Spatial and temporal analysis of mean and frequency variations of dry spells in Iran

Abstract

For probabilistic modeling of variations in averages and frequencies of length of dry spells in Iran and obtaining a set of suitable patterns in this relation, the data on daily precipitation from 41 synoptic meteorological stations for a 30-year period (1989-2018) were obtained from the Iran Meteorological Organization. A dry spell is defined as a sequence of consecutive dry days with precipitation equal to or less than 0.1 mm per day. In order to evaluate the probabilistic distribution of the mean length and frequency of short (1-10 days), medium (11-20 days), and long dry spells (21-30 days), a first-order Markov chain model, a second-order Markov chain model, an exponential distribution model, and a negative binomial distribution were used. The results showed that there were longer dry spells at more southern latitudes and shorter dry spells at more northern latitudes in Iran. At an annual scale, the longest dry spell in southeastern Iran was 351 days and the longest in northern Iran was 37 days. Simulation of seasonal variations in mean and frequency of different lengths of dry spells using the four probabilistic models showed that the exponential and the negative binomial distributions gave the closest approximations to the empirical mean of the length of dry spells at the studied stations. However, these estimates were relatively simulated well only for mean length of dry spells that lasted less than 10 days and were not simulated well at all for dry spells with lengths of more than 10 days. Moreover, considering that the average frequencies of medium dry spells (11-20 days) and long dry spells (21-30 days) indicated the length and severity of dry spells in different parts of Iran, it was observed that the exponential and negative binomial distributions provided closer approximations to the empirical averages compared to the other two models. In general, we can conclude that the selected models performed poorly in simulating the average length of dry spells but performed better in simulating the average number of short, medium, and long dry spells.