While the efforts of the scientific community have lead to aremarkable knowledge of the climate system, there are still many opentopics left to be resolved. One of these topics is concerned withupper-air climate change. There has been a debate about upper-airtrends in the last decades, and the controversy is stillongoing. Climate model data, reanalysis datasets, and observationalrecords show large uncertainty ranges in their trend values, many ofthem even disagree in important aspects of trend characteristics.This thesis investigates methods to analyze upper-air climatedata. Firstly, an innovative approach to deal with large atmosphericdatasets is presented: Interactive visual exploration is shown to be avaluable tool to complement classical statistical methods. It opensnew opportunities for data analysis because it does not require priorknowledge of data characteristics, thus enabling the user to come upwith new hypothesis about the data. Several datasets from climatemodels, reanalyses and observations are explored with sophisticatedinteractive visualization techniques, showing how these methods makeit easy to determine potentially unknown patterns and characteristicsin the data. Secondly, Radio Occultation (RO) as a recent upper-airdataset with high accuracy is employed as reference for stratosphericradiosonde and (Advanced) Microwave Sounding Unit ((A)MSU)climatologies. Special care is taken to account for sparse sampling inthe radiosonde and RO climatologies. The results show good agreementof radiosondes and RO, while (A)MSU and RO trends are found to differsignificantly. The advantages of RO with homogeneously distributedobservations, high vertical resolution and accuracy are confirmed,helping to overcome problems of conventional upper-air data. Thesuitability to serve as reference for other observational records isdemonstrated, thus narrowing the large structural uncertaintiesinvolved in today?s upper-air climate records.