In the fall of 2001, the Air Force Weather Agency (AFWA) is scheduled to replace its existing operational global cloud analysis capability with an upgraded system known as the Cloud Depiction and Forecast System II (CDFS II). In addition to new processing hardware and database management facilities, this upgrade will represent the first major improvement in retrieval methodology in the 30 plus year history of operational cloud analysis at AFWA. Existing capabilities rely primarily on two-channel data from the Operational Linescan sensor onboard the Defense Meteorological Satellite Program Block 5-D3 platforms. In CDFS II, DMSP data will be supplemented by civilian polar-orbiting satellites and the addition of half-hourly observations from the international constellation of geostationary systems. Coverage from the combination of both polar and geostationary satellites represents a major enhancement to the global observing capability, particularly in the tropics which are a recognized weakness of the current system. In addition to the new data sources, CDFS II features a completely new set of cloud analysis algorithms developed specifically for this program and designed to exploit the unique information content from each sensor system. The nephanalysis algorithms have been described elsewhere and will not be repeated here. The focus of this paper will be on the testing and validation of the algorithms as they have been adapted for global application in the CDFS II environment.

In preparation for system acceptance, CDFS II is undergoing an extended period of formal testing. A key consideration is validation of the new cloud analysis algorithms. Algorithm testing is three phased. Initially source code reviews are conducted to verify correct implementation of the algorithms. The second phase is preliminary testing to evaluate infrastructure consistency and low level functionality. Of primary importance is the coordinated use of satellite, satellite-derived and supporting (e.g., NWP) data. Final testing focuses on integrated algorithm performance for global application under real-world conditions. Final testing is facilitated by the binary design of CDFS II that consists of a Production system mirrored by a separate Development system with redundant database and processing capabilities. The design allows the development system unrestricted access to the dynamic databases that are continuously maintained and updated on the production system, without interfering with the ongoing production cycles. This allows testers to perform detailed analysis of the operational products while insuring that the data samples are representative and that no special processing is performed on the test data sets. This paper will provide results of the algorithm testing and descriptive information on the available CDFS II data products, including access to real-time and archived products.