Assessment of the accuracy of (re)analyses in the equatorial lower stratosphere

Abstract

AbstractThe accuracy of horizontal winds and temperature in the equatorial lower stratosphere is evaluated in different (re)analyses (European Centre for Medium‐Range Weather Forecasts (ECMWF) operational analysis, ERA Interim, and Modern‐Era Retrospective Analysis for Research and Applications) using an independent data set collected at low latitudes during long‐duration balloon flights in early 2010. The three analyzed wind products are found significantly less accurate than in the extratropics, with periods of disagreement with the observations lasting several days. To highlight the dynamical context in which the major disagreement events occur, case studies are carried out. The events are shown to be related to an improper representation of large‐scale equatorial Kelvin and Yanai wave packets with vertical wavelengths smaller than 5 km. Such events can induce large errors on trajectories computed with analyzed winds relatively to the actual (balloon) trajectory: 4000 km separation after 5 days of calculation. Reasons for analyses inaccuracy are discussed. The vertical resolution of the underlying model likely plays a role, but the main factor responsible for deficiencies appears to be the lack of wind observations. Indeed, errors in analyzed winds during the campaign have a strong longitudinal structure, with root‐mean‐square errors twice as large over the Indian Ocean and western Pacific, poorly covered by radiosounding stations, as over the Maritime Continent or South America. For the ECMWF analysis, this structure mirrors that of the analysis increments, which have largest amplitudes over observed regions. We argue that the reported events are more likely to happen during maximum shear phases of the quasi‐biennial oscillation.