Air Force Global Weather Center's (AFGWC) Relocatable Window Model (RWM) total cloud forecasts were validated using data for selected days in May, June, and July, 1996. Forecasts were generated twice daily (00 UTC and 12 UTC) to determine the RWM's ability to accurately forecast total cloud cover during the late spring and early summer. The RWM forecasts were post-processed using the Slingo cloud forecast algorithm and compared against AFGWC's operational real time nephanalysis (RTNEPH) cloud analysis model. As a minimal-skill baseline comparison to the RWM's total cloud forecast, the RTNEPH initial analysis hour was persisted and evaluated against the same RTNEPH analysis as the RWM forecasts. The results indicate RWM total cloud forecasts did not show improved skill, sharpness, accuracy or bias when compared against RTNEPH persistence through the 36-hour forecast period. The results also suggest the Slingo algorithm, as tested, is not appropriate for use in the RWM as an accurate total cloud forecast method for the late spring and early summer months. The RWM's total cloud forecast performance during the late spring and early summer over the North American Window should be improved in the short term by incorporating convective parameterization within the Slingo algorithm or replacing the Slingo algorithm with an alternative algorithm designed for more accurate and skillful total cloud forecasts. While the suggested short-term improvements are incorporated into the RWM, the results of this and other related studies must be carefully communicated to the operational users of the RWM products to be useful. In the long term, the RWM should be replaced with a state of the art forecast model capable of forecasting clouds deterministically, rather than diagnostically.

AFGWC has three cloud forecast models: Five-Layer (5LAYER), High Resolution Cloud Prognosis (HRCP), and Tropical Cloud Forecasting Model (TRONEW). These models satisfy a wide range of requirements and have been in operation since the early 1970s. The 5LAYER model makes extra-tropical forecasts for periods up to 48 hours. Forecasts of layer and total cloud, cloud type, layer temperatures, and icing and weather conditions are made. The 5LAYER model uses a quasi-Lagrangian approach to make these forecasts. The trajectories needed for these forecasts are computed from the AWS Global Spectral Model (GSM) derived wind forecasts. The 5LAYER moisture is initialized from the Real-Time Nephanalysis Model (RTNEPH) layer cloud-amount and the Multi-layer Analysis Model (MULTAN) layer dew-point depression. The 5LAYER temperatures are initialized from the High Resolution Analysis System (HIRAS) and GSM derived temperatures. The High Resolution Cloud Prognosis (HRCP) model combines RTNEPH analyzed cloud input with 5LAYER trajectories to produce high resolution (25 nm), short-range (out to 9 hour) cloud forecasts. The TRONEW model uses the analyzed RTNEPH cloud to make 24 hour persistence cloud forecasts for the tropics.

This technical note (TN) describes the automated analysis/forecast model system that currently exists at the Air Force Global Weather Central (AFGWC). The emphasis will be on the interrelation and ing of the various analysis/forecast models in the production cycle. This description of the automated analysis/forecast system was written for managers, programmers, computer operations personnel, users of the ultimate products, and the meteorological community at large. This TN will only address analysis/forecast models that are the primary meteorological data-base builders. Most applications programs that access the data bases are therefore not included.

This study was performed under a plan which allows complete traceability between user requirements, Air Force Global Weather Central operational functions, requirements levied upon the data system, a proposed component configuration, which meets the data system requirements, and a system specification designed to acquire a system which meets these requirements. The resultant system described has a number of unique features, including total hardware authentication separation of security levels, load leveling accomplished by assigning main processors in accordance with a dynamic priority queue of tasks, and a system-wide network control capability. Other key features incude a central data base processor to fill requests for data from other processors, computer operations centers, the use of array processors for accomplishing difficult numerical problems, and sophisticated forecaster console support. These elements have been designed to provde 99.5% reliability in meeting user requirements.

The RTNEPH (Real Time NEPHanalysis) model replaced the 3DNEPH (3- Dimensional NEPHanalysis) model in August, 1983 as the AFGWC (Air Force Global Weather Central) automated cloud analysis model. It uses a horizontal grid spacing of 25 nautical miles and uses four floating layers in the vertical. The RTNEPH uses visual and infrared data from meteorological satellites as well as surface observations, radiosonde data, and aircraft reported data to produce its analysis. A capability exists to manually modify the analysis to correct deficiencies. This nephanalysis is then used as input for AFGWC's cloud forecast models and is stored for climatological purposes by the USAF Environmental Technical Applications Center. The RTNEPH is scheduled to undergo a technical enhancement to incorporate multi-spectral techniques, use finer resolution satellite data, and produce a 12.5 nautical mile resolution analysis. Keywords: Cloud analysis, Meteorology, Nephanalysis, Polar orbiting satellite data.