Interactive tools to support a prototype high-resolution cloud analysis and forecast system

Recently, a prototype, high-resolution, automated cloud analysis and forecast system was developed for use at the Air Force Weather Agency (AFWA). The prototype produces automated cloud analyses and forecasts at approximately 6 km resolution using electro-optical imagery data from two polar-orbiting satellite systems, the Defense Meteorological Satellite Program (DMSP) and the NOAA Polar-orbiting Operational Environmental Satellite (POES) program. This system marks the first quantitative use of fine-mode (0.5 km resolution) data from the Operational Linescan System (OLS) imager on board the Defense Meteorological Satellite Program (DMSP) satellites. It also marks the first worldwide implementation of the cloud analysis algorithms developed under the Support of Environmental Requirements for Cloud Analysis and Archive (SERCAA) program that will be used on AFWA's future system, the Cloud Depiction and Forecast System II (CDFS II). The prototype products are 16 times higher resolution than the planned CDFS II products and 64 times higher resolution than the current operational cloud products at AFWA. The technological "firsts" combined with the increase in data volume presented significant challenges in the area of data analysis, algorithm validation, and product generation. To address these challenges, a significant portion of the effort involved development of interactive display tools. Additionally, visualization needs had to be considered at virtually every level of the design and development of the system.

Several steps in the system required visualization and analysis tools. First, raw imagery must be properly earth-located and calibrated. Visualization tools were needed to assess the accuracy of this step. Second, additional meteorological fields necessary for the cloud analysis algorithms such as time-coincident surface temperature predictions and snow cover have to be retrieved from gridded databases in real time. Therefore, tools to verify and validate the results of this step were needed. Third, cloud detection algorithms are applied at the pixel level (i.e., not projected to a map) and the results of the cloud tests needed to be compared to the original imagery. Fourth, the pixels for which clouds are detected are aggregated to grid points on a polar-stereographic grid where cloud bases, heights, type, and amount are determined. Finally, a method of generating automated products useful to the meteorologist and non-meteorologist was needed. It is clear there is a need for a variety of tools that operate on a variety of data formats that are all inter-related, and each of the tools had to have both quantitative and qualitative methods of examining the data.

This presentation gives examples of how we addressed these challenges using a 4th-generation data visualization software package as well as the unique challenges of working with satellite data and derived products. It will also demonstrate the value of interactive data visualization languages for rapid-prototyping purposes in today's world of high-volume datasets.