Geosat GM 9-Track SDR recovery summary John Lillibridge - NOAA/LSA 12/22/2004 ========== -------- Summary: -------- Previous analyses of the SDR files from Geosat's Geodetic Mission indicated that only 20 of the 549 daily SDRs had an inconsistency between the number of records reported in the header record and the actual number of data records within the SDR. Of these 20 tapes, 17 of the original 9-tracks were sent to Bill R. King's Data Recovery International firm in Texas for advanced forensics. The remaining 3 tapes with header/data record count mismatches could not be located at NODC. The following section details the status of each of these 20 SDRs following nearly complete success by Mr. King in re-reading the tapes. Additionally one SDR (85_101) was found which contained 17 duplicate records, and it is therefore included as an SDR with a header/data mismatch below. This brings the total count to 21 suspicious SDRs. After data recovery 11 SDRs are still 'incomplete' (assuming the header information is correct) while the other 10 SDRs are fully restituted. The total number of missing records from these 11 SDRs is 1358 seconds. If we ignore SDR 85_254, which apparently has always had a header/data mismatch, the total number of missing records due to tape read errors drops to only 217, ie 3' 37" out of 549 days! ---------------------- Detailed Descriptions: ---------------------- 85_101: This tape was not re-read. The original SDR had 17 duplicate records, which were removed by sdr_dupefix. The missing data are not the result of tape read errors. The original file had a 6 'bit-slip' errors (bit #3 set wrongly) requiring hand edits. The file also had a bad frame count that was replaced with the correct major+minor frame counts: 947284+25. 85_120: Completely recovered after rereading. Original had 209 data records, new file has full 88170. 85_182: Completely recovered. Original was seriously corrupted and had many duplicates, such that the number of data records was actually greater than the header value. Recovered file has no dupes, no corruption, and the full 87945 records. The original file also had 'bit-slip' errors that required two lines to be removed. The new file has no such errors. 85_183: Completely recovered. Original had 74613 records, new file has full 88171. 85_200: Completely recovered. Original had 55252 records, new file has full 88170. One bad frame count in new file was hand edited with correct major+minor frame counts: 3680492+30. Original also had duplicate errors, but not new file. 85_240: Recovered all but 19 records (1 tape block). Original had 68684 records, new file has 88151 of possible 88170 records. 85_242: Completely recovered. Original had 27460 records, new file has full 88170. Original file was corrupted and had duplicates. New file has none of these anomalies. 85_243: Completely recovered by merging portions of original and new files. New file was missing 38 records (2 tape blocks) which didn't overlap with the missing 29 records at the very end of the original file. Merged file has full 88170 records while original file had 88141 and recovered file had 88132. All extant records in original and recovered files matched exactly. 85_254: Original and recovered files were identical, both missing 1158 records, with 87012 data records compared to 88170 according to the header. Original SDR->GDR processing logs confirm this situation back in 1985 so this loss is NOT the result of tape read errors, but is presumably either a wrong value in the header or the tape was improperly created in the beginning at JHU/APL. 85_264: Recovered all but 19 records. Original had 54131 records, new file has 88151 of possible 88170 records. One bad frame count in new file was hand edited with correct major+minor frame counts: 5448503+19. 85_267: Completely recovered. Original had 88151 records, new file has full 88170. One bad frame count in new file was hand edited with correct major+minor frame counts: 5530728+29. 85_269: Completely recovered. Original had 88141 records, new file has full 88170. Two bad frame counts in new file were hand edited with correct major+minor frame counts: 5587081+23 & 5600503+29. 85_300: Unable to find original tape. Original has 88141 of 88170 records, missing 29. 85_305: Recovered all but 19 records. Original had 88141 records, new file has 88152 of possible 88171 records. 85_319: Completely recovered. Original had 74250 records, new file has full 88170. Original file had corruption errors, new file has none. 85_339: Unable to find original tape. Original has 88141 of 88170 records, missing 29. 85_345: Unable to find original tape. Original has 88141 of 88169 records, missing 28. 85_347: Recovered all but 19 records. Original had 88141 records, new file has 88151 of possible 88170 records. 86_006: Recovered all but 19 records. Original had 56525 records, new file has 88151 of possible 88170 records. 86_051: Recovered all but 19 records. Original had 88141 records, new file has 88151 of possible 88170 records. 86_232: This tape was not re-read. The original SDR had one 'bit-slip' error (bit #1 set wrongly) requiring hand edits. No missing records. 86_266: Completely recovered. Original had 64393 records, new file has full 88170 records.