Abstract
Radiosonde observations were conducted at Tsukuba in the warm seasons of 2011 and 2012 as a part of the research program Tokyo Metropolitan Area Convection Study for Extreme-Weather-Resilient Cities (TOMACS). The Aerological Observation Simulation (AOS) program, a system for simulating radiosonde track, was utilized in the observation for predicting the radiosonde landing area. The AOS program performance was validated based on the landing location data in operational radiosonde observations. Landing locations predicted by AOS were generally well correlated with those observed. The mean difference in landing location between simulation and observation was 16 km for 728 cases in the warm seasons of the past seven years. The AOS program adopted a Monte-Carlo method to account for uncertainties in (1) horizontal wind forecast, (2) radiosonde ascending speed, (3) pressure level reached, and (4) descending speed. This enables predicting the probability ellipse around the most likely landing location, which represents the possible landing area with a certain probability. Seventy-five percent of the observed landing locations were inside the 70% probability ellipse. Approximately 90% of the observed landing locations were inside the 90% probability ellipse. This means that the 90% probability ellipse is a reliable index of the landing area in the operational observations. However, 96% of the observed landing locations were inside the 99% probability ellipse, falling slightly short of the expected probability. The higher-than-expected rate of landing in the 70% probability ellipse was closely related to the performance of the horizontal wind forecast in the numerical weather prediction model. AOS prediction accuracy was lower on rainy days for multiple causes.
The AOS prediction was applied to the radiosonde observations in the 2011 and 2012 TOMACS Intensive Observation Period (IOP). In these observations, a 200 g balloon was chosen, and the heights reached were lower than those of operational observations. This reduced influences of summertime easterly wind in the lower stratosphere. Parameters used in the AOS probabilistic distribution (radiosonde ascending speed, pressure level reached, and descending speed for 200 g balloons) were revised using data obtained in the present observations. Parameter revision improved AOS predictions for 200 g balloons.
