In this paper, we determine the apparent soil thermal diffusivity of a dense sandy soil of the HAPEX-Sahel experiment. Several current methods are compared. Two temperature data sets are chosen. In the first, the assumption of steady periodicity is fulfilled and in the second, it is not. In both cases, we compare methods which assume a vertical homogeneity of the soil thermal properties with the NHS method which is based on the vertical inhomogeneity of the thermal diffusivity. Results obtained with both sets of data show that thermal properties are not homogeneous vertically. It is shown that the NHS method is not applicable when the steady periodicity assumption is not valid. In this case, when abrupt change in the temperature wave pattern occurs, as frequently happens in the Sahel just before or just after a rain, a method based on the Lapalce Transform with a corrective factor for the non uniformity of the initial temperature profile (CLTM) must be used. When the steady periodicity assumption is fulfilled, both the Harmonic (HM) and the CUM methods lead to somewhat greater values of the thermal diffusivity than the NHS method.