Multiscale surface roughness description for scattering modelling of bare soil

Multiscale surface roughness description suitable for natural targets such as agricultural fields is studied. Special attention is given to randomly rough, essentially periodic surfaces. Bare soil measurement results are demonstrated to verify the theoretical methodology. It turned out that bare soil shows multiscale behaviour even up to . The main idea of the multiscale surface roughness description is that the surface roughness is thought to be a superposition of various wavelengths. The roughness component included can (1) cover a semi-infinite band starting from zero or (2) consist of one or more finite wavelength bands. It is also possible to combine separate distinct periods or period bands with random roughness of various sizes. The basic assumptions used here are: (1) the correlation length depends linearly on distance and (2) the logarithm of the rms height depends linearly on the logarithm of distance. The shape of the single scale surface autocorrelation (and thus also the shape of the power spectrum) is left open, although only cases of gaussian, exponential and transformed exponential autocorrelation are studied explicitly. It is also possible to combine different autocorrelation types for different bands, which indeed turned out to be useful in the case of the harrowed field. For the ploughed field combining a distinct period with multiscale random roughness described the surface roughness well. The description of the surface roughness of the ploughed field in a direction slightly deviating from the row direction requires extra attention.