Chapter 83 Managing and Analyzing Geological Data

Year 4 · Thesis / Project Report · G447-L2

Preserve raw geological data, document processing and present analyses with suitable graphs, quality checks and uncertainty.

A repeatable workflow from a testable question to an evidence-based interpretation. Original Geology Hub teaching diagram.

This independent lesson follows the topic framework of the supplied University of Peshawar BS Geology curriculum, session 2004–05 onward. It is newly written study material, not an official university course or qualification. The source supplies broad objectives for this unit; the lesson structure, exercises and examples below are newly authored extensions.

Learning objectives

Preserve raw data and document transformations.
Apply quality checks before interpreting patterns.
Present numerical and spatial results with uncertainty.

Reproducibility starts with a recoverable record

Keep original measurements or downloaded files unchanged and create separate working copies. Assign stable identifiers linking samples, observations, photographs and analytical results. A data dictionary defines each column, unit, allowed value and missing-data code. Zero, not detected and not measured have different meanings. Conflating them can change averages, maps and conclusions. Record source, acquisition date, method and relevant precision so a later analyst understands the values.

Document every transformation, including unit conversions, exclusion criteria, coordinate changes and recalculated variables. A script or carefully logged spreadsheet can preserve the sequence from raw values to final figures. Record software and method versions where relevant. A clean-looking table is not enough if no one can reconstruct how it was produced. Review formulas and intermediate results using a small hand calculation before applying a workflow to the full dataset.

Analyze the question rather than every possible pattern

Check for impossible values, duplicate identifiers, inconsistent units and uneven sampling before statistical analysis. Plot distributions and spatial coverage. Mean and median describe different aspects of a dataset, especially where distributions are skewed or contain outliers. An outlier may be an error, a real unusual observation or a separate population. Investigate it with documented criteria rather than deleting it because it weakens the expected pattern.

Choose graphs that match the comparison. A scatterplot examines two measured variables; a distribution plot compares spreads; a map shows spatial relationships with reference system and scale. Uncertainty includes instrument precision, sampling variation and model assumptions. Correlation does not by itself establish causation. For geological interpretation, consider confounders such as facies, weathering, stratigraphic position and collection method. Present a result before proposing the process that explains it.

A data table must preserve meanings and units

Define each variable before analysis. A column called thickness could mean vertical thickness, true stratigraphic thickness or an incomplete borehole penetration. A number without that distinction is not ready for a geological comparison. The data dictionary should specify the definition, unit, measurement method, valid range and missing-value codes. A field recorded as not visible in a photograph is different from a measured absence in an exposed rock.

Keep source identifiers in the working table so each result can be traced to its original image, specimen or measurement. Unit conversions should be explicit: 25 centimetres is 0.25 metre, not 25 metres copied into a column with a new heading. Preserve the original unit and value alongside the converted quantity or in an unchanged raw file. This makes a later correction possible without reconstructing the evidence.

Quality checks distinguish errors from geological variation

Check duplicate identifiers, inconsistent location references, impossible percentages and values outside the instrument range. A porosity fraction of 0.18 and a porosity percentage of 18 are equivalent only after recognizing their different scales. A missing decimal can create an apparent geological anomaly. Automated checks help flag such values, but the original record must decide whether the value is an error or a valid unusual observation.

Repeated measurements quantify one kind of variability; independent sites quantify another. Ten grain measurements from one specimen describe that specimen more densely but do not substitute for ten specimens from a formation. State the level of sampling at which a comparison is made. Otherwise, a large numerical sample count can imply more independent geological evidence than the design actually contains.

Presenting distributions and uncertainty honestly

For the synthetic values 1, 2, 2, 3 and 12, the median is 2 and the mean is 4. The high value influences the mean, but it should not be removed merely to make the two agree. Show the distribution or individual observations so the reader can see that effect. A summary number is useful when paired with information about spread and sample size.

