Temperature Measurements in the 30- to 40-kilometer Region

When comparing rocketsonde and balloonsonde (radiosonde) temperature data, one notes systematic differences at levels of 25 km and above that increase with height. During the past 10 yr, systematic errors in the rocketsonde have been largely eliminated (at least for the region from 20 to 45 km), but little attention has been paid to the balloonsonde sensor. A study of temperature data for June 1967 suggests about half the difference is due to infrared cooling of the balloonsonde thermistor and much of the remaining difference may be due to the thermistor riding up through the wake of a balloon cooled by radiative and adiabatic processes. These errors make balloonsonde temperature data above 30 km unsuitable for many types of study. During the past dozen years, the advent of the rocketsonde and the high-altitude radiosonde has opened up new areas of exploration at levels of 30 km and above. Meteorologists of many specialties have been examining these new data to determine dominant patterns of wind and temperature, typical behavior, tidal motions, estimates of momentum and energy budgets, and factors affecting ozone photochemistry. As might be expected, the quality and quantity of observation have improved considerably over the years; however, at present we do find significant systematic differences between temperatures as measured by radiosondes (hereafter we shall refer to these as “balloonsondes”) and as measured by rocketsondes, at levels from 30 to 40 km. An inspection of June 1967 data, for example, revealed differences of about 1°K at 25 km, 4” to 10°K at 35 km, and 10” to 18°K at 40 km, with the balloonsonde temperature measurements being lower. The measurement of temperatures aloft using a rocket is a rather difficult task; and not surprisingly, many early measurements suffered errors from many technical difficulties. I n this system, a rocket is fired; it rises to 50 km or above; a package is ejected; a parachute opens allowing the package to descend at speeds of 50-100 m s-l a t 50 km, slowing to 15-20 m s-l by 30 km, while the package telemeters data to a ground station and is tracked by precision radar. Synoptic meteorologists questioned much of the early temperature data; and the engineers responded with bctter sensors, bettcr mounting, and standard data reduction techniques. Thus, much of the day-to-day fluctuation noted in midsummer observations of earlier years is no longer observed; and while there is concern over apparently excessive day-night differences measured near 50 km (Lindzen 1967), there are no longer any technical reasons to expect large systematic errors in the 30to 40-km region. On the other hand, synoptic meteorologists working with balloonsonde data from 16 to 30 km have long been familiar with instrumental difficulties that lead to systematic difference between measuring systems of different countries &s well as fictitious day-night differences with each system. In general, the more advanced systems measured lower temperatures and smaller day-night difference than the older systems. At these altitudes, the air is very thin” and consequently is relatively inefficient in removing heat from a sensor that has been affected by spurious energy from electronic equipment or the sun. Thus, meteorologists have grown up with the situation that, if the sondes are in error, they are probably reading too high. For illustrating the systematic temperature differences, figure 1 shows the 5-mb monthly-mean temperature for June 1967 as a function of latitude. The strong, relatively undisturbed easterly winds at these levels imply very little longitudinal temperature gradient , particularly when averaged over a month. However, when examining the balloonsonde data, we see a wild scattering of points that must be smoothed to reveal the true meridional gradients. In marked contrast, the rocketsonde data show a smooth systematic latitudinal variation that inspires confidence. Looking further at the balloonsonde data, one notes that the temperatures from hypsometerequipped sondes (solid circles) tend to be higher than those based on baroswitches (crosses), suggesting a systematic error in the baroswitch such that the sondes were actually near the 6-mb level when the baroswitch indicated 5 mb, thus giving a temperature 2” to 3°K lower than a hypsometer-sonde would have indicated. Within the hypsometer-sonde data, there was a marked dependence on solar elevation angle such that a given L L