Modeling the multi-scale temporal variability of fog harvesting potential in the coastal Atacama region

The inland advection of the well-formed marine stratocumulus cloud deck in the tropical Southeast Pacific produces semi-permanent fog banks in the hyperarid coastal mountains of the Atacama Desert. These fog banks represent the sole water input to highly adapted xeric ecosystems and can serve as a potentially tappable water resource for human consumption. Whether to sustain ecosystems or domestic water consumption, our understanding of long-term fog-harvesting variability is very limited, as observations are short-term and intermittent. This observational gap makes it difficult to understand what is driving fog-harvesting variability at interannual (relation with ENSO), seasonal, and sub-diurnal scales. Therefore, hindering our ability to assess the feasibility of exploiting this natural resource at the long term. In this work, we propose using the Advective fog Model for Arid and semi-arid Regions Under climate change (AMARU; Lobos-Roco et al., 2025) to study the long-term variability of harvesting potential resulting from the interaction of stratocumulus clouds with coastal topography. The model inputs are ERA5 reanalysis time series between 1950 to 2023, which have been downscaled to meteorological observations using artificial neural networks. Model outputs are compared with historical fog water harvesting observations from 1997 to 2023 in Alto Patache fog oases (20.8°S; -70.1°W), showing R²~0.8 and a regression slope~1. Our modelling results show that the coastal Atacama Desert is a promising site for fog harvesting, with water volumes ranging from 2.9 L m^-2 per day to 9.5 L m^-2 per day over seven decades, and a subtle trend toward an increase of 3.46 L m^-2 pear year. At the interannual scale, fog harvesting is modulated by the (in)harmonization between the El Niño Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO) phases. For example, during warm PDO, ENSO correlates positively with fog harvesting, while during cool PDO, ENSO correlates negatively with fog harvesting. The modulation of ENSO-PDO in fog harvesting has decreased over the decades, probably due to climate change. From 1950 to 2023, fog-harvesting seasonality has been narrowing, with the fog-harvesting season starting later and ending earlier, but with higher water volumes during the fog season. Finally, at the diurnal scale, our model results demonstrate that fog harvesting is more controlled by air-liquid water content (cloud density) at night and by wind speed (cloud density transport) in the afternoon. Our study contributes to disentangle fog-harvesting variability across multiple temporal scales, thereby enhancing our capacity to assess ongoing and future multipurpose and large-scale fog-harvesting projects in coastal deserts.