Abstract
Measurement of rain is made difficult by the high variability of the precipitation process, down to raindrop scale. Point measurements are generally accurate, but their lack of spatial representativeness is a significant limitation. Weather radars indirectly measure rainfall over large regions, but the microphysical properties of the rain being measured must be known or inferred in order to compute rainfall quantities from radar data. The raindrop size distribution (DSD) statistically describes the microstructure of rain. While the DSD is often assumed to be uniform in space, it is in fact highly variable. The work in this thesis contributes to the understanding of the small-scale variability of the DSD and its effects on the measurement of rainfall. The methods shown were developed using data from a network of disdrometers and radars over a 13 x 7 km2 field site in Ardeche, France. This area experiences heavy Mediterranean rainfall. A technique to improve the accuracy of DSD measurements made by Parsivel disdrometers is proposed. The method uses a 2D-video-disdrometer as a reference instrument. A new geostatistical method for spatial interpolation and stochastic simulation of the experimental DSD is provided. It can estimate or simulate the non-parametric DSD at an unmeasured location, conditional on nearby measurements. Leave-one-out testing showed that estimates were produced with minimal bias. The correction and simulation techniques were used together to investigate the small-scale variability of the DSD in the study region. DSD variability was studied in detail over two typical scales, corresponding to the footprint size of the Global Precipitation Mission (GPM) space-borne weather radar, and a typical size for an operational numerical weather model pixel. It is shown that the assumption that a point measurement of the DSD represents an areal estimation introduces error that increases with areal size and drop size. Satellite and weather model rainfall retrieval algorithms that correspond to these two typical domains were tested, and while it was found that rain intensity and radar reflectivity were well retrieved, other DSD properties were often not representative of the sub-grid process. Double-moment normalisation provides a compact representation of the DSD, under the assumption that the normalised version DSD is invariant. This assumption was tested using instrument network data in France, Switzerland, and the United States. It is shown in this work that for practical purposes, the double-normalised DSD can be assumed invariant through horizontal and vertical displacement. Using this assumption, a new method for retrieval of the DSD from polarimetric radar data is proposed. The new DSD-retrieval technique performs as well or better than an existing method. An application of multifractal analysis to high-resolution snowfall data from the Swiss Alps is presented. Scaling of snowfall was observed in reconstructed vertical columns, at scales from about 35 metres to two metres, with no scaling observed at smaller scales. Weak scaling was observed in time series. The results indicate that at small (sub-metre or sub-minute) scale, snowfall appears homogeneously distributed.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.