UNCERTAINTIES IN CLIMATE TRENDS Lessons from Upper-Air Temperature Records

and reconciling rapid surface warming with upper-air temperatures, which were reported to have exhibited little if any warming over the satellite era (Folland et al. 2001; NRC 2000). This has led to a number of new satellite(Christy et al. 2003; Mears et al. 2003; Grody et al. 2004), radiosonde(Lanzante et al. 2003; Thorne et al. 2005), and reanalyses(Uppala et al. 2005) based upper-air temperature datasets. Many of these agree in characterizing specific short-time-scale atmospheric features [e.g., El Niño–Southern Oscillation (ENSO), volcanic response], but diverge substantially in their long-term large-scale mean trends (Seidel et al. 2004). Their construction has actually served to increase the spread in reported long-term trends, nominally increasing our uncertainty. We can no longer absolutely conclude whether globally the troposphere is cooling or warming relative to the surface. Clearly, however, the climate system has evolved in one unique way. Hence the challenge to the climate science community is to understand the reasons for the coherent differences between available datasets, and to discern the true climate evolution. The key first step is to understand the likely sources and causes of errors and biases. Only with this knowledge can we hope to truly reconcile the differences and gain a more complete and accurate picture of the true climate system evolution.