We have performed power spectral analysis of surface temperatures, velocities, and magnetic fields, using spectropolarimetric data taken with the Hinode Solar Optical Telescope. When we make power spectra in a field of view covering the supergranular scale, kinetic and thermal power spectra have a prominent peak at the granular scale while the magnetic power spectra have a broadly distributed power over various spatial scales with weak peaks at both the granular and supergranular scales. To study the power spectra separately in internetwork and network regions, power spectra are derived in small subregions extracted from the field of view. We examine slopes of the power spectra using power-law indices, and compare them with the unsigned magnetic flux density averaged in the subregions. The thermal and kinetic spectra are steeper than the magnetic ones at the subgranular scale in the internetwork regions, and the power-law indices differ by about 2. The power-law indices of the magnetic power spectra are close to or smaller than –1 at that scale, which suggests the total magnetic energy mainly comes from either the granular scale magnetic structures or both the granular scale and smaller ones contributing evenly. The slopes of the thermal and kinetic power spectra become less steep with increasing unsigned flux density in the network regions. The power-law indices of all the thermal, kinetic, and magnetic power spectra become similar when the unsigned flux density is larger than 200 Mx cm–2.