of 1857, London, 2 Vols.

• Tinsley, J.C. and T.E. Fumal (1985), "Mapping Quaternary Sedimentary Deposits for Aereal Variations in Shaking Response," Evaluating Earthquake Hazards in the Los Angeles Region - An Earth Science Perspective, J.I. Ziony, U.S. Geological Survey Professional Paper 1360, pp. 101-125.

• Yamazaki, F., K. Wakamatsu, J. Onishi, and K.T. Shabestari (2000), "Relationship between Geomorphological Land Classification and Site Amplification Ratio Based on JMA Strong Motion Records," Soil Dynamics and Earthquake Engineering, Vol. 19, pp. 41-53.

Contents

Summary

• Site Classification Using Remote Sensing Imagery

• Development of Reference Shear Wave Velocity Profiles

• Stochastic Approach for Modeling Rock Motions

• Site Response Analyses

• Attenuation Relationships