Abstract The northern extratropical stratosphere shows very large variability during an extended winter period and little variability in summer. Our diagnostic estimates of the forecast impact of radiosonde data suggest that stratospheric profiles in this area are much more useful in winter than in summer – despite the average burst height being lower in winter. Observing‐system experiments show similar features but less seasonality in the impact of radiosonde data. The reasons for this are explored. Both techniques suggest that in the stratosphere the radiosonde winds have a larger direct impact than the radiosonde temperatures – partly because satellites provide much more temperature than wind information. In the southern extratropical stratosphere the maximum variability is in the austral spring and there is a less clear‐cut variation in radiosonde impact, but a clear maximum in the impact of high‐peaking satellite channels. We support the World Meteorological Organisation (WMO) recommendation of two profiles a day from radiosondes with some sampling to at least 10 hPa, but suggest that the sampling up to 10 hPa should be increased (using larger balloons where necessary) in the boreal winter/spring to benefit both forecasting and climate studies.
Ingleby et al. (Wed,) studied this question.