NUMBER 44

DATE: 04/06/87

Sector Processing Center (SPC):

During February, ESA, JMA, and CSU collectively processed greater than 97% of A data into B1 and/or B2 data.

Normal operations were reported by ESA, CSU, JMA, and NOA. The February monthly report was received from NOA, but no February B2 processing matrix was received.

NOAA shipped NOAA-9 B2 data for November 1985, eliminating the backlog of NOAA-9 data from 1985. No NOAA-9 B2 data has been received for December 1986, January 1987, or February 1987. NOAA also shipped NOAA-8 B2 data for January 1984.

AES reported that they have been operationally processing GOES-6 data since last year; however, they are dissatisfied with the reliability of their connection to the actual receiving station. Manpower to work on the improvement of the connection becomes available in April. They are also modifying their software to accommodate the new GOES AAA format, which was scheduled to be implemented at the end of March. AES is ready to begin processing GOES-7 data as soon as it becomes operational. They have sent a new sample B2 tape to the GPC.

Special Area Processing Center (SAPC):

No reports were received this month.

Satellite Calibration Center (SCC):

The SCC reports the completion of the BC data for GOES-6 for July and October 1985. GOES-7 AC data were requested from AES for April 1987. Contradictory information about the planned shift of GOES-6 back to its original WEST location was received from CSU and NOAA. The SCC received new GOES-6 AC data from UWS for April and July 1984; data for January 1984 are delayed at UWS by navigation problems.

Global Processing Center (GPC):

The GPC continued to receive B1 and/or B2 data from JMA, ESA, CSU, and NOA and correlative data from NOAA/NESDIS in a routine manner.

The new production version of the GMS B3 code, with the corrected navigation calculation, has been completed. Trial production of B3 data for July 1983 is underway; after checking these data they will be used to produce the first C1 data. Some further error testing remains, but full production should begin in late April.

Merging of GOES-6 data from CSU and UWS has progressed through December 1983. Six more months remain to be processed.

A new B2 sample tape was received from AES. Documentation is complete and accurate. We have been able to read the sample data, but are just beginning to attempt navigation to insure that the navigation information is complete and accurate. The format of the tape is acceptable. A complete report will be sent to AES once we have checked the navigation information.

The final step in the B3 quality inspection is manual inspection of all images labeled as suspicious by the automatic steps. This inspection has been completed for the 19 months of METEOSAT-2 data in the archive. Inspection of the NOAA-7 data has progressed through September 1983. These three months suggest that only about 1% of the NOAA-7 images are being flagged as suspicious. The problem with these images usually appears as dropped scan lines or scan lines containing bad data; in about one image a month, several tens of scan lines appear to have noisy data or spurious values. A list of all images finally judged to be too poor in quality to retain will be published.

We have completed processing of the overlapping three weeks of NOAA-7 and NOAA-9 data, using the AVHRR calibration monitoring procedure. Once a normalization is available for NOAA-9, the overlapping B3 data will be delivered to the ICA. Comparison of surface reflectances over the whole Earth measured by channel 1 on NOAA-7 and NOAA-9 gives a ratio of NOAA-9/NOAA-7 = 1.14. This would appear to indicate a calibration diffence between the two radiometers of this magnitude; however, this result does not account for the change in average solar zenith angle between NOAA-7, which had a local overflight time of about 16:30, and NOAA-9, which had a local overflight time of about 14:30. As part of the calibration monitor procedure, we collect results for about 15 targets as a function of solar zenith angle. Based on these results for the Sahara, for example, the change in reflectance with solar zenith angle should have made NOAA-9 appear to be about 2-3% (absolute) darker than NOAA-7. This suggests that the actual ratio of NOAA-9/NOAA-7 is about 1.18 - 1.2. The SCC has also estimated the ratio of NOAA-9 to NOAA-7 to be about 1.17, using METEOSAT as the comparison standard. We plan further studies of the solar zenith angle effect to insure a proper normalization of NOAA-9 to NOAA-7, including a refinement of the NOAA-7 calibration evolution over the period from July 1983 to January 1985 and a verification of the SCC results using simultaneous B3 measurements from METEOSAT, NOAA-7 and NOAA-9. A similar procedure will be used to normalize NOAA-8 to NOAA-7.

The inter-system I/O error in the radiation analysis part of the cloud algorithm has been avoided by placing all components of the cloud algorithm on a single operating system. The recoding of the algorithm and the production management software has been completed. All of the data for July 83 from NOAA-7, METEOSAT, GOES-5 and GOES-6 have been analyzed by both the detection and radiative model parts of the algorithm. Also the C1 production code has been completed; this software reduces the pixel-by-pixel results to the 2.5 degree grid, calculates the statistics discussed at the last WGDM meeting, and merges the results from all the satellites to produce C1 data. Some additional tests of the preliminary C1 product are planned for this week. The C1 tape format design has been completed and the final software that produces the tape header files and writes the C1 data onto tape is being written. In order to fit the C1 data onto two tapes for one month, we have had to store the data physically on the tape in an equal-area map grid; however, the tape READ software (that is included on the C1 tape) automatically produces the output data in a square latitude/longitude map grid that is identical to the ERBE map grid.

Testing of the accuracy of the description of cloud property distributions using different numbers of cloud types shows that the error incurred with 5 ranges in VIS and IR radiances is about 1-2% in VIS and about 3-4K in IR (we are using the radiances in these first tests to "classify" clouds for convenience). Most of the error in the IR is associated with cloud layers that happen to have temperatures in between the pre-defined classes. This suggests that the error can be reduced by using more ranges in the IR than in the VIS. These tests will be repeated with TAU (cloud optical depth) and PC (cloud top pressure) as the determinants of cloud classes; we plan to test a scheme with seven levels in PC and five levels in TAU (which will fit on the C1 tape). The first C1 data will be delivered using the original 5X5 scheme.

ISCCP Central Archive (ICA):

The ICA continued to receive B1 data from JMA, CSU, ESA, and NOA in a routine manner.