Answers are grounded in this paper and cite their sources.
A stochastic hourly precipitation model is developed by using the Iowa City‐Ralston Creek hourly precipitation record of 33 years length. Six divisions of the year are used to account for seasonal nonstationarity. Related wet time intervals, corresponding to storm events, are scheduled by an exponentially distributed interarrival time model. Intrastorm structure is described in terms of ‘storm segments,’ corresponding to the passage of a storm rainfall cell or a group of cells. It is found that the location in time and the duration of storm segments can be adequately modeled by independent random variables, thus avoiding traditional computation difficulties in modeling the persistence effects encountered in hourly data. Observed ‘persistence’ is preserved, since persistence results from the clustering of wet hours and the clustering of dry hours. The intensity and distribution of precipitation within storm segments are modeled by fitted log normal intensity probability distributions and by cataloging sample storm segment shapes. Generated data analysis indicates that very precise precipitation modeling and extremely high speed data generation result.
No takes yet. Share an insight, caveat, or question.
Croley et al. (1978) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: