The polar mesosphere is characterized by a very high radar backscattering coefficient. Both radar and in situ rocket spectra indicate that two qualitatively different scattering mechanisms are operating: one turbulent and one edge-like. Fourier analysis methods applied to in situ data have been very useful in showing that spectra are steeper than the canonical −5/3 turbulent form. Such steep spectra accompany the edge-like structures. However, since Fourier methods assume that the energy is distributed throughout a given window, they are not very useful for describing a finite number of sharp edges in a data set, which is the normal situation. Here we use the techniques of wavelet analysis to partition a rocket data set taken during a polar mesosphere summer echo event characterized by the sounding system radar in Norway. We find that the edges can be either isolated in space or collocated with turbulence and, likewise, that turbulent regions can be found without steep edges. Our partitioning verifies previous notions concerning the interpretation of shallow and steep spectra. The first clear evidence for theoretically predicted viscous-convective and viscous-diffusive subranges is also presented. In a companion paper [Alcala and Kelley, this issue] we use the well-known scattering properties of our wavelet basis functions to calculate the scattering from the region sampled by the rocket probe in a manner that accounts for multiple edges in the same scattering volume.