The impact of aircraft digital weather data in an adaptive observation strategy to improve the ensemble prediction of hurricanes

Modern meteorological data assimilation techniques and ensemble weather prediction methods involve the minimization of error variance. Both of these techniques are used here to assess the impact of assimilating digital aircraft weather data, in an adaptive observation strategy, to improve hurricane forecasting. Many operational and research centers utilize ensemble prediction systems to improve their weather forecasts. This study utilizes a Monte Carlo technique to randomly generate a series of 50 separate perturbed model states. These "random" perturbations are scaled to the typical error values observed daily in meteorology. From the family of 50 ensemble members a field of sea level pressure variance, at the 48-hour forecast point, is calculated using deviations of the 50 members from the control (unperturbed state). The bull's eye of maximum error variance is then backward correlated to various meteorological fields such as temperature, wind and humidity at the 24-hour point of the forecast. The areas of highest correlation represent model sensitivity to random error growth and thus targets for intensive observations.;In August and September of 1998, a team of scientists participated in a NASA-sponsored field campaign termed the Third Convection and Moisture Experiment (CAMEX-3). The purpose was to conduct an intensive study of Atlantic hurricanes. The CAMEX-3 data were utilized in this research to assess the impact of targeted observations on subsequent forecasts. Specifically, a series of experiments were conducted utilizing: all of the CAMEX-3 data, a portion just over the model-sensitive target areas, and from a random selection of data throughout the near-storm environment. These data were assimilated using both a multivariate optimal interpolation technique and a 4-dimensional variational assimilation method. Results from these experiments showed that these data, once assimilated into the model's initial state, improved subsequent hurricane forecasts. Statistical measures of track and intensity forecasts, for a limited set of storms, validated the adaptive observation strategy as a means of targeting specific areas for intensive observation. Results were mixed regarding which assimilation method best incorporated the digital data. Operational constraints limited the number of case studies for the adaptive observation strategy, thus overall the number of realizations inhibited the statistics.