On the value of reanalyses prior to 1979 for dynamical studies of stratosphere-troposphere coupling

Studies of stratosphere-troposphere coupling, particularly those seeking to understand the dynamical processes underlying the coupling following extreme events such as major stratospheric warmings, suffer significantly from the relatively small number of such events in the "satellite" era (1979 to present). This limited sampling of a highly variable dynamical system means that composite averages tend to have large uncertainties. Including years during which radiosonde observations of the stratosphere were of sufficiently high quality substantially extends this record, reducing this sampling uncertainty by up to 20 %. Moreover, many open questions in this field involve aspects of tropospheric dynamics likely to be better constrained by "conventional" (i.e. radiosonde and surface-based) observations. Based on an intercomparison of reanalyses, a quantitative case is made that for many purposes the improved sampling obtained by including this period outweighs the reduced precision of the reanalyses in the Northern Hemisphere. Studies of stratosphere-troposphere coupling should therefore consider the use of this period when using reanalysis data. These results also support continued attention on this period from centres producing reanalyses.