Continuous temperature observations at the stratosphere and lower mesosphere are rare. Radiometry opens the possibility of observing microwave emissions from two oxygen lines to retrieve temperature profiles at all altitudes. In this study, we present observations performed with a temperature radiometer (TEMPERA) at the MeteoSwiss station at Payerne for the period from 2014 to 2017. We reanalyzed these observations with a recently developed and improved retrieval algorithm accounting for the Zeeman line splitting in the line center of both oxygen emission lines at 52.5424 and 53.0669 GHz. The new temperature retrievals were validated against MERRA2 reanalysis and the meteorological analysis NAVGEM-HA. The comparison confirmed that the new algorithm yields an increased measurement response up to an altitude of 53-55 km, which extends the altitude coverage by 8-10 km compared to previous retrievals without the Zeeman effect. Furthermore, we found correlation coefficients comparing the TEMPERA temperatures with MERRA2 and NAVGEM-HA for monthly mean profiles to be in the range of 0.8-0.96. In addition, mean temperature biases of 1 and -2 K were found between TEMPERA and both models (MERRA2 and NAVGEM-HA), respectively. We also identified systematic altitude-dependent cold and warm biases compared to both model data sets.