عنوان مقاله [English]
The North Atlantic Oscillation (NAO) is one of the large climate signals that affect partly climate variabilities in northern hemisphere. Various studies have considered covariation between precipitation in Iran with NAO mainly at monthly timescale. No research has been conducted on a weekly time scale. In this study, the covariate between ERA-Interime gridded precipitation data over Iran and NAO index was investigated. The challenge of this study is implementing correlation alalyses in arid and semi-arid areas of the country, where the frequency of zero precipitation data across the year is high, at weekly timescale.
Materials and methods
In the study was used the daily data of precipitation amount obtained from reanalysis ERA-Interim (ECMWF) and the daily data NAO indices obtained from the url http://www.cgd.ucar, both of them related to the cold months of the period of 1979-2016 (November (N), …, April (A); or NDJFMA) were considered. The gridpoints cover 44°E - 64°E and 25° N - 40° N (about geographical area of Iran) with 0.5°* 0.5° spatial resolution. At timescales less than one month, a large number of zero values may appear in precipitation time series, leading bias estimate in correlation coefficient between precipitation and NAO. The present study calculated moving average of the data for smoothing time series and removing the undesirable effect of enormous zeroes in the precipitation time series. Also, Modified Pearson correlation coefficient,, was used to unbias the estimation of traditional Pearson correlation coefficient. centers nonzero data around their average and works with new time series. Each year, the relationship between precipitation and NAO was calculated in terms of ; therefore, a set of 37 Modified Pearson correlation coefficients corresponding to 37 years of the record period was obtained. Besides, in each year, was calculated for the lag times of 1-45 weaks. Among the simultaneous and lagged , the significant coefficient that its absolute value was greatest was selected for each year. This allows identifying gridpoints with the significant impact of NAO on precipitation. The minimum, maximum, median and standard deviation of selected values at each gridpoint were estimated and showed with ,, and notations, respectively.
Results and Discussion
The results of this study are summarized as follows:
s (modified Pearson correlation coefficient) between precipitation and NAO had wide range from negative to positive values across the study area.
Based on the probability of precipitation occurring in years it seems that this method can better quantify the effect of the NAO negative phase, NAO-, than NAO+ on year-to-year variability of precipitation.
The mapping showed that most regions of Iran have the positive correlation with the NAO index. Also, most parts of Saudi Arabia (east and central parts), the highlands of Afghanistan, the Persian Gulf and Afghanistan have either a positive relationship with the NAO or are more or less affected by the positive phase. But this correlation is negative in large areas to the west and southwest, in small parts in the east, along the Afghanistan-Turkmenistan border, in the south-east of Iran (the Oman Sea coast), the western regions of the Caspian Sea, northern Turkey and Azerbaijan. Given that the areas with negative correlation coefficient (), have bigger precipitation event probability in the NAO-, it seems that they are affected by NAO- more than other areas. Activity of low pressure in the eastern Mediterranean increases during NAO-; therefore, the precipitation over Iran increases. But areas with positive correlation coefficients do not necessarily receive more precipitation in NAO+.
The largest values of are in the northwest, east, parallel to the southwest-northeast and the southeast of Alborz Mountains. This index showed that there is more variance of coefficients in these areas.
This paper considered the relationship between precipitation and NAO index at weekly timescale. The results showed that the precipitation occurrence in Iran coincides with the negative phase of NAO. The proposed approach for bias correction of Pearson correlation coefficient acted successfully in investigating the effect of NAO on the cold months precipitation over Iran.