Having recently been involved with discussions with one of the pioneers of TDR soil water content measurement, who is a soil physicist semi-retired from Agriculture Canada, I can pretty much guarantee that agronomists use volumetric water content, and calculate it from volume of water divided by bulk volume.
In response to jdonville's post, I was also amazed to learn that when the volumetric water content is determined with electromagnetic methods (TDR,GPR, and FDR), and the equipment is calibrated to the particle minerals present in the soil or aggregate, the result is independent of the bulk density or void ratio, and it is unnecessary to have undisturbed samples, etc. It is only when the result has to be referenced to gravimetric values that the problems jdonville points out arise.
This kind of begs an important question or reflection, why do we (geotechnical and civil engineers) think in terms of gravimetric water content, on a dry basis? Is this simply a result of the ease with which this can be determined? Since the mechanical and hydraulic behaviour are more closely related to the degree of saturation than the gravimetric water content maybe we should be looking outside of our box?
The geotechnical/civil convention has been accurately described by others, although I am finding more younger geotechnical engineers with backgrounds in unsaturated soil mechanics or evironmental areas thinking in terms of degree of saturation or volumetric water content.
I have no idea what the roller compacted concrete people use.
As an additional point of confusion, combustion engineers consider the water content of a fuel (e.g., hog fuels) on a gravimetric basis, based on wet total weight.
Talk about a need for some inter-silo translation.