An Investigation of spatio-temporal variability of annual rainfall and maximum daily rainfall in western Iran

Extended abstract

Introduction

One of the important consequences of climate change is a change in the frequency and intensity of rainfall. In fact, it can be said that as a result of this phenomenon (climate change), in many parts of the world, the frequency and intensity of maximum rainfall have increased. However, the type and severity of these changes vary from region to region (IPCC, 2012: 5). In recent decades, in the framework of climate change, several studies have been conducted on the trend of rainfall in most parts of the world. Most of these studies are based on parametric or non-parametric methods such as linear regression analysis, Spearman's test, age gradient test, and Mann-Kendall test. However, the use of these methods requires some limiting assumptions. In addition, these tests only show the trend of changes in the average time series, while they do not provide any information about the trend of changes in different time series classes. In this regard, Sen (2012: 1044) proposed the ITA method, which addresses the above-mentioned problems in addition to allowing a trend to be identified in a series of "low", "medium" and "high" values.



Materials and methods

In this study, in order to achieve the defined goal, the basic method of percentiles was initially used. Since the choice of the threshold value of the base percentile is a matter of taste and does not reflect the intensity-frequency of heavy rainfall, the type 1 equation probabilistic distribution function (Gamble) was used to determine the heavy precipitation threshold to provide a criterion that results in intensity-frequency of occurrence, while it is based on time series distribution of the precipitation data. For this purpose, first the data of rainy days related to 6 synoptic stations in the west of the country from 1961-2019 were obtained from the Meteorological Organization. Then, in order to calculate the maximum amount of daily rainfall during different return periods, the Gamble distribution function was used, based on which the heavy rainfall threshold was defined for the stations in the region and the maximum daily rainfall values during the return periods of 2, 5, 10, 15, 20, 25, 50 and 100 years were calculated. Finally, MK and ITA tests were used to determine the trend of these precipitations. The ITA method was proposed by Sen. Despite its simplicity, it does not require any presuppositions and is more capable than other non-parametric methods because this method is able to identify hidden trends and internal trends in time series in addition to uniform trends. Therefore, in this study, ITA method was used to identify the uniform and non-uniform trends in the maximum daily rainfall and the total annual rainfall in the west of the country.



Research Findings

After determining the thresholds of heavy rainfall using the Gamble distribution function, it became clear that the average rainfall of more than 37 mm in the western region of the country is considered heavy rainfall. The heavy rainfall index of Dezful station was higher than other stations, which indicates the high intensity of daily rainfall in this station. In this study, heavy precipitation threshold was calculated for all stations by applying the percentile method on rainy days with a minimum threshold of 1 mm. Then, ITA method was used to analyze the trend and determine the behavior of total annual rainfall and maximum daily rainfall in the study area. Application of this method on the total annual rainfall of Khorramabad station showed that middle floor precipitation (430-650) of this station is decreasing, while upper and lower floor precipitation was decreasing. In Hamedan station, the precipitation of the middle class (240-240) showed a significant decreasing trend, but no significant trend was observed in the lower class (less than 240) and the upper class (above 440). In Kermanshah, Dezful, Ahvaz and Abadan stations, the trend specified in all classes was decreasing and uniform. The application of Man-Kendall method on the total annual rainfall in the study area showed that the trend of these rains is decreasing in all stations and is significant in Khorramabad, Hamedan, Kermanshah and Dezful at the level of 95%. Regarding the application of ITA method for maximum daily rainfall of Gamble, the results showed that in Khorramabad station, rainfall with a return period of 5, 10, 15, 20 and 25 years of this station increased and precipitation with a return period of 2 years showed no particular trends. At Abadan station, rainfall with a return period of 50 years showed a decreasing trend. In Dezful and Ahvaz stations, the trend marked in all classes was a decreasing trend. In general, no specific trend was observed in Hamedan and Kermanshah stations. Regarding the maximum daily rainfall of Gamble, the results of Mk showed a negative trend in Hamedan, Kermanshah, Ahvaz and Abadan stations and a positive trend in Khorramabad and Hamedan, but only in Ahvaz station, the trend was significant at 95%. In the present study, some conflicting results have been obtained by comparing the ITA and MK methods. This shows the advantage of this method over other process tests. For example, while the Mann-Kendall test on Khorramabad station showed a significant decrease in the total annual rainfall of this station, the Sen method showed a different trend from this method. In fact, according to ITA, it was found that the total rainfall in Khorramabad had an uneven trend, which was divided into three classes, and the precipitation class showed 430 to 650 decreasing trends less than 430 and more than 650 increasing trends in this station.

Result

The results of ITA method showed that on an annual scale, there is a non-uniform trend in rainfall in Khorramabad and Hamedan. But in Kermanshah, Dezful, Ahvaz and Abadan stations, the trend marked in all classes was decreasing and uniform. On a daily scale, the results showed that precipitation with a return period of 5, 10, 15, 20 and 25 years of this station is an increasing trend and precipitation with a return period of 2 years is without a trend. At Abadan station, rainfall with a return period of 50 years showed a decreasing trend. In Dezful and Ahvaz stations, the trend marked in all classes was a decreasing trend. In Hamedan and Kermanshah stations, in general, no specific trend was observed, but in more detail, it can be said that the trend of rainfall on the upper floor of Hamedan station was increasing, whereas the trend was decreasing in Kermanshah station. The results of Mann-Kendall test also showed that on an annual scale, all stations have a decreasing trend and on a daily scale, Hamedan, Kermanshah, Ahvaz and Abadan stations have a negative trend, although Khorramabad and Hamedan stations showed a positive trend in precipitation. In fact, when there is a non-uniform trend in the time series, the MK test shows that the trend is insignificant, but the ITA method detects such non-uniform trends and makes hidden time series information available.

Keywords: Gamble, ITA, Limit rainfall, West of Iran