Soil stratigraphy from seismic piezocone data and multivariate clustering in alluvial soil deposits: Experience in the Lower Tagus Valley region

Molina-Gómez, Fausto; Cordeiro, D.; Ferreira, C.; Fonseca, A. Viana da

doi:10.1201/9781003308829-84

Cited by 2 publications

(10 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For sequentially ordered data such as CPT data, the nearest neighbor matrix is tri-diagonal with ones on the diagonal and the two adjacent diagonals, and zeros elsewhere. This differs from the approach of Molina-Gómez et al 8 that used a more fully populated nearest neighbor matrix. In our approach, a particular data point is constrained to belong to the same cluster as the point above and the point below (or both) or to constitute its own cluster, but it cannot belong to the same cluster as a distant neighbor unless all of the points in between are part of the same cluster.…”

Section: Agglomerative Clusteringmentioning

confidence: 86%

“…Thus, results from this method are not reported in Figure 7. Molina-Gómez et al 8 utilize the silhouette method to define the number of clusters in their algorithm. We also apply an alternative method in which K is selected as the point where J (from Equation 13) is minimized.…”

Section: Elbow and Min(j) Methodsmentioning

confidence: 99%

“…Ntritsos and Cubrinovski 7 caution that the algorithm may result in fictitious layers at layer boundaries and indicate that their algorithm is not intended to replace engineering judgment. Molina-Gómez et al 8 more recently utilized a multivariate hierarchical clustering approach to identify stratigraphic layers at a site in the Tagus River Valley where gel push sampling was performed in combination with CPT testing to confirm soil types. Layers need not be vertically continuous in their algorithm (e.g., a layer may have another layer within it).…”

Section: Existing Layering Algorithmsmentioning

confidence: 99%

“…9,10 Such profiles are required for the development of new liquefaction triggering and manifestation models. Our algorithm was developed independently from, and concurrently with, the methods by Ntritsos and Cubrinovski 7 and Molina-Gómez et al 8 and bears some similarities to both methods, as well as having some advantages. It is similar to the method by Ntritsos and Cubrinovski 7 in that it seeks a set of layers that reduces the within-layer variance.…”

Section: Motivation For Automated Layer Identification Algorithmmentioning

confidence: 99%

“…It differs from their method in that it uses unsupervised machine learning rather than prescribed rules for assigning layers, which is advantageous since the algorithms are widely available in Python packages. Our method is similar to that of Molina-Gómez et al 8 in that it utilizes an unsupervised machine learning technique, hierarchical clustering, to identify layers. However, it differs from their method in two important respects.…”

Section: Motivation For Automated Layer Identification Algorithmmentioning

confidence: 99%

See 4 more Smart Citations

Unsupervised machine learning for detecting soil layer boundaries from cone penetration test data

Hudson

Ulmer

Zimmaro

et al. 2023

Earthq Engng Struct Dyn

View full text Add to dashboard Cite

Cone penetration test (CPT) data contains detailed stratigraphic information that is useful in a wide variety of applications. Separating a CPT profile into discrete layers is an important part of many analyses such as critical layer selection in liquefaction triggering analysis, effective stress seismic ground response analysis, analysis of pile shaft and tip resistance, and soil‐pile interaction analysis. The discretization of the profile into layers is often done manually, relying on the judgment of the analyst. This manual approach is cumbersome for datasets that include large numbers of CPT profiles (such as the Next Generation Liquefaction [NGL] database and the New Zealand Geotechnical Database) and it may not be consistent or repeatable because different analysts may discretize a given CPT log in different ways. To overcome these difficulties, we present an approach to automatically divide a CPT profile into discrete layers. Automated layer detection is performed using an unsupervised machine learning technique called agglomerative clustering in combination with two cost functions to identify an optimal number of layers. The algorithm is illustrated using CPT profiles from the NGL database, where the approach is being used in the development of liquefaction triggering and manifestation models. Although the algorithm shows promise for replicating our judgment regarding layering, we recommend visual review of the layering produced by the algorithm to check for reasonableness given the site geology and intended use of the CPT data.

show abstract

Section: Agglomerative Clusteringmentioning

confidence: 86%

Section: Elbow and Min(j) Methodsmentioning

confidence: 99%

Section: Existing Layering Algorithmsmentioning

confidence: 99%

Section: Motivation For Automated Layer Identification Algorithmmentioning

confidence: 99%

Section: Motivation For Automated Layer Identification Algorithmmentioning

confidence: 99%

See 3 more Smart Citations

Unsupervised machine learning for detecting soil layer boundaries from cone penetration test data

Hudson

Ulmer

Zimmaro

et al. 2023

Earthq Engng Struct Dyn

View full text Add to dashboard Cite

show abstract

A cone penetration test database for multiple thin-layer correction procedure development

Yost,

Yerro,

Martin

et al. 2024

Earthquake Spectra

View full text Add to dashboard Cite

Cone penetration tests (CPTs) are a commonly used in situ method to characterize soil. The recorded data are used for various applications, including earthquake-induced liquefaction evaluation. However, data recorded at a given depth in a CPT sounding are influenced by the properties of all the soil that falls within the zone of influence around the cone tip rather than only the soil at that particular depth. This causes data to be blurred or averaged in layered zones, a phenomenon referred to as multiple thin-layer effects. Multiple thin-layer effects can result in the inaccurate characterization of the thickness and stiffness of thin, interbedded layers. Correction procedures have been proposed to adjust CPT tip resistance for multiple thin-layer effects, but many procedures become less effective as layer thickness decreases. To compare or improve these procedures and to develop new ones, it is critical to have pairs of measured tip resistance ( q m) and true tip resistance ( q t) data, where q m is the tip resistance recorded by the CPT in a layered profile, and q t represents the tip resistance that would be measured in the profile absent of multiple thin-layer effects. Unfortunately, data sets containing q m and q t pairs are extremely rare. Accordingly, this article presents a unique database containing laboratory and numerically generated CPT data from 49 highly interlayered soil profiles. Both q m and q t are provided for each profile. An accompanying Jupyter notebook is provided to facilitate the use of the data and prepare them for future statistical learning (or other) applications to support multiple thin-layer correction procedure development.

show abstract

Soil stratigraphy from seismic piezocone data and multivariate clustering in alluvial soil deposits: Experience in the Lower Tagus Valley region

Cited by 2 publications

References 0 publications

Unsupervised machine learning for detecting soil layer boundaries from cone penetration test data

Unsupervised machine learning for detecting soil layer boundaries from cone penetration test data

A cone penetration test database for multiple thin-layer correction procedure development

Contact Info

Product

Resources

About