NUMBER 121

DATE: 09/13/93

Sector Processing Center (SPC):

During July, NOA, JMA, AES, ESA and CSU processed greater than 99% of A data into B1/B2 data. Except for three NOAA-11 images and some METEOSAT-3 water vapor images, data collection for July was nearly 100%!Normal operations were reported by NOA, AES, ESA, JMA and CSU. ESA also reported normal operations in June.

Special Area Processing Center (SAPC):

UWS produced and shipped to the GPC replacement GOES-7 B2 data; 139 images from April 87 through October 88 filled the larger holes in the data collection.

Satellite Calibration Center (SCC):

The SCC received AC data for June/July 93 for NOAA-11, GMS-4, GOES-7 and METEOSAT-4. Problems with processing METEOSAT-3 AC data for the 75W position have been encountered; re-processing of these data is underway at ESA. BC data were sent to the GPC for April/May 93 for GOES-7 and for April 93 for METEOSAT-4.

Global Processing Center (GPC):

The GPC continued to receive B2 data, BC data and correlative data in a routine manner.Production of NOAA-11 (preliminary) B3 data is current.Production of NOAA-12 (preliminary) B3 data is current.Production of GOES-7 (preliminary) B3 data is current. Replacement B2 data for images missed by AES have been received from UWS covering the largest holes in period from April 87 through October 88 (139 images). Once transformed into B3 format, these data will be merged with the AES data to improve coverage.Production of METEOSAT-3 B3 data has not started yet. Final testing of software continues: the only remaining problem is to devise a way to substitute calibration information from "regular" images into "archive" images. During the early collection of data, CSU used two sources of images, the regular transmission and an alternative, local archive which did not include a key calibration coefficient.Production of METEOSAT-4/5 (preliminary) B3 data is current.Production of GMS-4 (preliminary) B3 data is current.Production of all INSAT-1 (preliminary) B3 data has been completed. These data will not be delivered to the ICA until the short-term calibration variations are removed (see below).Investigation of the relatively large short-term calibration variations in the INSAT-1 B2 data was expanded to include investigation of a few large deviations that occasionally occurred in the processing quality checks for C1 data. The quality check simply compares the retrieved surface and cloud properties from all overlapping observations, but focuses on comparisons between the geostationary and polar orbiting satellite results. These results showed generally small deviations (< 3% for VIS and < 1 K for IR), but occasionally larger deviations. By examining the data more carefully and testing the sensitivity and precision of our quality check procedure, we have confirmed that these larger deviations represent real calibration differences. During the re-processing, we will alter the calibrations for particular satellite-months (about 20 - 30 cases) to reduce the magnitude of these discrepancies. The same procedure will be used to determine the final INSAT-1 calibrations.Production of the ice/snow and TOVS correlative datasets is current.Testing of the revised cloud detection algorithm has been completed. We have examined the nature of the changed results, compared the characteristics of the newly detected clouds with those of the "old" clouds and the clear sky, and intercompared results from different wavelength channels on the AVHRR. The major changes in the standard algorithm are a refined analysis of the clear sky composite and lowering of some thresholds. The refined clear sky analysis eliminates some spurious effects and cloud contamination (only small amounts of a few percent but in some regions approaching 10%). The reduced thresholds slightly increase cloud amounts over high latitude land areas in winter and over some high latitude ocean areas. The newly detected clouds appear to resemble "old" clouds more than clear sky. Of the additional channels on AVHRR, only Channel 3 has been found to be useful, since almost everything detected by either Channel 2 or Channel 5 was already detected by the standard algorithm. Channel 3 detections will significantly raise summertime cloud amounts (by about 15 -20%) in the central Arctic and slightly increase cloud amounts in wintertime (by 5%), making the results more consistent with other information. The cloud amounts over the South Pole also increase (by 10 - 15%), but not by enough to make them more consistent with surface observations. However, on-going studies of surface observations may revise the cloud amounts inferred from them. In general, the algorithm revisions have significantly reduced the notable errors in first results.As decided by the Working Group on Radiative Fluxes, we will archive the pixel-level dataset for the revised version and future data. To that end we are designing a format that is easier to use than the "internal" system-specific format used for CX data the last time. This new dataset, called DX, will report up to 33 bytes per pixel, where nighttime pixels may as few as 13 bytes (eg, GOES satellites) and polar satellite pixels have 6 extra bytes. To make use of this very large dataset simple, yet keep data volume under control, each pixel is labelled by both its actual latitude-longitude and an array index for image positioning; however, only pixels containing data will be present. The number of bytes per pixel is determined by an image-level label (satellite identification) and by a day/night flag in the pixel. We have also added a single "no-cloud" or "cloud" bit to provide a simple guide to the decision. With the introduction of a new ice microphysical model, but a large uncertainty about the temperature at which liquid changes to ice, the DX dataset will contain the results of both analyses for all clouds, even though the new C1/C2, called D1/D2 will not.The overall format of the DX is structured so that individual "images" can be isolated as separate datasets for users who wish to obtain very specific samples. This is a natural division for the geostationary satellites; however, for the polar orbiter data, we will create four datasets: north pole, south pole, low latitudes (day) and low latitudes (night). This will produce a dataset for polar studies that is equivalent to having a specific satellite viewing the poles.Due to renovations and construction at the GPC, our mainframe computer system will be shut down from 17 September through 25 October. Consequently, we will be almost "out of business" for over one month. However, we have managed to move a number of activities to our workstation network, so that some progress on completing changes will be made. There will be no monthly report early in October, rather the next report will be issued in early November. E-mail may not be available, but all other forms of communication will be functioning, so that datasets and reports should be sent to the GPC as usual.

ISCCP Central Archive (ICA):

The ICA continued to receive B1 data from NOA, AES, ESA, CSU and JMA and B3 and C data from the GPC in a routine manner.