AN OPTIMUM INTERPOLATION METHOD APPLIED TO THE RESAMPLING OF NOAA AVHRR DATA

Two main problems must be solved in the geometric processing of satell ite data: geometric registration and resampling. When the data must be geometrically registered over a reference map, and particularly when the output pixel size is not the same as the original pixel size, the quality of the resampling can determine the quality of the output, not only in the visual appearance of the image, but also in the numerical ly interpolated values when used in multitemporal or multisensor studi es. The ''optimum'' interpolation algorithm for AVHRR data is defined over a 6 x 6 window in order to consider overlapping effects among adj acent pixels. The response for each new pixel R(x, y) is determined as a linear combination of the response R(i)(x(i), y(i)) of the surround ing pixels in the window (i = 1, 36). The weighting coefficients mu(i) are calculated from the ground projection of the effective spatial re sponse function for each AVHRR pixel, taking into account the particul ar viewing angle and geometry of the pixels on the ground. This method is intended to give an optimal interpolation of AVHRR scenes along al l the scanline, in order to compensate for off-nadir radiometric alter ations associated to the varying spatial resolution (change in the IFO V size and shape on the ground) and the blurring introduced by the pix el overlaps. The optimum method, as mathematically defined, is highly expensive in CPU time. Then, a big effort is necessary to implement th e algorithms so that they could be operationally applied. Two approach es are considered: a general numerical method (assuming a realistic sp atial response for function) and a pseudoanalytical approximation (ass uming a simplified Gaussian pulse as spatial response function). The a nalytical method requires only 2% of the CPU-time required by the full y numerical approach. Some examples are given by comparing the optimum interpolation technique with some other traditional methods. A Landsa t TM image corresponding to the same date of the AVHRR image is used t o test the quality of the radiometric interpolation procedure. The mai n advantage of the optimum interpolation is given by the fact that the resulting interpolated image ''loses the memory'' of the original pix el spacing in the image, which is not true for classical interpolation approaches.