Commercial Microwave Links in Urban Drainage Modelling: On Deriving Precipitation Estimates for Various Link Lengths

It is currently unknown whether quantitative precipitation estimates (QPEs) from shorter or longer commercial microwave links (CMLs) are better suited to be used as input for urban drainage models. Short CMLs could deliver high spatial resolution, longer CMLs are probably less disturbed by systematic errors due to antenna wetting. We investigate the effect of various link lengths on urban runoff predictions using real-world CML data from an experimental catchment in Prague, Czech Republic. We run a rainfall-runoff model repeatedly, using QPEs from only a single CML (out of 19 available) every time and also their mean. We compare model predictions to observations at the drainage system outlet for 67 rainfall-runoff events from 3 summer seasons. We use i) unadjusted CML QPEs based on the power law between rainfall intensity and radio signal attenuation and ii) QPEs adjusted to 15-min rainfall data from distant rain gauges from a permanent network. We show that, when using the unadjusted CMLs, model performance is better for CMLs which do not provide overestimated QPEs, i.e. typically the longer ones. When using the adjusting-based method, we recommend using the mean of QPEs from all available CMLs. We could additionally benefit from CML QPEs if information about temporal dynamics of heavy rainfalls was not compromised when reducing the bias in these QPEs.