OPTIMIZING A LOCAL WEATHER FORECAST MODEL

THERE IS A NEED FOR IMPROVING THE FORECASTING OF LOCAL WEATHER ELEMENTS SUCH AS PRECIPITATION, SURFACE TEMPERATURE,ETC. CURRENT FORECASTS ARE MADE BY A STATISTICAL ADAPTATIONOF LARGE SCALE NUMERICAL FORECASTS. WE PROPOSE TO USE, INSTEAD, A MODEL THAT COMPUTES THE PHYSICAL PROCESSES THAT LOCALLY MODULATE THE LARGE SCALE METEOROLOGICAL FIELDS. THEPARAMETERS OF SUCH A MODEL NEED TO BE ADJUSTED TO THE LOCAL CONDITIONS. WE PROPOSE TO USE THE ADJOINT MODEL TECHNIQUE, A TOOL OF OPTIMAL CONTROL THEORY, TO DO THIS ADJUSTMENT, I.E. TO FIND THE SET OF PARAMETERS THAT MINIMIZE THE LOCAL FORECAST ERRORS. PHASE I WILL DEMONSTRATE THE EFFECTIVENESSOF THIS METHOD. IN PHASE II, A COMPLETE PROTOTYPE FORECASTING SYSTEM WILL BE BUILT. THIS RESEARCH WILL ALLOW THE DEVELOPMENT OF MODELS THAT CAN BE TAILORED TO THE SPECIFIC NEEDS OF USERS OF LOCAL FORECASTS, SUCH AS UTILITY INDUSTRIES, AIRPORTS, AGRICULTURE, ETC. OTHER POTENTIAL APPLICATIONS OF THE RESEARCH ARE THE CALIBRATION OF SATELLITE OBSERVATIONS OF GROUND CHARACTERISTICS, AND THE INDIRECT MEASUREMENT OF QUANTITIES, SUCH AS SOIL MOISTURE, THAT ARE DIFFICULT TO MEASURE ROUTINELY IN SITU.