Strateole 2 superpressure balloons reveal persistent errors in reanalyzed winds in the tropical lower stratosphere

Winds in the tropical lower stratosphere raise difficulties for numerical weather prediction models: without geostrophy, winds decouple from temperature and direct observations are scarce. The Strateole 2 project explores the tropical lower stratosphere using superpressure balloons that drift for up to three months between 18 and 21 km altitude. Wind measurements from the technological campaign (2019–2020) and the first scientific campaign (2021–2022) are used to assess errors in the ERA5 reanalysis for latitudes between 18° S and 10° N. The comparison reveals significant errors, with standard deviations of 3.76 m s−1 for zonal and 3.24 m s−1 for meridional wind. Relative to a previous comparison in 2010, only a modest decrease of 20 % and 10 % is found, revealing the persistent difficulty of modeling winds in the tropical lower stratosphere. 

Additionally, the errors in modelled balloon trajectories are also assessed, with a focus on the predictability of the trajectories. It is shown that the initial error in the wind gives a reliable indication on the skill of the subsequent forecast. Trajectory calculations have very variable skill, with median errors after 24 h of 260 km, but a tenth of the errors larger than 600 km. Factors leading to large errors, such as initial wind error and latitude are identified. 

Certain instruments onboard Strateole 2 balloons measure features below the balloons (temperature, thin cirrus, water vapour..). While the sampling of air at balloon flight level is quasi-Lagrangian, observations of features below the balloon describe both spatial and temporal variations. In order to disentangle these and facilitate the interpretation of observations made below the balloons, we document the dispersion of air below the balloons (altitudes between about 15 and 21 km). Trajectory dispersion of air below the balloon is very variable, depending on the initial shear. The persistent errors highlight the need for regular obsevations of winds in the tropical lower stratosphere. 

Overall, we emphasize the need for caution when using trajectory calculations for process studies.