A rainfall disaggregation scheme for sub-hourly time scales: Coupling a Bartlett-Lewis based model with adjusting procedures

Abstract

Many hydrological applications, such as flood studies, require the use of long rainfall data at fine time scales varying from daily down to 1 minute time step. However, in the real world there is limited availability of data at sub-hourly scales. To cope with this issue, stochastic disaggregation techniques are typically employed to produce possible, statistically consistent, rainfall events that aggregate up to the field data collected at coarser scales. A methodology for the stochastic disaggregation of rainfall at fine time scales was recently introduced, combining the Bartlett-Lewis process to generate rainfall events along with adjusting procedures to modify the lower-level variables (i.e., hourly) so as to be consistent with the higher-level one (i.e., daily). In the present paper, we extend the aforementioned scheme, initially designed and tested for the disaggregation of daily rainfall into hourly depths, for any sub-hourly time scale. In addition, we take advantage of the recent developments in Poisson-cluster processes incorporating in the methodology a 2 Bartlett-Lewis model variant that introduces dependence between cell intensity and duration in order to capture the variability of rainfall at sub-hourly time scales. The disaggregation scheme is implemented in an R package, named HyetosMinute, to support disaggregation from daily down to 1-minute time scale. The applicability of the methodology was assessed on a 5-minute rainfall records collected in Bochum, Germany, comparing the performance of the above mentioned model variant against the original Bartlett-Lewis process (non-random with 5 parameters). The analysis shows that the disaggregation process reproduces adequately the most important statistical characteristics of rainfall at wide range of time scales, while the introduction of the model with dependent intensity-duration results in a better performance in terms of skewness, rainfall extremes and dry proportions.

Event-based stochastic point rainfall resampling for statistical replication and climate projection of historical rainfall series

“…In the literature, the most acknowledged methods for stochastically generating synthetic rainfall series are based on Poisson cluster processes and rectangular pulse models such as Bartlett-Lewis (Koutsoyiannis and Onof, 2001;Wheater, 1994, 1993;Segond et al, 2007;Onof and Arnbjerg-Nielsen, 2009;Paschalis et al, 2014;Kossieris et al, 2016) or Neyman-Scott (e.g. Entekhabi et al, 1989;Cowpertwait, 1991Cowpertwait, , 2010Cowpertwait et al, 2002;Fowler et al, 2005;Burton et al, 2008;Paschalis et al, 2014;Sørup et al, 2016).…”

A doubly stochastic rainfall model with exponentially decaying pulses

“…by adding a jitter to a rectangular shape (Onof and Wheater 1994) or replacing the pulse by the clustering of a sequence of instantaneous pulses (Cowpertwait et al 2007); 9. Bartlett-Lewis rectangular pulse models were used in a disaggregation framework (Koutsoyiannis and Onof 2001;Kossieris et al 2016).…”

Censored rainfall modelling for estimation of fine-scale extremes

“…2, there is an ongoing tendency for stochastic mechanistic models to underestimate extremes at hourly and sub-hourly scales. Consequently, the practitioner is required to employ additional methods for better extreme value performance, including disaggregation Onof, 2000, 2001;Onof et al, 2005;Onof and Arnbjerg-Nielsen, 2009;Kossieris et al, 2018) and model fitting with more information about the variability of precipitation (Kim et al, 2013a). We propose a censored approach to mechanistic rainfall modelling for improved estimation of fine-scale extremes by focussing model fitting on the heavy portion of the rainfall time series.…”