3D downscaling model for radar-based precipitation fields

Abstract

The generating of rainfall fields with a higher resolution than so far observed and with realistic features is a challenge with multiple applications. In particular it could be useful to quantify the uncertainty introduced by the different sources of error affecting radar measurements, in a controlled simulation framework. This paper proposes a method to generate three-dimensional high-resolution rainfall fields based on downscaling meteorological radar data. The technique performs a scale analysis of the first radar tilt field combining a wavelet model with Fourier analysis. In order to downscale the upper radar elevations and with the aim of preserving the vertical structure, a homotopy of the observed vertical profiles of reflectivity is performed. Preliminary evaluation of the technique shows that it is able to generate realistic extreme values and, at the same time, partially reproduce the structure of small scales. Zusammenfassung

Evaluation of an early flood warning system in Bamako (Mali): Lessons learned from the flood of May 2019

“…Their difference lies in the systems and models used to simulate water levels and inundation zones. Data‐driven systems rely on historical flood databases and use statistical and stochastic models (Llort et al, 2014; Yang et al, 2015) and, more recently, machine learning techniques (Garcia et al, 2016; Lee et al, 2020), to predict water levels. The second family of systems uses hydrological (Bouvier et al, 2018; Javelle et al, 2016) or hydraulic models (Chitwatkulsiri et al, 2021; Hofmann & Schüttrumpf, 2019) for water level, discharge, or inundation plane prediction.…”

A study of the structure of radar rainfall and its errors

“…The wavelet transform gives a representation of the signal as a function of both frequency and time, and allows us to study the local variability of fields at different scales (see, e.g., Foufoula-Georgiou and Kumar 1995). The second downscaling approach is based on the fact that the standard deviation of rainfall fluctuations standardized by the scaling component (defined via the wavelet transform) seem to obey a simple scaling law over the different available scales (Harris and Foufoula-Georgiou 2001;Llort et al 2006;Perica and Foufoula-Georgiou 1996a;Venugopal et al 1999). Extrapolating this law permits to simulate rainfall fluctuations over the non-observed smaller scales.…”

“…In this method, rainfall signals are decomposed using the wavelet transform (Mallat 1989) with the Haar base (Haar 1910). The sample distributions of the fluctuation components standardized by the scaling component are assumed to be Gaussian with zero mean (Llort et al 2006;Perica and Foufoula-Georgiou 1996a). Moreover, standard deviations of those signals follow a simple scaling law.…”

“…With the idea of not only simulating the right distribution of the standardized wavelet fluctuations when downscaling, but also their autocorrelation, we studied the Fourier power spectrum of the fluctuations field. But using this information in the simulation process did not much influence the final field's final autocorrelation (see Llort et al 2006).…”

“…To generate these reference fields, we use reflectivity information measured close to the radar as the starting point of a downscaling technique. The downscaling technique used (described in detail in Llort et al 2006 and in Chapter 2 of this document) is based on a combination of wavelet scaleanalysis and homotopic techniques. In this technique, the first radar tilt is downscaled using a 2D wavelet model: by means of a given Haar-base wavelet scale variability analysis, the observed variability at large scales is extrapolated to the new small scales created.…”