Measurement uncertainty differs from natural variability and model uncertainty. A precise ruler does not make five poorly representative sites describe a basin. Conversely, a variable rock unit can be measured carefully even though its observations have a wide spread. Distinguish these sources in figure captions and interpretation. Error bars should state what they represent rather than appearing as an unexplained symbol of rigor.

Reproducibility requires an inspectable processing path

Record exclusion decisions, transformations, formula definitions and analysis settings. A spreadsheet should expose inputs and calculations rather than replace results with unexplained pasted numbers. Test a formula on a small hand-calculated case before applying it across the dataset.

Finally, ask another learner to reproduce one figure from the original table and written instructions. A failure identifies missing documentation, not automatically a failed geological result. Repairing that path improves the reliability of every later interpretation.

Worked activity

Use the explicitly synthetic measurements 2, 3, 3, 4 and 18 millimetres. The mean is 6 millimetres and the median is 3 millimetres. Neither calculation is wrong, but they summarize the skewed set differently. Check the original record for 18 rather than silently removing it. If it is valid, report its presence and explain why the median better describes the middle observation in this particular small set. Label every table and figure synthetic practice data so these values cannot be mistaken for a real field result.

Practice and interpretation

Build a miniature reproducibility package with raw table, data dictionary, processing log and final figure. Ask a peer to recalculate one result without verbal help. If that is impossible, improve the documentation. In a spatial project, include the coordinate reference and the distinction between observed and interpolated values. This exercise assesses whether the interpretation can be traced to evidence, even when the analysis uses only a small desk-based dataset.

Self-check

Why keep raw data unchanged?
Is an outlier automatically an error?
Does correlation demonstrate cause?

Check your answers

They preserve the original evidence and make processing decisions reversible and reviewable.
No. Its origin must be investigated rather than assumed.
No. Alternative explanations and study design must be considered.

Sources and further study

The explanations and teaching examples are original. Use these institutional resources to check terminology and explore the subject further.

Chapter 83 multiple choice questions

Choose one best answer for each question. Open each answer panel after making your choice.

For the synthetic values 2, 3, 3, 4 and 18 millimetres, what are the mean and median?
1. Mean 6 millimetres and median 3 millimetres
2. Mean 3 millimetres and median 6 millimetres
3. Mean 5 millimetres and median 4 millimetres
4. Mean 6 millimetres and median 4 millimetres
Answer and explanation
A. Mean 6 millimetres and median 3 millimetres The sum is 30, so the mean is 30/5 = 6. The ordered middle value is 3.
A dataset uses zero for both not measured and an actual zero result. What is the main problem?
1. The coding guarantees unbiased averages
2. Distinct meanings are being combined and may distort analysis
3. Missing values automatically become exact measurements
4. The dataset no longer requires a data dictionary
Answer and explanation
B. Distinct meanings are being combined and may distort analysis Zero and missing observations carry different information. The lesson recommends explicit missing-data definitions.
An unusually large value weakens the expected pattern. What should the analyst do first?
1. Delete it solely because it weakens the hypothesis
2. Change it to the median without a record
3. Investigate its original record using documented criteria
4. Declare it impossible without checking its source
Answer and explanation
C. Investigate its original record using documented criteria An outlier may be an error, a real unusual value or another population. Its treatment should follow evidence and transparent criteria.
Which workflow best supports reproducibility?
1. Replace the original measurements with the final rounded values
2. Keep only the final chart because it is easier to read
3. Record units only if a reviewer asks later
4. Retain unchanged raw data and document transformations in separate working products
Answer and explanation
D. Retain unchanged raw data and document transformations in separate working products Preserved originals and documented processing let another analyst reconstruct the results. Final products alone hide decisions.
Two geological variables are correlated in an observational dataset. Which next step is most appropriate?
1. Examine possible confounders and alternative explanations
2. Announce that one necessarily causes the other
3. Assume the strongest correlation removes sampling bias
4. Treat spatial coverage and facies as irrelevant
Answer and explanation
A. Examine possible confounders and alternative explanations Correlation alone does not establish causation. Facies, weathering, collection methods and other factors may influence the relationship.

All 84 chapters

Search This Blog

Geology Hub