The Spatial Dependence of near-Surface Soil Moisture at the Meter Scale.

Near-surface soil water content plays a central role in many important land surface processes. These include: evaporation and plant transpiration; the land surface albedo; the fraction of precipitation or irrigation that infiltrates into the soil; and sensible heat transfer among the soil, vegetation, and atmosphere. A better characterization of the temporal change and spatial structure of soil moisture would lead towards a better understanding of land surface processes and eventually to improved predictions of weather, crop growth, surface run off, tile drainage, and flooding.

The spatial variation of any quantity must be characterized with measurements that represent a unit that is assumed to be homogeneous at spatial scales that are smaller than the spatial scale of interest. For example, ``point measurements'' made either through gravimetric measurement or other methods are often used to quantify the spatial variability of soil moisture. These measurements normally represent a soil volume of a few centimeters.

There are many examples in the scientific literature where point measurements of soil moisture have been used to characterize the spatial variability of soil moisture at scales on the order to 10 to 1000 m. Typically only one or a just a few (two or three) point measurements are made at one location under the assumption that the mean of these measurements adequately represents the soil moisture at the smaller spatial scale.

We hypothesized that significant spatial variability in soil moisture exists at the meter scale that has been overlooked. To test this hypothesis, we made 500 point measurements of near-surface soil moisture randomly-spaced on a centimeter-scale grid over a 9 by 7 m area of bare soil on 6 separate days with varying levels of mean soil moisture. Point measurements were made with a hand-held impedance probe with a sample depth of 6 cm, sample volume of 75 cm^3, and a soil-specific calibration. We assumed that the mean of these 500 measurements was an accurate representation of the true mean soil moisture. We found that:

• a clear spatial dependence of soil moisture exists at the meter scale;
• an exponential variogram model could explain this spatial dependence;
• the sill and range of the variograms varied with mean soil moisture conditions;
• we can predict the number of point measurements needed to estimate the mean soil moisture at the meter scale to within a specified accuracy; and
• the number of point measurements needed to estimate the mean can be reduced if the spatial dependence is exploited.

During the summer growing season of 2010 we will extend our investigation to include vegetation-covered soil. We hypothesize that a simple exponential variogram will not be sufficient to characterize the spatial variability of soil moisture under a vegetation canopy because of the effect of plant roots and the regular variations in soil surface height caused by tillage. In addition to what we have discovered concerning the spatial variability of soil moisture for bare soil, we also intend to present some of our initial findings for soil moisture beneath a soybean canopy.