Expanding direct evapotranspiration (ET) measurements with accessible and low-cost Variance Bowen Ratio instruments

Evapotranspiration (ET) is the largest terrestrial water loss, accounting for 60% of precipitation and driving the regional and global water cycle. Land-use modifications, agricultural practices, and atmospheric warming can intensify ET demand, thereby presenting novel societal challenges and necessitating policies to conserve, monitor, and penalize overconsumption of freshwater. Unfortunately, the spatially and temporally aligned ET data products needed for operational policy practices, including precision agriculture and municipal water conservation efforts, are generally inaccessible due to the high costs of infrastructure and instruments, as well as the level of expertise required to operate these systems. Thus, practitioners are often limited to coarse spatially interpolated ET products, reference ET estimates, or alternative proxies to generate protocols for water-use applications.

Here, we introduce a new direct ET measurement sensor, the ATMOS 51 Variance Bowen Ratio (VBR) Direct ET Sensor, that leverages the Variance Bowen Ratio and energy-balance closure methods. This technology employs high-frequency temperature and specific humidity measurements to compute the Bowen ratio and ET from measured and modeled energy fluxes within the sensor footprint. The VBR technique, in general, and the implementation in the ATMOS 51 specifically, provides a compact design that allows for quick deployment, minutes compared to days, owing to the minimal infrastructure requirements. Moreover, its low power requirements make it ideal for field logging and seamless integration into cloud-based installations. Thus, providing a user-friendly experience and reducing the barrier to applying ET measurements to operational irrigation and water management decisions.

In 2025, we conducted extensive intercomparisons of the ATMOS 51 VBR Direct ET Sensor against field-standard measurements, namely eddy covariance towers and weighing lysimeters. The intercomparisons spanned numerous agroecological systems (e.g., potatoes, beets, barley, pasture, deciduous hardwood forest, desert shrubland) to characterize the best sensor application and practices.

Across the spectrum of agroecological systems, the ATMOS 51-measured ET closely matched the eddy covariance-derived ET fluxes, with a root-mean-square error (RMSE) among half-hourly measurements ranging from 0.02 to 0.07 mm. As expected, the ET measured by ATMOS 51 was 6-10% higher than that from the eddy covariance, attributed to the differences between the open- and closed-energy-balance approaches. Due to the reliance on energy-closure-based methods, the Variance Bowen Ratio method and ATMOS 51 perform best in systems with moderate-to-high ET rates and homogeneous footprints. More xeric locations, which exhibit higher sensible heat fluxes, will likely require more deliberate constraints on the energy balance terms, including soil heat flux, to optimize ET estimates.