Fit for purpose evapotranspiration measurements: A multi-site comparison of lower-cost approaches across croplands and shrublands

Yeonuk Kim , Todd G. Caldwell, Justin Huntington, Meetpal Kukal, Maria Isabel Zamora Re, Dragomira Zheleva, Richard Jasoni, Christian Dunkerly, Murphy Gardner, Clarence Robison, Dinesh Gulati, Shawn Naylor, Christopher Pearson, John Volk, Bruno Comini de Andrade, Robert Lotspeich

Evapotranspiration (ET) links the water and energy cycles, yet research-grade eddy covariance (EC) systems may not be the best option to deploy operationally for water management. Emerging lower-cost ET approaches may expand fit-for-purpose observations, but their reliability across contrasting ecosystems, climates, and instrumentation configurations remains uncertain. Here, we evaluate lower-cost ET approaches across a heterogeneous network spanning agriculture and shrublands. We assess LI-710 limited-diagnostics 1-D EC, variance Bowen ratio (VBR), and residual energy (RE) ET against collocated benchmark EC, while accounting for systematic uncertainty associated with energy balance non-closure and random uncertainty arising from turbulent sampling.
Raw half-hourly LI-710 fluxes frequently contained spikes, but after quality control they generally agreed well with benchmark EC, with discrepancies at shrubland sites approaching the random uncertainty of EC. LI-710 nevertheless tended to underestimate ET across land covers, including in a replicated sensor experiment where inter-unit precision was high but a common negative bias remained. Energy-balance-closure-derived uncertainty ranges often overlapped between LI-710 and EC, demonstrating the continued value of measured available energy for diagnosing systematic uncertainty. VBR showed relatively low random uncertainty and performed best under intermediate evaporative conditions, but overestimated ET in dry shrublands and underestimated ET in high-ET, advective croplands. A sign-aware commercial VBR implementation partly reduced the latter bias, although departures from the assumed similarity of temperature and water-vapor turbulent transport remained an important limitation. RE performed comparatively well in high-EF irrigated croplands but poorly in shrublands.
These results provide field-based evidence for a fit-for-purpose framework: LI-710 is most useful when paired with rigorous quality control and energy-balance measurements, VBR is most robust under intermediate conditions, and RE is best suited to ET-dominated systems. Overall, the value of lower-cost ET measurements depends not only on sensor cost, but also on post-processing transparency, available energy measurements, ecosystem context, and the intended application.