Generating Hourly Precipitation Time Series by Power Spectrum Simulation
Abstract
This study presents a precipitation time series generator by utilizing the Fourier transform. The described method can be employed in weather generators for design purposes, as well as hydrological and rainfall-runoff models. Precipitation time series at hourly time scale have high temporal intermittence properties. As a result, simulating precipitation time series only based on the distribution function will be insufficient. The Introduced power spectrum simulation method using the Fourier transform enables us to generate precipitation time series by preserving the statistical properties and occurrence of precipitation for the hourly time scale. In this regard, two types of transformations have been taken into account. First, a transformation between indicator and Gaussian correlations for preserving rain and no-rain hours. Second, a transformation between time and frequency domain and then power spectrum simulation for keeping the statistical measures. It is shown that the dry spells of observed and simulated precipitation time series are preserved. Furthermore, the auto correlation and the distribution function are compared between observed and simulated precipitation time series and the results show that the observed time series are within the simulation bounds not only for the simulate time scale, but also for the aggregated time scales.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H15U1287M