Quantifying the uncertainty of evaporation estimates with different sensors above homogeneous and heterogeneous terrain

Evaporation is typically the largest flux of the water balance on a yearly basis in the mid latitudes, often larger than discharge. Yet, it is the most difficult component to measure directly. For this reason, hydrological models often include evaporation as a fraction of potential evaporation or as a residual term. Especially above heterogeneous terrain, determining evaporation is complex and its accuracy highly dependent on e.g. measurement scale, wind speed and direction (footprint), and type of vegetation.

Multiple measurement techniques are available to estimate evaporation (like eddy covariance, scintillometry, or flux variance method). More recently new, lower cost eddy covariance sensors by LiCor (Lincoln, NE, USA) have been introduced to the market, i.e. the LI-710 and LI-720. Some recent intercomparison studies have shown the relevance and downsides of lower cost eddy covariance systems compared to conventional systems. However, these studies did not include other techniques besides eddy covariance techniques, to estimate evaporation. All measurement techniques have their particular uncertainties due to differences in measuring principle, footprint, sensor accuracy, etc. This can lead to large (~35% at mid-day) differences in the obtained evaporation estimates. In this study we aim to quantify the differences in evaporation estimates by comparing several measuring techniques during a summer season over a homogeneous and a heterogeneous terrain in the Netherlands: Cabauw (homogeneous well-watered grass) and Herenboeren Wenumseveld (heterogeneous agricultural site). The measurement techniques included in the intercomparison are: a conventional eddy covariance system, a dual wavelength scintillometer system (optical and microwave), the flux variance method, and the novel LI-710 and LI-720 sensors.

First results, based on a measurement campaign for the summer of 2025, show that the difference in determined evaporation expressed as the latent heat flux between the different sensor techniques is highly variable (up to 250 W/m²), especially above heterogeneous terrain, peaking around mid-day. Above homogeneous terrain, the different sensors are much more in agreement (up to 100 W/m²). During the summer of 2026, the measurement campaign will be repeated to gain more insights in the sensor uncertainty range by adding a LI-720 sensor and by performing a tilt experiment with the LI-710. The latter is meant to test sensor sensitivity for wind direction and turbulence variations, since that has not been done in previous studies with these specific lower cost sensors. This will result in a better understanding of the uncertainties associated with evaporation measurements, which can be especially valuable with regards to water and energy balance closure and for catchment modelling.