The specification of the upper atmosphere strongly relies on solar proxies that can properly reproduce the solar energetic input in the UV. Whilst the microwave flux at 10.7 cm (also called F10.7 index) has been routinely used as a solar proxy, we sh
ow that the radio flux at other wavelengths provides valuable complementary information that enhances their value for upper atmospheric modelling. We merged daily observations from various observatories into a single homogeneous data set of fluxes at wavelengths of 30, 15, 10.7, 8 and 3.2 cm, spanning from 1957 to today. Using blind source separation (BSS), we show that their rotational modulation contains three contributions, which can be interpreted in terms of thermal bremsstrahlung and gyro-resonance emissions. The latter account for 90% of the rotational variability in the F10.7 index. Most solar proxies, such as the MgII index, are remarkably well reconstructed by simple linear combination of radio fluxes at various wavelengths. The flux at 30 cm stands out as an excellent proxy and is better suited than the F10.7 index for the modelling the thermosphere-ionosphere system, most probably because it receives a stronger contribution from thermal bremsstrahlung. This better performance is illustrated here through comparison between the observed thermospheric density, and reconstructions by the Drag Temperature Model.
