A conceptual model is proposed for the characteristic sub-ranges in the velocity and temperature spectra in the boundary layer of tropical cyclones (hurricanes or typhoons). The model is based on observations and computation of radial and vertical profiles of the mean flow and turbulence, and on the interpretation of eddy mechanisms determined by shear (namely roll and streak structures near the surface), convection, rotation, blocking and sheltering effects at the ground/sea surface and in internal shear layers. The significant subranges, as the frequency increases, are associated with larger energy containing eddies, shear and blocking, inertial transfer between large and small scales, and intense small-scale eddies generated near the surface caused by waves, coastal roughness change, and the buoyancy force associated with the evaporation of spray droplets. These sub-ranges vary with the locations at which the spectra are measured, i.e. the level z in relation to the height z max of the peak mean velocity and the depth h of the boundary layer, and the radius r in relation to the eyewall radius R ew and the outer-vortex radius R ov . For two tropical cyclones (Nuri and Hagupit), experimental data were analyzed. Spectra were measured where r is near to R ew and R ov using four 1-h long datasets at coastal towers, at 10-and 60-m heights for tropical cyclone Nuri, and at 60-m height for tropical cyclone Hagupit at the south China coast. The field measurements of spectra within the boundary layer show significant sub-ranges of self-similar energy spectra (lying between the length scale 1,000 m and the smallest scales less than 40 m) that are consistent with the above conceptual model of the surface layer. However, with very high wind speeds near the eyewall, the energy of the independently 123 L. Li et al. generated intense surface eddy motions, associated with surface waves and water droplets in the airflow, greatly exceeds the energies of the small scales in the inertial sub-range of the boundary layer, over scales less than about 3-40 m depending on the height z and the radius r . This rise in the small-scale frequency weighted spectra (nS u (n), where n is natural frequency, and S u (n) is the energy spectrum of the longitudinal wind component) is consistent with the hypothesis that these processes are only weakly correlated with the main boundary-layer turbulence.