Detecting Trends in Extreme Temperature and Precipitation events with Different Return Periods in Iran

In recent years, research on climate change has increased due to its economic and social importance and the damages of increasing extreme events. In most studies related to climate change, detecting potential trends in the long-term average of climate variables have been proposed, while studying the spatio-temporal variability of extreme events is also important. Expert Team on Climate Change Detection and Indices (ETCCDI) has proposed several climate indices for daily temperature and precipitation data in order to determine climate variability and changes based on R package.Various methods have been presented to investigate changes and trends in precipitation and temperature time series, which are divided into two statistical categories, parametric and non-parametric. The most common non-parametric method is the Mann-Kendall trend test. One of the main issues of this research is the estimation of each index value in different return periods. The return period is the reverse of probability, and it is the number of years between the occurrence of two similar events (Kamri and Nouri, 2015). Accordingly, choosing the best probability distribution function is of particular importance in meteorology and hydrology.Despite of the enormous previous studies, there is no comprehensive research on the estimation of extreme indices values for different return periods. Accordingly, this study focuses on two main goals: First, the changes in temperature and rainfall intensity are analyzed by analyzing the findings obtained from extreme climate indices (15 indices) and then (second) estimating the values of the indicators for three different return periods (50, 200 and 500 years).

In this research, the daily data of maximum, minimum and total annual precipitation of 49 synoptic stations for 1991-2020 were used to analyze 15 extreme indices of precipitation and temperature. Namely, FD, Number of frost days: Annual count of days when TN (daily minimum temperature) < 0oC; SU, Number of summer days: Annual count of days when TX (daily maximum temperature) > 25oC, ID, Number of icing days: Annual count of days when TX (daily maximum temperature) < 0oC; TXx, Monthly maximum value of daily maximum temperature; TNx, Monthly maximum value of daily minimum temperature; TXn, Monthly minimum value of daily maximum temperature; TNn, Monthly minimum value of daily minimum temperature; DTR, Daily temperature range: Monthly mean difference between TX and TN; Rx1day, Monthly maximum 1-day precipitation; Rx5day, Monthly maximum consecutive 5-day precipitation; SDII Simple precipitation intensity index; R10mm Annual count of days when PRCP≥ 10mm; R20mm Annual count of days when PRCP≥ 20mm; CDD. Maximum length of dry spell, maximum number of consecutive days with RR < 1mm; CWD. Maximum length of wet spell, maximum number of consecutive days with RR ≥ 1mm. Finally, the trends of indices were estimated using the non-parametric Mann-Kendall test and the values of these indicators were estimated for 50, 200 and 500 years return periods.In order to calculate values of each indicator for a given return period, the annual time series and its probability of occurrence are estimated and the most appropriate statistical distribution function that can be fitted on the data is selected from among twelve functions. In this estimation, EASY-FIT (a hydrology software), which supports a higher range of distribution functions, is used. The intended significance level for 500, 200 and 50 years return periods were 0.998, 0.995 and 0.98, respectively. The functions used in this research include: Lognormal (3P), Lognormal, Normal, Log-Pearson 3, Gamma (3P), Gumbel, Pearson 5 (3P), Log-Gamma, Inv. Gaussian, Pearson 6 (4P), Pearson 6, Gamma. Kolmogorov–Smirnov test is used to assess the goodness of fit of the estimation from three return periods.

The results indicate that while the trend of precipitation indices except for the Maximum length of dry spell (CDD) is decreasing, the trend of temperature indices was increasing, except for two indices of the days with daily maximum and minimum temperatures below zero degrees. From a spatial perspective, hot indices in the northwestern regions, cold indices in the southern half of the country shows an increasing trend, and the Caspian Sea, Oman Sea, Persian Gulf coastal regions, and the Zagros foothills are the most affected areas as a result of the increasing trends. Also, the index values were estimated for 50, 200 and 500 years return periods. As a result of the investigations, for temperature indices the north-west of the country has the highest values by different return periods. The increase in the values of R10, R20, RX1day and RX5day indices in the different return periods was more in the Zagros and Alborz mountain ranges, and the CWD, CDD and SDII indices have the highest values in the Caspian Sea and Persian Gulf Coastal areas.