Evaluation of Balloon Trajectory Forecast Routines For GAINS

Abstract

The Global Air-ocean IN-situ System (GAINS) is a global observing system designed to augment current environmental observing and monitoring networks. GAINS is a network of long-duration, stratospheric platforms that carry onboard sensors and hundreds of dropsondes to acquire meteorological, air chemistry, and climate data over oceans and in remote land regions of the globe. Although GAINS platforms will include balloons and Remotely Operated Aircraft (ROA), the scope of this paper is limited to balloon-based platforms. A primary goal of GAINS balloon test flights is post-flight recovery of the balloon shell and payload, which requires information on the expected flight path and landing site prior to launch. Software has been developed for the prediction of the balloon trajectory and landing site, with separate versions written to generate predictions based upon rawinsonde data and model output. Balloon positions are calculated in 1-min increments based on wind data from the closest rawinsonde site or model grid point, given a known launch point, ascent and descent rate and flight duration. For short flights (< 6h), rawinsonde winds interpolated to 10-mb levels are used for trajectory calculations. Predictions for flight durations of 6 to 48h are based upon the initialization and 3 h forecast wind fields from NOAA's global aviation- (AVN) and Rapid Update Cycle (RUC) models. Given a limited number of actual balloon launches, trajectories computed from a chronological series of hourly RUC initializations are used as the baseline for comparison purposes. These baseline trajectories are compared to trajectory predictions from the rawinsonde and model-based versions on a monthly and seasonal basis over a 1-year period (January 1 - December 31, 2001) for flight durations of 3h, 6h and 48h. Predicted trajectories diverge from the baseline path, with the divergence increasing with increasing time. We examine the zonal, meridional and net magnitudes of these deviations, and attempt to determine directional biases in the predictions. This paper gives an overview of the software, including methods employed, physical considerations and limitations, and discusses results of this evaluation.