Use a log scale when:

• Your data spans multiple orders of magnitude.
• The process shows exponential growth or decay.
• Relative change (percent increase or fold-change) matters more than absolute difference.
• You need to reveal patterns hidden on a linear scale.
• Your data has a high dynamic range.

Example scenarios include the following: contaminant concentrations, grain-size distributions, hydraulic conductivity, earthquake magnitudes, frequency distributions, or anything that grows or declines multiplicatively.