2D ≠ 3D: CONVERTING GRAIN SIZE DISTRIBUTIONS MEASURED IN THIN-SECTION TO VOLUMETRIC GRAIN SIZES
Grain sizes of most granular materials have approximately log-normal distributions, and larger grains have a much greater impact on the volumetric calculation than they have on area calculations. In addition, random planar slicing of statistically spherical objects produces a mean surface diameter that is (i.e., 0.866) of the true mean grain diameter. Correcting for these two factors gives the volumetric mean grain size: Mv = Mts + logβ(2/√3) + ln(β) σβ2, where Mv and Mts are the logarithmic values for volumetric grain size and thin-section grain size means, and β and σβ are the log base and standard deviation in that log base. Logarithmic standard deviation is constant, i.e. σv = σts. For σ = 1.0Φ, the correction of mean size is Mv = Mts - 0.9Φ, and for σ = 0.5Φ, the correction is Mv = Mts - 0.38Φ.
Where there is a bimodal distribution, the two populations need to be treated separately, i.e. as separate populations with distinct means and standard deviations. Where large irregular objects (e.g., shells) lie within the grain package, estimation of the length normal to the plane of the thin section is necessary to properly estimate the proportional volume. Usually in such cases, there are few enough of those objects that the volumetric calculation can be made for each object and the results summed to get an approximation of the true volumetric proportion.