Comparison of lower-tropospheric temperature climatologies and trends at low and high elevation radiosonde sites

Abstract

Observations of rapid retreat of tropical mountain glaciers over the past two decades seem superficially at odds with observations of little or no warming of the tropical lower troposphere during this period. To better understand the nature of temperature and atmospheric freezing level variability in mountain regions, on seasonal to multidecadal time scales, this paper examines long-term surface and upper-air temperature observations from a global network of 26 pairs of radiosonde stations. Temperature data from high and low elevation stations are compared at four levels: the surface, the elevation of the mountain station surface, 1 km above the mountain station, and 2 km above the mountain station. Climatological temperature differences between mountain and low elevation sites show diurnal and seasonal structure, as well as latitudinal and elevational differences. Atmospheric freezing-level heights tend to decrease with increasing latitude, although maximum heights are found well north of the equator, over the Tibetan Plateau. Correlations of interannual anomalies of temperature between paired high and low elevation sites are relatively high at 1 or 2 km above the mountain station. But at the elevation of the station, or at the two surface elevations, correlations are lower, indicating decoupling of the boundary layer air from the free troposphere. Trends in temperature and freezing-level height are generally upward, both during 1979-2000 and during longer periods extending back to the late 1950s. However, some negative trends were found at extratropical locations. In many cases, statistically significant differences were found in trends at paired high and low elevation stations, with tropical pairs revealing more warming (and greater increases in freezing-level height) at mountain stations than at low elevations. This result is consistent with both the observed retreat of tropical glaciers and the minimal change in tropics-wide tropospheric temperatures over the past two decades. Overall, the analysis suggests that, on diurnal, seasonal, interannual, and multidecadal time scales, temperature variations at mountain locations differ significantly from those at relatively nearby (a few hundred kilometers) low elevation stations. These differences are greatest at the two surface levels, but can persist up to 2 km above the mountain site. Therefore, to determine the nature of climate variability at high elevation sites requires local observations, since large-scale patterns derived from low elevation observations may not be representative of the mountain regions. Conversely, temperature change in mountain regions should not be viewed as necessarily representative of global surface or tropospheric trends.