SBIR-STTR Award

The objectives will be: (1) an assessment of existing analytical tools for predicting cloud-free fields-of-views (CFFOV) and (2) a specification of a new model for improving CFFOV forecasts. A review of existing cloud measurement and modeling efforts will be made. An improved model (`CFFOV predictor`) based on generalized, nonlinear temporal prediction by neural networks trained on CFFOV data will be specified with a workstation, graphical user interface (GUI) and off-the-shelf hardware. Predictor inputs will be ground and satellite cloud data and vertical wind profiles from sites at the U.S. Army kwajalein Atoll (USAKA) and neighboring atolls. Outputs will be best CFFOV's for a range of times in the near future. The predictor will estimate cloud altitudes, fractional cloud cover in the FOV's and time windows for clear viewing. Thus IR viewing plans can be made to see through cloud decks in the near future. Government benefits will be a review of sky coverage data and models, a feasibility assessment of an improved cffov predictor to aid IR collection, ground-based laser and satellite laser communications. Commercial benefits include predicting visual flight rule (VFR) conditions and solar insolation for alternative energy sites. Short-term prediction of visibility will benefit air traffic control tasks. Ground-based laser defense systems (e.g., SDIO program) and satellite found communications (e.g., submarine blue-green laser data links) will operate better with reliable predictions of CFFOWV's. Long-term prediction of insolation will allow evaluation of available sunlight for agricultural needs and exposure to UV rays.

KTAADN will develop a workstation, called the Cloud-Free Line-of-Sight Optimal Locator for Sensor Aircraft or 'CLOSA,' manages RV launch and trajectory parameters, collection aircraft flight constraints, geographical descriptors and current weather inputs. These variables are manipulated by the flight planner on the CLOSA overlay a chart of the probability of a cloud-free line-of-sight for all geographical locations against flight experiment ground track. Individual cloud forecasts for each allocation are provided by neural network elements selected for contextual matches (geography, season, air mass) and trained on extensive sequences of weather satellite cloud image data from within that context. CLOSA is capable of enhancing the neural network performance by retraining against new image data and validation using contingency table evaluation methods. Benefits include: optical collection against distant targets, and agricultural operations requiring estimates of radiation cooling to make a decision on allocating resources to protecting crops from freezing and excessive drying conditions.