A novel interpolation technique is applied to assessment of the quality of sea surface temperature (SST) observations and quantitative analysis of the subpixel variability within satellite footprints of different size. Using retrieved satellite data as input, the new, global, multistage interpolation technique generates a trigonometric polynomial, providing a representation of the underlying physical SST field in functional form. The resulting interpolating function can be efficiently and accurately evaluated anywhere within the domain over which it was derived and its moments calculated to estimate the mean and variance of the field over desired sub-regions. Application of the technique is demonstrated for SST retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS), Spinning Enhanced Visible and Infrared Imager (SEVIRI), and Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) sensors. Comparison of the functional form with the data from which it was derived demonstrates how the technique can potentially help to identify small observational artifacts such as MODIS scan striping and residual cloud contamination. Integrals of the interpolating functions over successively larger spatial scales successfully emulate the retrieved SST at the different effective spatial resolutions and the second moments are consistent with the direct sample variances, and hence representative of the spatial SST variability of the available finer-resolution observations over the coarser scales. Using the approach, the variability of 1-km-resolution SST observations on open ocean grids of both 5- and 25-km resolution is found to be ~0.07 K. In regions of sharper gradients such as associated with strong localized diurnal warming, the variability within 25-km-resolution grids increases to as much as 0.4 K for sampling at 1-km resolution. The variability of 1-km observations on a 25-km-resolution grid is about 2.4 times greater than that on a 5-km-resolution grid. Broader application of the technique globally could help better quantify regional variations in the spatial variability, which would subsequently improve uncertainty estimates for existing satellite-based SST products.