We investigate how evaporation changes depending on the scales in the Altiplano region of the Atacama Desert. More specifically, we focus on the temporal evolution from the climatological to the sub-diurnal scales on a high-altitude saline lake ecosystem. We analyze the evaporation trends over 70 years (1950–2020) at a high-spatial resolution. The method is based on the downscaling of 30 km ERA5 reanalysis data at hourly resolution to 0.1 km spatial resolution data, using artificial neural networks to analyze the main drivers of evaporation. To this end, we use the Penman open-water evaporation equation, modified to compensate for the energy balance non-closure and the ice cover formation on the lake during the night. Our estimation of the hourly climatology of evaporation shows a consistent agreement with eddy-covariance (EC) measurements and reveals that evaporation is controlled by different drivers depending on the time scale. At the sub-diurnal scale, mechanical turbulence is the primary driver of evaporation, and at this scale, it is not radiation-limited. At the seasonal scale, more than 70 % of the evaporation variability is explained by the radiative contribution term. At the same scale, and using a large-scale moisture tracking model, we identify the main sources of moisture to the Chilean Altiplano. In all cases, our regime of precipitation is controlled by large-scale weather patterns closely linked to climatological fluctuations. Moreover, seasonal evaporation significantly influences the saline lake surface spatial changes. From an interannual scale perspective, evaporation increased by 2.1 mm yr−1 during the entire study period, according to global temperature increases. Finally, we find that yearly evaporation depends on the El Niño–Southern Oscillation (ENSO), where warm and cool ENSO phases are associated with higher evaporation and precipitation rates, respectively. Our results show that warm ENSO phases increase evaporation rates by 15 %, whereas cold phases decrease it by 2 %.