Abstract
The images in a short wave channel 15 (4.45µm) of GOES 15 Sounder show considerable striping effects. The study by ITT Exelis group showed that these striping effects can be characterized by the detector to detector (D2D) and scan to scan (S2S) stripings. The D2D striping is caused by the overall sinusoidal oscillation behavior across east-west scan direction for each detector with the fixed wavelength of 350 pixels. The phases of the sinusoidal oscillation for detector 1 and 3 is opposite to that for detector 2 and 4 so that the striping effects reaches to the maximum when the oscillation magnitude for 1 and 3 (1+3) or 2 and 4 (2+4) are at maximum. The D2D is location dependent as well as time dependent with a diurnal behavior. The S2S striping is overall the difference of same detector between East to West and West to East scans. This striping feature is not location dependent in an image, however, it is time dependent with the diurnal behavior.
A real time de-striping algorithm implemented in GOES ground systems is presented in this study. The algorithm corrects both D2D and S2S stripings on a scan by scan basis to minimize the impact on the latency requirements of the GOES Sounder data processing in the ground system. The de-striping is done sequentially with D2D striping being corrected first. The algorithm retrieves the sinusoidal oscillation function from an offset function between detectors 1+3 and 2+4 through the Fast Fourier transformation (FFT), as the offset function is dominated by the D2D striping function. The D2D striping is corrected for a detector in the scan by subtracting the value of the oscillation function at the corresponding position. The S2S striping characteristics is obtained from the difference between the mean values of a detector at a given scan direction and the whole image. For the real time de-striping operation, the S2S striping for an image is corrected by an average of S2S striping terms from the images at the same time in previous two days. Before the correction is made, the data are collected for the current scan that will be used to evaluate the S2S striping characteristics, which is calculated at the end of each frame to replace the old value in the data buffer which will be used for the image at the next day at the same time.
The results show that the algorithm has been very effective to provide systematic corrections to both D2D and S2S stripings. Independent evaluation of the test data shows that the S2S and D2D striping noises meet the specification after the application of correction algorithm. It is also very efficient so that it has no significant impact on the latency of the Sounder data processing in GOES ground system.
Manuscript
A Real Time De-Striping Algorithm for Geostationary Operational Environmental Satellite (GOES) 15 Sounder Images
The images in a short wave channel 15 (4.45µm) of GOES 15 Sounder show considerable striping effects. The study by ITT Exelis group showed that these striping effects can be characterized by the detector to detector (D2D) and scan to scan (S2S) stripings. The D2D striping is caused by the overall sinusoidal oscillation behavior across east-west scan direction for each detector with the fixed wavelength of 350 pixels. The phases of the sinusoidal oscillation for detector 1 and 3 is opposite to that for detector 2 and 4 so that the striping effects reaches to the maximum when the oscillation magnitude for 1 and 3 (1+3) or 2 and 4 (2+4) are at maximum. The D2D is location dependent as well as time dependent with a diurnal behavior. The S2S striping is overall the difference of same detector between East to West and West to East scans. This striping feature is not location dependent in an image, however, it is time dependent with the diurnal behavior.
A real time de-striping algorithm implemented in GOES ground systems is presented in this study. The algorithm corrects both D2D and S2S stripings on a scan by scan basis to minimize the impact on the latency requirements of the GOES Sounder data processing in the ground system. The de-striping is done sequentially with D2D striping being corrected first. The algorithm retrieves the sinusoidal oscillation function from an offset function between detectors 1+3 and 2+4 through the Fast Fourier transformation (FFT), as the offset function is dominated by the D2D striping function. The D2D striping is corrected for a detector in the scan by subtracting the value of the oscillation function at the corresponding position. The S2S striping characteristics is obtained from the difference between the mean values of a detector at a given scan direction and the whole image. For the real time de-striping operation, the S2S striping for an image is corrected by an average of S2S striping terms from the images at the same time in previous two days. Before the correction is made, the data are collected for the current scan that will be used to evaluate the S2S striping characteristics, which is calculated at the end of each frame to replace the old value in the data buffer which will be used for the image at the next day at the same time.
The results show that the algorithm has been very effective to provide systematic corrections to both D2D and S2S stripings. Independent evaluation of the test data shows that the S2S and D2D striping noises meet the specification after the application of correction algorithm. It is also very efficient so that it has no significant impact on the latency of the Sounder data processing in GOES ground system.