Use of commercial microwave links as scintillometers: potential and limitations towards evaporation estimation

A scintillometer is a common instrument to measure path-integrated evaporation and sensible heat fluxes. It consists of a transmitter and a receiver separated along a line of sight of several hundreds of meters to a few kilometers. Turbulent eddies and the associated refractive index fluctuations along the path between transmitter and receiver cause diffraction of the transmitted microwave beam (known as the scintillation effect). Scintillometers have been designed to measure the full spectral range of the signal intensity fluctuations caused by this phenomenon and quantitatively link these fluctuations to the turbulent heat fluxes. Commercial Microwave Links (CMLs), such as used in cellular telecommunication networks, also make use of microwave signals. However, CMLs are obviously not designed to capture scintillation fluctuations. We investigate if and under what conditions CMLs can be used to obtain the structure parameter of the refractive index, C_nn, which would be a first step in computing turbulent heat fluxes with CMLs using conventional scintillation theory. To do so, we use data from three collocated microwave links installed over a 856 m path at the Ruisdael Observatory near Cabauw, the Netherlands. We compare received signal intensity fluctuations sampled at 20 Hz from two 38 GHz CMLs formerly employed in telecom networks in the Netherlands, a Nokia Flexihopper and an Ericsson MiniLink, with measurements from a 160 GHz microwave scintillometer (RPG-MWSC) sampled at 1 kHz and an eddy-covariance system. After comparison of the unprocessed C_nn, we reject the Ericsson MiniLink, because its 0.5 dB power quantization was found to be too coarse. Based on power spectra of the Nokia Flexihopper and the microwave scintillometer, we propose two methods to correct for the white noise present in the signal of the Nokia Flexihopper: 1) we apply a high-pass filter and subtract the noise based on a calibration with the microwave scintillometer; 2) we select parts of the power spectra in which the Nokia Flexihopper behaves similar to the microwave scintillometer and correct for the missed part of the scintillation spectrum based on scintillation theory. Calibration on the microwave scintillometer provides the best possible C_nn estimates for the Nokia Flexihopper, although this is usually not feasible for CMLs. Both of our proposed methods show an improvement of C_nn estimates in comparison to unprocessed estimates, though with a larger uncertainty than the reference instruments. Overall, our study illustrates the potential to use CMLs as scintillometers, especially due to their large global coverage, but also lays out major drawbacks, most of which are related to unfavourable design choices made for commercial microwave links.