عنوان مقاله [English]
Iran is located in an arid region, so planning to adaptably cope with drought consequences in this arid climate regions is essential and a constitutes important water-resources/water-demands managerial program at country level. In order to compile a drought plan, we need to carefully study the historic rainfall period using a appropriate drought indices. Drought is a climatic phenomenon that is statistically random and is studied by several indicators. One of these indicators that has triggered a major paradigm change is the one proposed by McKee et al. (1993) known as Standardized Precipitation Index (SPI). The index was developed to define and monitor drought and determine rainfall intensity anomalies in time scales of 3 to 48 months in Colorado. In the calculation of SPI, the total precipitation of a cumulation period consisting of several consecutive months are compared with its counterpart during the studied years. Based on the appropriate statistical distribution, the probability of similar precipitation is calculated. The index number is obtained after normalizing the corresponding cumulative probability distribution function value using standard normal distribution. In this method, the rainfall of different months is included in the calculation without considering the distance or proximity to the present time and using a uniform weight. It has been observed that extreme monthly rainfall events, several times heavier than long-term a verge keeps the index value high for several consecutive months. The dry periods that follow the such a wet month are hidden from the SPI, and this index mistakenly show that period as wet, even though we know that these precipitations, especially in arid areas, are mostly in the form of showers and are quickly out of reach and will not have much effect on the wet season.
Materials and methods
In this study, SPI drought index in12 provinces with26 synoptic stations located in the center, south and east of Iran with a dry climate (determined by Domarten method) has been calculated and analyzed over a period of 34 years (1364-1397) for a cumulation period of twelve months. The most abundant and the least precipitated stations were Safiabad-Dezful and Nehbandan, respectively. To calculate the SPI index, gamma distribution was used to fit long-term precipitation data. Since the SPI index does not take into account the effect of precipitation timing by definition, it is likely that in the period under review to estimate unreal wet period. Therefore, in this study, for the first time, the a new EP-SPI correction index is presented, in which the effective precipitation data obtained by damping the effect of precipitation with the time factor, as used as input to SPI Generator software, developed in2018 by The National Drought Mitigation Center was used. In this way, the effect of precipitation in the first month close to the event is 12 times, in the second month 11 times and also in the previous months compared to 12 and the farthest month. As far as the twelfth month is concerned, in practice it has not had much effect on the current drought, and only once it has been included in the calculation in the ratio of 1/12. Thus, the effect of older rainfall is reduced and the index provides a better estimate of drought.
Results and discussion
Since the two methods used have the same logic, the regression of EP-SPI versus SPI values shows a significant correlation at the5% level with a coefficient of determination of 0/673 and a slope of 0/82. To compare the two methods SPI and EP-SPI, we need to perform statistical tests. If the data has a normal distribution, it is possible to use a parametric test, otherwise we must use a non-parametric test. Using the Kolmogorov-Smirnov test, we determine the normality of the data distribution. According to the result of this test with error α=0.05, the distribution of normal SPI and EP-SPI values and the difference between the two methods were not normal. In reviewing the studied stations, out of 9895 events, the cases in which SPI was positive and EP_SPI negative included 880 cases, of which 277 were differences in the values of SPI and EP-SPI index more than 1. In 67 cases, the degree of drought was 2 to 4 levels of difference with SPI, which indicates the presence of drought in the study areas. Three acute incidents occurred in Qom, Mehrabad and Omidiyeh stations in January and February 1996 and December 1998. In these three stations at least 4 levels of drought have been shifted. According to the study, the lowest number of disputes was related to Safiabad station (Dezful) in Khuzestan province with 6 incidents and the highest number was related to Omidieh (Aghajari) stations in Khuzestan province and Minab in Hormozgan province with 18 incidents. it is arrived 86% of the incidents were seen with a difference of more than one in the months of November to February and the highest difference between the two methods was equivalent to four levels of drought index shift from very severe to mild drought. Also, in terms of average, the difference between the two methods was more severe in December and January.
The monthly pattern of more than one difference between SPI and EP_SPI twelve months in the rainy months of December to February and low rainfall in June to October was in accordance with the rainfall regime in the region. In general, the difference in precipitation from year to year, more at the beginning or end of the rainy season, greatly affects the SPI values, while in the EP-SPI method, their effect dampens over time and a better understanding of the real state of drought/wet period has provided in the area. In this study, it was found that8.9% of the twelve-month drought events in dry climate were hidden from the SPI index. This was consistent with a comparison of NDVI maps that were typically received from NOAA satellite AVHRR sensors on5 dates. In these 5 dates, the difference between SPI and EP-SPI index has been observed in more than ten stations. Therefore, it can be said that the efficiency of EP-SPI method for detecting droughts in a period of twelve months is confirmed. Although no similar study has been found to date, Nadi and Sheikhi Soghanloo (2020) observed that the difference between SPImod and SPI indices is more obvious in stations with drier climates. Therefore, they concluded that using SPImod instead of SPI provides more accurate results due to the elimination of seasonal rainfall effects. It can be concluded that the results of this study were consistent with the result of this study.
Keywords: Correction Index EP-SPI, arid zone, Drought, SPI.