Interpretability of negative latent heat fluxes from Eddy Covariance measurements during dry conditions

Non-rainfall water can play a critical role in many ecosystems, but is understudied in most regions due to a lack of continuous, specialized measurements. One of the most commonly used techniques to quantify in situ ecosystem water fluxes is Eddy Covariance (EC). However, its use for the quantification of the two most famous non-rainfall water sources, dew and (radiation) fog, is limited because they often occur under humid conditions and nighttime stable stratification, making EC measurements particularly uncertain or non-valid.

Here we describe how a non-rainfall water input observed under dry conditions, namely water vapor adsorption by soil particles (VWA), can be monitored using existing eddy covariance datasets, giving insight into this little-studied soil water source. Unlike dew and radiation fog, atmospheric stability is not a prerequisite for WVA. Instead, WVA is driven by a highly negative soil matric potential inducing water vapor to condensate already at relative humidity < 100 %. Therefore, EC measurements may be more suitable to detect and quantify this flux than for dew and fog.

In this study, we test EC measurements for inferring WVA by comparing them to observations from large-weighing lysimeters, where the latter can be considered as a reference system for the measurement of WVA. Our aim is to explore the potential and limitations of the EC technique to detect and quantify WVA. We assess the quantitative and qualitative agreement between WVA estimated with the lysimeters and negative (downward) LE fluxes from EC. Our analysis uses four years of observations from a semi-arid tree-grass ecosystem and one year of a temperate agricultural ecosystem during the 2018 drought.

Our results show that during dry conditions the water vapor gradient between the relatively humid atmosphere and the dry soil pores leads to WVA in both ecosystems. We find a decent agreement between the timing of fluxes detected as WVA with lysimeters and with EC instruments, but the magnitudes (i.e. the amount of flux) differ. Furthermore, we aim to characterize the conditions under which negative LE fluxes from EC measurements can and should be interpreted as WVA. This way, our study expands the possibilities to investigate the relevance of WVA as a non-rainfall water source and, more generally, sheds light on a mostly overlooked aspect of land-atmosphere interaction during dry conditions in different ecosystems.