Mountain permafrost modelling in remote, continental mountain ranges (e.g. Russian Altai Mountains) holds several difficulties due to the limitations these environments pose. The lack of meteorological input data and impossibilities for BTS-validations (Bottom Temperatures of winter Snow cover) makes conventional modelling strategies inapplicable. Statistical methods, however, based on correlation coefficients between different parameters, offers good alternative but requires lots of observations to be significant. As a solution, spatially covering land surface temperature (LST) values might be used as a proxy replacing the interpolated air and near ground surface temperatures. This article proposes 2 strategies, one statistical and one adapted TTOP (temperature at the top of permafrost), based on remote sensing data and ground measurements. Although these methods seem promising, they require a detailed understanding of the relation between LST and the air and near ground surface temperature. Therefore, before installing field equipment, we compared filtered MODIS LST time-series with corresponding ground temperature measurements recorded by Sergei Marchenko (Geophysical institute of Alaska, Fairbanks) in the Ulandryk Valley. Despite the cloudy conditions of this test site, a good correlation is showed between both time-series.