Simulation Study of Sampling Experiment over Washita ’92 Soil Moisture Field

Abstract

This study evaluates the studies by Kim, Yoo, and Yoo and Ha on the sampling of soil moisture to show if their studies are applicable to field experiments. The objective of this study was accomplished by a simulation of sampling experiments over Washita ’92 soil moisture data field to estimate the sampling errors, and by comparing them with those analytically derived. As a result of the study we found that the ensemble means of sampling errors estimated by repeated simulations of the sampling experiment agree well with those analytically derived. Also, sampling errors from evenly spaced and from randomly located sensor networks do not have any obvious difference. These findings indicate that the formulations by Kim, Yoo, and Yoo and Ha for the sampling errors are reliable enough to be used for evaluating the sampling networks as well as for planning purposes. Also, the model for soil moisture dynamics, even though it is a simple diffusion model with only two parameters, may have no serious problems and can be used for sampling related studies. However, the variance of the mean square sampling errors from the evenly spaced sensor networks was found to be much smaller than those from the randomly located sensor networks. This indicates the advantage of using evenly spaced sensor networks even though the expected mean square sampling error value is the same. Similar results can also be found for the soil porosity field.