Geotechnical Data Analysis

Geotechnical data analysis is the engineering interpretation step between site investigation and design. It starts with measured or observed information such as boring logs, SPT blow counts, CPT soundings, groundwater readings, laboratory tests, rock core logs, and field observations. The goal is to determine what the ground is likely to do when it is loaded, excavated, saturated, compacted, drained, or left in service.

The output is not just a spreadsheet of test values. A useful analysis produces an interpreted subsurface profile, design groundwater assumptions, soil and rock parameter ranges, construction considerations, and risk flags. Those outputs support downstream topics such as foundation design, settlement analysis, retaining wall design, earthworks, and slope-related decisions.

Most geotechnical projects combine several data sources because no single test fully describes the ground. A boring log may show soil layering, but it does not fully define strength. A lab test may measure a sample precisely, but the sample may not represent the controlling field condition. Good analysis compares multiple lines of evidence.

This is why soil testing and soil mechanics sit directly upstream of data analysis. Testing produces measured properties, while soil mechanics explains what those properties mean under load, water, and time.

Boring logs, SPT results, and CPT soundings are often reviewed together because each one tells a different part of the story. A boring log provides visual descriptions and samples. SPT data provides discrete penetration resistance values. CPT data provides a more continuous resistance profile. The strongest interpretation looks for agreement between the three instead of relying on one source alone.

One of the most important skills in geotechnical data analysis is separating factual information from interpretation. Factual data is what was observed, measured, logged, or tested. Interpretive information is the engineer’s conclusion based on that evidence.

A strong geotechnical report should make this distinction clear. Readers should be able to trace how factual data led to the interpreted model, how the model led to selected parameters, and how those parameters led to design recommendations.

A strong analysis workflow moves from completeness to interpretation to design relevance. The order matters because selecting design parameters before checking data quality can create false confidence in numbers that are not representative.

1. Organize field, lab, and groundwater records

Start by grouping boring logs, sample records, lab reports, field test results, groundwater observations, and construction or monitoring data by location and depth. This makes it easier to compare trends across the site instead of treating each result as an isolated number.

2. Check quality before interpreting trends

Review sample condition, missing intervals, inconsistent classifications, questionable groundwater readings, unusual blow counts, and lab values that do not match visual descriptions. A technically precise test result can still be misleading if the sample was disturbed or not representative.

3. Build an interpreted subsurface model

Use the evidence to define likely layers, groundwater conditions, weak zones, fill limits, weathered rock surfaces, and zones of uncertainty. The model should be detailed enough to support design but honest about where the data is sparse.

The most common mistake in reading subsurface data is assuming layer boundaries are known continuously across the site. In reality, those boundaries are inferred between borings and should be treated with more caution where the spacing is wide or the geology is variable.

4. Select design parameters for the actual problem

Choose strength, compressibility, permeability, unit weight, modulus, and groundwater assumptions based on the controlling design condition. The right parameter for a temporary excavation may not be the right parameter for long-term settlement or lateral earth pressure.

5. Document assumptions, uncertainty, and design sensitivity

The final analysis should make clear which data controlled the interpretation, which assumptions were conservative, where uncertainty remains, and which design decisions are most sensitive to parameter changes.

Real geotechnical data is rarely smooth. SPT values jump, CPT traces change quickly, lab results scatter, groundwater readings vary, and soil boundaries shift between borings. The job of analysis is not to erase that messiness. The job is to decide what the scatter means for the project.

Trends are usually more important than single values

Engineers compare data by depth, elevation, layer, and location. A single low SPT value may be less important if it is isolated and unsupported by other evidence. The same value becomes more important if nearby borings, CPT data, samples, or lab tests show a consistent weak zone.

Outliers are not automatically errors

An outlier may be a test problem, but it can also represent gravel, cemented material, fill, organic soil, a soft seam, a weathered rock pocket, or a localized construction risk. Discarding it without investigation can remove the most important data point in the dataset.

Averages can hide controlling conditions

Averaging all results in a layer may produce a value that looks statistically clean but ignores the weak portion that controls settlement, bearing, slope stability, or excavation performance. In geotechnical work, lower-bound behavior, characteristic values, and design sensitivity often matter more than a simple average.

Design parameters are selected values used in calculations, models, and recommendations. They are not simply copied from the highest, lowest, or most recent test result. Engineers usually evaluate ranges, data quality, correlations, field observations, and project consequences before choosing values.

