This communication assesses six methods estimating UltraViolet A and B (UV-A and UV-B) fluxes from satellite imagery, numerical weather models or ground measurements. The UV estimates from each method are compared to coincident 15 minutes in-situ measurements collected at one location in Uruguay from September 2015 to January 2019. The first method "LAAM" (Locally-Adapted Antón Martínez) combines Global Horizontal Irradiance (G) measured on site with satellite-derived daily Ozone concentration. The second method "Wald" uses an empirical model onto satellite-retrieved solar broadband irradiance at the surface (SSI) produced by HelioClim-3 version 5 (HC3v5) to derive UV fluxes. The third method named "CAMS-UV", is one of the outputs of ECMWF numerical weather model. The three remaining methods are respectively named "Weighted_Kato HC3v5", "Discretized _Kato HC3v5" and "DWD SARAH-3". They rely on more sophisticated modelling of the atmosphere in cloud-free conditions using radiative transfer modelling combined to a cloud modification factor (or cloud extinction) derived from HC3v5. Outside an underestimation observed for the UV-B range for both CAMS-UV (-20 %) and for the empirical model (-29 %), methods demonstrated their ability to collect the temporal variability of the signal of the instrument on-ground; biases ranges from-2 to 4 % for UV-A and from 0 to 10 % for UV-B, RMSE are close to 15 % and almost all correlation coefficients exceed 0.96. This analysis gives precious elements for discussion about the performance of models mainly developed and validated over Europe and Africa.