Scaling and fractal properties of rainfall - analysis and modeling of rain gauge data
Abstract: Fractal theory involving scaling properties has been applied in a wide range of research areas, particularly within natural sciences. In the present thesis, fractal theory has been applied to the rainfall process. The main questions to be answered were: (1) is rainfall characterized by scaling and fractal properties, and, if so, (2) how can these properties be modeled and used in applied hydrology? Concerning question (1), a literature review revealed that two main types of scaling have been suggested for rainfall, simple scaling and multiscaling. Analysis results in the literature indicated that temporal and spatial rainfall intensities generally are characterized by multiscaling, whereas the fluctuations, i.e., intensity changes are typically simple scaling. Analysis of large amounts of rainfall data from a temperate and a monsoon climate performed within the present thesis generally supported the above findings. Furthermore, some new observations were made, particularly that a simple scaling behavior of spatial rainfall fields could be connected to frontal passages and a multiscaling behavior to convective fields. Concerning question (2), the literature review indicated that the observed scaling behavior may be due to a cascading of the rainfall from larger to smaller scales, an hypothesis that has been argued for on both theoretical and empirical grounds. For rainfall, various multiplicative random cascade modles have been developed, fitted to rainfall observations, and shown to reproduce the scaling behavior. In the present thesis, a muliplicative random cascade model involving some new concepts was developed for temporal rainfall. The possibility to use the model for rainfall disaggregation was tested, and the model proved able of disaggregating temporal rainfall volumes with several fundamental properties reproduced. To conclude, the application of fractal theory revealed that the rainfall process in time and space is characterized by symmetries extending over large scale ranges, which allow a description of the process over a range of scales simultaneously. This is a significant advancement as compared to previous descriptions, and the present approach has therefore a potential to improve our ability to model rainfall.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.