Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


A method of indirect measurement of the apparent thermal diffusivity of soil, in situ, was developed, tested, and evaluated. The method was based on an analysis of diurnal variations in soil temperature measured at several depths near the surface of an undisturbed mineral soil (a silty clay loam), near Baton Rouge, Louisiana. A thermocouple insertion technique was developed and implemented to precisely locate the sensors in undisturbed soil, thereby reducing experimental error in soil temperature measurement. The apparent thermal diffusivity was determined from an implicit relationship derived from transient one dimensional heat conduction theory. Nonlinear regression analysis was used to select the estimate of apparent thermal diffusivity as a function of observed soil temperature profiles. The regression model included a Fourier series representation of the upper boundary condition. Increasing the number of harmonics in the series beyond one did not have a significant effect on the estimate of apparent thermal diffusivity. Exclusion of observations of temperature near the surface affected the estimate of apparent thermal diffusivity, which could be an indication of nonhomogeneous soil properties. Further evidence of nonhomogeneity on some days was detected by noting that the residuals from the regression, varied sinusoidally in time (with a 24 h period) at one or more depths. Best results were obtained when nearly periodic temperatures were observed, during a succession of sunny, rain-free days. The results of the nonlinear regression method were compared with three simpler methods which used daily maximum and minimum soil temperature data as input. The nonlinear regression method, which used hourly values of soil temperature, was substantially more precise. The results from the nonlinear regression method also agreed fairly well with the estimated range of thermal diffusivity predicted by the De Vries (1963) method. Overall, the method was easy to perform experimentally, required somewhat complex computation, and yielded precise results.