Global land surface temperature (LST) data derived from satellite-based infrared radiance measurements are highly valuable for various applications in climate research. LST is a fundamental state variable for land surface processes and has long been available from satellite observations in the thermal infrared (TIR). LST is also increasingly important for studies assessing land surface conditions, e.g., studies of urban climate, evapotranspiration, and vegetation stress. LST is usually retrieved from satellite-based radiance measurements in the infrared (IR) or microwave (MW) range and it is well suited to provide global coverage. Due to the spatial scale mismatch between ground and satellite-based measurements and the heterogeneity of natural land surfaces, the validation of satellite LST data sets is a challenging task. However, in situ validation is essential for obtaining quantitative information on the accuracy of LST satellite products. Permanent, continuous in situ measurements of up- and downwelling TIR radiance allow the analysis of long timeseries of satellite LST observations, which can reveal seasonal cycles and potential deviations; these can originate from surface anisotropy, topography, heterogeneous land cover, or spatial variations in soil moisture. Many of the validation results obtained over the Namib gravel plains demonstrate the maturity of the LST products investigated over the past 15 years. They also highlight the need to carefully consider their temporal and spatial properties when using them for scientific purposes. Total uncertainty of in situ LST obtained from the TIR radiance measurements at the Gobabeb wind tower is estimated as 0.8 ± 0.12°C, which is highly accurate for a bare soil site with diurnal LST amplitudes of up to 40°C. Analyses of spatial representativeness performed on the meter to kilometer scale near Gobabeb Namib Research Institute yielded an absolute bias of 0.5°C compared to in situ LST, a value mainly achieved thanks to the Namib’s hyper-arid desert climate and the spatial homogeneity and temporal stability of the gravel plains. The Namib gravel plains were found to be suitable for validating LST with pixel sizes of up to 100 km2 and the continued availability of the in situ measurements from Gobabeb is of high importance for accurately validating and monitoring current and future satellite LST products.