Insights Into Water Vapor Uptake by Dry Soils Using a Global Eddy Covariance Observation Network

Abstract

The exchange of water vapor between soil and atmosphere is a key component of land–atmosphere interactions, especially under dry conditions. Soil water vapor adsorption (SVA) occurs when the atmospheric water vapor pressure is greater than the soil air vapor pressure, which triggers the transport of water vapor from the atmosphere to the soil and its retention on the soil particle surface in liquid form. This process is largely caused by soil hydraulic properties and may play a significant role in dryland hydrology, yet remains understudied due to a lack of continuous, direct observations. In this study, we use globally distributed eddy covariance flux tower data to detect and characterize patterns of soil water vapor adsorption. We verify the consistency between negative latent heat fluxes as an indicator of water vapor movement toward the ground and the theoretical understanding of SVA. Our results reveal a relationship between the direction of the vapor gradient, as indicated by the direction of the latent heat flux, soil moisture, and near-surface relative humidity, which is consistent with the understanding of a phase equilibrium at the pore scale of the soil. Distinguishing between random noise and physically explainable occurrences of negative latent heat fluxes enables the characterization of SVA occurrence in eddy covariance observations. SVA is detected most frequent in arid and semi-arid regions, particularly in ecosystems with sparse vegetation such as savannas and dry shrublands. On average, SVA occurs for 4 ± 1.1 h per night, and may last up to 7 h in some locations. In certain sites, SVA occurs on more than 150 nights per year. These findings suggest that the eddy covariance method can help monitor SVA occurrence. Mapping the spatiotemporal patterns of SVA enhances our understanding of dryland land–atmosphere water fluxes and uncovers a previously overlooked component of the terrestrial water cycle.