Can we estimate evaporation using commercial microwave links?

Spatial evaporation estimates are essential information for studying the water cycle, yet the amount of direct observations, such as Eddy-Covariance (EC) networks, are limited. Satellites can also provide spatial evaporation estimates, but these are based on indirect measurements of surface conditions and contain many assumptions. As a new method, we explore the potential of commercial microwave links (CMLs), such as used in cellular telecommunication networks, to be used as scintillometers. Scintillometers are dedicated instruments to measure path-integrated latent and sensible heat fluxes, which transmit electromagnetic radiation that is diffracted by turbulent eddies between transmitter and receiver, the so-called scintillation effect. CMLs are also line-of-sight devices that transmit electromagnetic radiation at similar frequencies as microwave scintillometers. Here, we estimate 30-min latent heat fluxes and daily evaporation estimates using the received signal level from a CML sampled at 20 Hz. To do so, we use data of a 38 GHz Nokia Flexihopper CML (formerly part of a telecom network) installed over an 856 m path at the Ruisdael Observatory near Cabauw, the Netherlands. We compare our results with estimates of a combined optical and microwave scintillometer setup, as well as an EC system.

Before obtaining flux estimates, we correct for the white noise present in the signal of the CML, based on power spectra of the CML and the microwave scintillometer, and obtain 30-min estimates of the structure parameter of the refractive index C_nn. Subsequently, to obtain the flux estimates from these C_nn estimates, we apply the two-wavelength method, in combination with the optical scintillometer, as well as a standalone energy-balance method (EBM), requiring net radiation estimates. Also, we consider the free-convection scaling of Monin-Obukhov similarity theory (MOST), instead of the complete scaling. An advantage of this scaling is that it removes the need for horizontal wind speed measurements, which are more difficult to obtain in complex environments. For the net radiation estimates, we use in-situ measured radiation and data products provided by the Satellite Application Facility on Land Surface Analysis (LSA SAF) of EUMETSAT.

Considering both turbulent heat fluxes, the two-wavelength method outperforms the EBM. The standalone EBM shows a reasonable performance, but depends heavily on the quality of the net radiation estimates. When aggregating our 30-min latent heat fluxes to daily evaporation estimates, the overall performance for both methods remains comparable. These daily evaporation estimates could also be useful for hydrological applications, e.g., for catchment-scale water budget studies. Moreover, application of the free-convection scaling instead of the complete MOST scaling results in a comparable performance for all methods. Before adoption of our methods to obtain evaporation estimates using CML networks, the influence of different CML design types and their sampling strategies in operational networks on the obtained flux estimates needs to be studied. If these are successfully addressed, CMLs could show a large potential to estimate evaporation, especially considering that existing CML networks are present at locations where evaporation observations are lacking.