The hot spots in the satellite images in IR channels could represent the fire pixels in the images, in which the temperature on the hot spot could be tens of degrees higher than that for the neighboring pixels. If the edges on hot spots in satellite images are significantly sharper, then the FFTR resampling also creates the artifacts in the neighboring pixels around the hot spots. Figure 8 shows an example of the artifacts, in which the shift value is 0.5 that represents the maximum resampling uncertainty in the FFTR algorithm. The data in Fig. 8 are the local area of the full Earth scan with the size of 5208, and the hot spot corresponds to the dark pixel near the center in Fig. 6. In this case, the pixel count difference between the neighboring pixels could be as high as 250 counts for the hot spot pixels, while the average difference between the neighboring pixels counts is less than 10 counts. The hot spot pixels in a scan line on a satellite image, in fact, could be regarded as a discontinuity point in the data, and the Fourier expansion does not produce the smooth behavior between the two neighboring pixels around hot spots. The output of the FFTR algorithm shows the sinc function, , behavior in the neighboring pixels around the discontinuous point for the noninteger shifts. This behavior is most significant with the shift value . The sinc function behavior for discontinuity points is generally true for the FFTR algorithm, and the discontinuity points may correspond to the hot pixels in the satellite images or the images with sharper edges that need to be treated differently.