Scaling and intermittency in time of urban precipitation in Warsaw, Poland

Abstract

<div lang="en" class="abs"><p class="abs_header">Abstract</p><p><em>Study region:</em> A network of 25 rain gauges roughly covering an area of 517.2 km² of Warsaw, in Poland.</p><p><em>Study focus:</em> Scaling and intermittency of data series of rainfall intensity are investigated for each of the network rain gauges using five multifractal framework methods: spectral density analysis, functional box-counting, trace moment, probability distribution/multiple scaling, and double trace moment. Synthetic precipitation series are then generated by means of continuous universal random cascade models. Cascade generators are defined by universal parameters calculated for selected Warsaw rain gauges, or rain gauge clusters displaying similarity of multifractal parameters. The generated synthetic precipitation time series are subject to statistical evaluation by comparing the complementary cumulative distribution function and the intermittency calculated for synthetic versus observed data. Finally, a novel filtering algorithm is proposed to correct the intermittency characteristics of synthetic precipitation time series.</p><p><em>New hydrological insights:</em> Based on our analyses, the temporal structure of the recorded Warsaw precipitation time series is found to be a multifractal set characterized by a scaling behaviour over a wide range of scales. Furthermore, it has been observed that most Warsaw rain gauges, have a distinct similarity of multifractal properties. It has also been demonstrated that, for the first time in Poland, the universal continuous cascades could be used in practice for the generation of synthetic rainfall series at fine (1-min) temporal resolution.</p></div>