Data quality review should identify conditions that may change the interpretation. A red flag does not automatically mean the data is wrong, but it does mean the engineer should pause before relying on it for design.

Consider a small building site with four borings across the footprint. The borings show 3 to 6 ft of variable fill over clay, a sand layer below the clay, and groundwater observed between 12 and 15 ft. Two borings also show a loose sand lens with lower penetration resistance than nearby locations.

Data interpretation

The engineer would not simply average all values across the site. The fill would be treated as a separate material because it may be uncontrolled. The clay layer would be reviewed for settlement potential using water content, index tests, and consolidation data. The loose sand lens would be flagged as a possible weak or variable zone, especially if it lies below heavily loaded columns.

Engineering meaning

The interpretation may lead to shallow foundations with settlement checks, selective undercutting of poor fill, additional borings near heavy loads, or a recommendation to evaluate ground improvement. Groundwater would also affect excavation planning and the assumptions used for effective stress, seepage, and construction staging.

Digital tools can make geotechnical data easier to manage, plot, compare, and review. Spreadsheets are often used for quick comparisons, parameter summaries, and simple plots. CAD, GIS, and modeling tools help relate boring locations to site geometry. Dedicated geotechnical software and databases help manage logs, lab results, cross sections, and monitoring data.

Software improves organization, but it does not replace interpretation. A clean database can still contain poor samples, unstable groundwater readings, wrong units, missing metadata, or overconfident layer boundaries. The engineer still has to decide whether the data is reliable enough for the design problem.

• Spreadsheets: Useful for parameter tables, outlier checks, summary statistics, and plotting test results by depth or layer.
• GIS and CAD: Useful for comparing boring locations, structure footprints, slopes, utilities, drainage patterns, and cross-section alignments.
• Geotechnical databases: Useful for storing boring logs, lab data, metadata, test results, and project history in a traceable format.
• Monitoring dashboards: Useful for reviewing settlement, pore pressure, inclinometers, and construction-stage performance trends.

A separate geotechnical software workflow may be useful when projects involve many borings, large datasets, repeated reporting, complex cross sections, or long-term instrumentation.

Geotechnical data analysis is most valuable when it changes a design choice, construction plan, or risk response. The same dataset may be interpreted differently depending on whether the project is controlled by bearing, settlement, excavation stability, seepage, slope movement, or constructability.

• Foundation selection: Data may show whether shallow foundations are feasible or whether deep foundations, ground improvement, or a mat foundation should be evaluated.
• Settlement control: Compressibility data and soft-layer thickness determine whether movement is tolerable or whether preload, removal, replacement, or ground improvement is needed.
• Retaining and excavation systems: Strength, groundwater, and soil layering affect earth pressure assumptions, drainage details, temporary support requirements, and movement risk.
• Earthwork and subgrade decisions: Classification, moisture-density behavior, organic content, and field proofing influence whether soil is reusable, needs conditioning, or should be undercut.
• Monitoring and observational updates: Settlement points, piezometers, and inclinometers help confirm whether the original model is performing as expected.

Geotechnical data analysis is judgment-heavy because soil and rock are natural materials. A boring may miss a buried channel, fill lens, perched water condition, weathered seam, or soft pocket. A lab result may be accurate for the tested specimen but not representative of the controlling field zone.

Experienced engineers look for agreement between independent clues. A soft clay layer should make sense in the boring description, index tests, water content, strength data, consolidation response, and site geology. If those clues disagree, the answer is not to blindly average the numbers; it is to understand why the evidence conflicts.

Geotechnical data analysis becomes unreliable when the model becomes more detailed than the data can justify. The risk is not only a wrong calculation; it is a confident design built on weak assumptions. This is where geotechnical risk assessment becomes important.

• Sparse exploration: Widely spaced borings may miss localized soft zones, undocumented fill, variable rock elevation, buried debris, or groundwater changes.
• Poor sample quality: Disturbed or dried samples can produce misleading strength, compressibility, and moisture-related results.
• Unstable groundwater assumptions: One-time readings may not represent seasonal highs, perched water, delayed stabilization, or construction-period seepage.
• Misused correlations: SPT, CPT, and index-property correlations are useful screening tools, but they are not substitutes for judgment and project-specific validation.
• Unclear design objective: A parameter chosen for short-term undrained strength may not be appropriate for long-term drained performance, settlement, or seepage.

Most interpretation errors come from treating geotechnical data as more certain, more continuous, or more transferable than it really is. A clean table of numbers can hide weak sampling, inconsistent assumptions, or a site model that does not match field behavior.