Abstract. Atmospheric temperature and humidity fields as well as information on other meteorological parameters are nowadays retrieved from radiance measurements recorded by operational meteorological satellites. Up to now, the inversion procedures used only take into account crude information on the topography of the Earth's surface. However, the applied radiative transfer codes have to consider the Earth's surface as the lower boundary of the atmospheric model and, therefore, need a more precise mean elevation and a classification of the roughness of the Earth's surface. The influence of the topography of the Earth surface on retrieved temperature profiles is studied by using a physico-statistical inversion method. An objective analysis is made of the more precise mean elevation and derivation of roughness parameters using a new high-resolution digital elevation model (DEM) with a resolution of 500 m×500 m. By means of a geomorphological process and a newly developed topography rejection test, areas with a high surface roughness are localized and singled out. The influence of topography on the retrieved temperature profiles is illustrated by case studies. Changes are found predominantly in areas with a high variation of topography. Using the new high-resolution DEM and the topography rejection test, the geographical position of the calculated temperature profiles tends to be shifted towards areas with a small vertical variation of topography. The mean elevation determined by the new elevation model better characterizes the area observed. Hence, the temperature profiles can be calculated down to lower atmospheric levels. Furthermore, a guess profile better describing the atmospheric situation is selected by the more precise elevation. In addition, the temperature profiles obtained near the coast are improved considerably by the more precise determination of the surface property `sea' and `land,' respectively. Integration of an independent physical information such as topography leads, on average, to a slight improvement of the results of the physico-statistical inversion procedure. In some cases, however, significant improvements have been achieved regarding the desired accuracy of temperature profiles of the order of 1 K. In future, the spatial resolution of new high-resolution sounding instrumentation on the next generation of operational meteorological satellites will be increased. To exploit the resolving power of this new instrumentation, the different variation of the topography of the Earth surface, especially in regions with a high variation of topography, can be taken into account more precisely by using a high-resolution DEM.