Microwave links as path-average rainfall sensors: Uncertainties associated with space-time variability

Recent studies have shown that microwave links are suitable for the measurement of rainfall. The very dense network of such links used for cellular communication (e.g. ~12,000 links in The Netherlands, i.e. ~1 km per square km) could potentially be used to complement operational radar rainfall estimates. Analyses of the uncertainties and errors related to issues raised in the recent literature for different link settings (e.g. frequency, link length, etc.) are presented here. These analyses can be used to better assess the uncertainties associated with rainfall estimates obtained using a single microwave link, and to determine the optimum settings for microwave links in a given climatology. To assess the uncertainty in microwave link rainfall estimation caused by the macrostructure of rainfall, rainfall intensities estimated by a high-resolution (120 m range resolution, 1.875 degrees azimuth resolution and 16 s sampling time) scanning X-band radar (SOLIDAR) are used. Time series of profiles of rainfall intensities are used to compute the path-integrated attenuation measured by a virtual microwave link and the path-average rainfall intensity over such a link. Uncertainties related to spatial and temporal sampling, power resolution and wet antenna attenuation are studied using these data for different link lengths and frequencies. To study uncertainties caused by the microstructure of rainfall, time series of drop size measurements in combination with wind velocity data are used. Range profiles of the rainfall intensity and attenuation at different microwave frequencies are generated using these data. Analyses of uncertainties are focused on the effects of inter- and intra-storm variation of the drop size distribution.