Spatial downscaling of rain fields using satellite and high-resolution radar data
Loading...
URL
Journal Title
Journal ISSN
Volume Title
Insinööritieteiden korkeakoulu |
Master's thesis
Authors
Date
Department
Major/Subject
Mcode
ENG29
Degree programme
Language
en
Pages
44
Series
Abstract
Precipitation modeling relies heavily on the fact that precipitation has been long known to scale both temporarily and spatially. There have been attempts to blend rainfall data from multiple sources with promising results. Hydrological models (such as runoff models) require high-resolution data to produce reliable results, but quite often available data does not have the necessary resolution. Downscaling is a technique that infers high-resolution data from a source with low resolution. A realistic rain field can be synthetically generated using just four parameters - average of rain intensity, standard deviation of rain intensity, wetted area ratio (proportion of "rainy" pixels over all pixels), and spectral exponent (parameter that describes scaling behavior). In this thesis, the extent to which publicly available satellite data (low resolution) can be spatially downscaled for use in hydrological models was investigated. The region under examination was the Bonn area in western Germany. Available data consisted of a time series of rain intensity fields, from ground-based radar and space-based satellite observations. The rain intensity fields were converted to reflectivities. A rain event identification scheme was proposed and tested on the available data, which returned similar results for the events determined from the radar and the satellite observations. Pairwise relationships between the average of field reflectivity and three parameters (wetted area ratio, standard deviation of the field reflectivities, and spectral exponent) were established. Satellite data was then downscaled based on a method that involved statistical downscaling and parameter estimation from the established relationships. Downscaled data was compared to the high-resolution radar data, and an additional comparison between the upscaled radar and original satellite data was conducted. The downscaling method showed promising results. The comparisons did not show significant differences in the observed rain intensities, reflectivities or standard deviation of reflectivities. Maximum values of reflectivities of rain fields were lower for the downscaled satellite data compared to the fields measured by the radar. Future research may determine whether similar parameter relationships exist in different areas of the world (in particular oceans and mountainous areas where the radar and rain gauge coverage is lacking) and whether the relationships depend on the type of rain. Additionally, this work may be expanded by incorporating the temporal aspect to the downscaling scheme.Description
Supervisor
Koivusalo, HarriThesis advisor
Kokkonen, TeemuNiemi, Tero