Earth's surface, fashioned by multiple processes operating over different timescales, has more than 5 statistical dimensions of variability. Two-parameter spectral or unifractal models can at best only provide approximations which may serve
as initial surfaces and nul hypotheses. Much of the interest of geomorphology, however, lies in deviations from such models - in scale specificity, allometry, lineation, structuring and non-invertibility. Multifractal models hold much promise, but must