Using Virtual Dropsondes to Estimate the Sensitivity of Dropsonde Analyses to Sampling Errors

Dropsondes are an important observational tool that have been used in numerous field campaigns to aid in our understanding of tropical cyclone structure and dynamics, including the recent TCI and HS3 experiments. Even with intensive sampling strategies, the coverage of dropsondes is still relatively sparse in space and infrequent in time. Therefore, it is unclear how representative or robust dropsonde-analyzed structures are. For example, from the TCI dropsondes, we can calculate radial profiles of azimuthal mean radial winds in the outflow layer. However, sampling errors and/or biases may result in errors in the analyzed outflow, both in magnitude and structure. From the observed dropsondes alone, it is difficult to estimate such errors.

In this study, we emulate dropsondes within COAMPS-TC simulations of real cases from TCI and HS3, by calculating modified parcel trajectories that have a fall speed relative to the air. We then sample these virtual dropsondes in a manner analogous to the actual sampling done in TCI and HS3. We compare analyses derived from the virtual sondes to the full simulated fields, and so determine the errors that result from limited and/or biased sampling. To the extent that the COAMPS_TC simulations are reasonably realistic, we can estimate the errors in analyses derived from the observational datasets. Finally, we also assess which potential and feasible sampling strategies are most optimal, and how many dropsondes are necessary to resolve the structures of interest. Hopefully, this will help guide future observational campaigns.