Soil water retention parameters in the dry range—what is the physics?

Low-water content soil moisture relations are increasingly important given current trends of climate change, desertification, and growing interest in extreme environments like those of Antarctica and Mars. Whereas most parametric models of soil water retention were developed for the intermediate and wet ranges of moisture, some alternatives published in the last three decades address water retention down to oven dryness. Such models can be strengthened and made more versatile with a deeper understanding of the physical meaning of parameters used in them.

For the shape of the dry-range retention curve, a logarithmic relation has repeatedly been shown to work well, and is consistent with accepted theories of adsorption. Fitted values of the log function’s coefficient relate closely to the specific surface area of the medium.

The lower limits of water content and matric potential require more explication. Various observers have noted problems that arise with the use of a nonzero residual water content as the lower limit. In practice, this quantity is not measured but obtained as a fitting parameter, whose value depends not on a physical property but on how far the available measurements extend into the dry range. In parametric models it can be useful for applications in which the water content never goes below the intermediate range dominated by capillary processes.

The actual lower limit of water content depends on how its zero is defined. The most common definition is based on equilibration in an oven at a particular temperature, commonly 105° C. Ambiguity arises from the dependence of the soil water on the generally uncontrolled relative humidity within the oven. Application of the Kelvin equation with reasonable assumptions about the outside air and its exchange with the inside air can indicate an equivalent matric potential of the oven-dry state, typically about -1 GPa. Logarithmic extrapolations of dry-range retention measurements intersect the water content=0 axis at values comparable to this, with variations likely related to particular conditions in the lab and oven. A way of resolving this ambiguity is to define zero water content not in terms of oven temperature but rather a specified matric potential of equilibration. An attractive possibility, convenient in SI units, is to make it exactly -1 GPa.

Neither the traditional nor this proposed definition of zero water content identifies a state where no water molecules remain in the soil. Measurements at temperatures of hundreds of degrees C show that soil water contents can be lower than these defined zero levels by as much as 2% or more. Our standard scale of water content, therefore, is a relative scale, analogous to the Celsius scale for temperature. Consequently negative values of soil water content have a valid physical meaning. To acknowledge this fact and resolve ambiguities, more rigorous definitions as proposed here are thus necessary for applications dealing with the extremely dry conditions that are becoming increasingly important.