Abstract

This paper quantifies the impact of the El Niño–Southern Oscillation (ENSO) on the intensity and occurrence probability of dry and wet periods in Iran during boreal autumn and winter. Three phases (warm, cold, and neutral) were defined based on the Southern Oscillation (SO) status, and precipitation composites were constructed for each phase. The 30th and 70th percentiles of neutral phases were used as the thresholds for distinguishing normal conditions from dry and wet anomalies, respectively. The shifts in the amount and occurrence probability of these thresholds associated with warm and cold ENSO phases were then quantified. It has been found that, compared to the neutral period, warm events substantially reduce (increase) the intensity and occurrence probability of autumnal drought (wet) periods, particularly for southern districts. On the other hand, when a vigorous La Niña prevails, the chance of wet (dry) conditions is low (high) and the probability of severe autumnal drought is intensified. During winters of warm ENSO phases, although most of the country receives precipitation above the drought threshold, in the southeastern and northwestern districts of Iran, the risk of winter drought is high. For these phases, there is little chance that precipitation in these areas is above the wet threshold. A mechanism is proposed to justify the seesaw fluctuation of winter precipitation over the southwestern and southeastern Caspian Sea coasts. It is likely that the interaction between the Siberian high and ENSO controls rainfall variability over these coasts. It was found that during cold ENSO phases, winter drought (wet) periods in southern Iran are mostly coincident with wet (dry) conditions over the tropical Bengal Gulf (TBG) region. Such a strong coincidence was not found when rainfall in southern Iran and the Indian Ocean Extension region was compared. For western and northwestern parts of Iran, the probability and intensity of winter drought was found to be low during La Niña events.

1. Introduction

The Islamic Republic of Iran (Fig. 1) has an area of 1 648 000 km2 and a population of about 66 million people (2002 estimate). The average annual precipitation over the country is estimated to be 250 mm, less than one-third of the global average. Iran is mostly categorized as having arid or semiarid climates. Eastern and southern halves of the country are generally drier than the western and northern halves.

Fig. 1.

The relief map of Iran and geographical locations of the rainfall stations. The station names associated with the presented numbers and the record lengths (in parentheses) of precipitation time series (ended by 1997) are shown below

Fig. 1.

The relief map of Iran and geographical locations of the rainfall stations. The station names associated with the presented numbers and the record lengths (in parentheses) of precipitation time series (ended by 1997) are shown below

The country lies within the western Alpine–Himalayan chains with two major mountain systems, the Alborz and Zagros Ranges (Fig. 1). These two ranges play an influential role in determining the amount and distribution of rainfall over the country. While the total annual precipitation over the highlands of these two ranges and the southwestern Caspian Sea coasts exceeds 1800 mm, there is less than 60 mm over central and southeastern parts of Iran. The occurrence of drought and flash floods are common, and the nation has frequently suffered from these devastating events. Due to the significant impact of these hazards on human life, agriculture, livestock, forestry, water resources, tourism, construction, industry, and many other local activities, long-range prediction of precipitation is important for the mitigation of these hazards.

The El Niño–Southern Oscillation (ENSO) phenomenon contributes to the predictability of rainfall anomalies in many tropical and subtropical regions, including Australia, North America, South America, India, and some parts of Africa (e.g., Allan et al. 1995, 1996; Drosdowsky 1995; Ropelewski and Halpert 1996; Nazemosadat and Cordery 1997; Lough 1997; Richard et al. 2000; Cook 2001; Curtis et al. 2001; Kane 1999, 2002). Nazemosadat and Cordery (2000a,b), Nazemosadat (2001), and Ghasemi (2003) have also demonstrated that ENSO affects Iranian autumn and winter rainfall, which supply about 80% of Iranian water resources. In autumn, strong relationships were found between the Southern Oscillation index (SOI) and rainfall over the northwestern provinces. For the other regions, the impact of ENSO on autumn precipitation is substantial during intense warm and cold episodes. The measure of this impact, however, has not been comprehensively studied. In contrast to autumn, the influence of ENSO on precipitation is less during wintertime and varies from place to place (Nazemosadat 2001). Although, the above-normal (below normal) winter rainfall tends to be coincident with cold (warm) ENSO phases, it is not a general rule, and further research is needed to quantify and justify the SOI–rainfall relationships during warm and cold ENSO phases. Furthermore, the impact of ENSO on shifting the intensity and probability of seasonal drought and wet periods has not yet been comprehensively investigated. Barlow et al. (2002) have reported an inverse association between the anomalies of November–April precipitation over two distinct regions, southwest Asia and the Indian Ocean precipitation extension (IPX) area. The present study further evaluates the influence of ENSO and IPX rainfall on the occurrence of drought and excess rainfall in southern Iran.

Ropelewski and Halpert (1996) studied strong, consistent ENSO–precipitation relationships for 19 regions of the globe using the percentile analysis method. They have also quantified these relationships based on shifts in the statistical distribution of precipitation amounts with emphasis on shifts in the median, during warm and cold phases of the Southern Oscillation (SO). Their study, however, did not include precipitation analyses in Iran, where the influence of ENSO is comparable with eastern Australia (Nazemosadat et al. 2003).

For the consolidation of the previous findings in Iran (SO–rainfall relationships) and documentation of instabilities in the relationships, the percentile analysis method (Ropelewski and Halpert 1996) with some modifications was adopted for the present study. The thresholds of wet and dry periods were defined, and the measure of the influence of ENSO extremes on the intensity and occurrence probability of these thresholds is addressed quantitatively. The possible interaction between the Siberian high and ENSO on the variation of rainfall over the southern coasts of the Caspian Sea is also presented. The alternation of dry and wet years between the IPX region and southern parts of Iran was also studied, and a new index that justifies rainfall variability over southern Iran is proposed. Since the impact of ENSO on the precipitation in Iran is partially inverted as the season changes from autumn to winter, analyses were performed separately for autumn and winter seasons.

2. Data and methods

The basic data used in this study were total monthly precipitation from 51 synoptic stations during the period 1951–97 (Fig. 1). In addition to precipitation, the time series of monthly SOI and IPX data were also used. The SOI data (1951–97) were extracted from the Web site of the Australian Bureau of Meteorology, and the IPX data (1979–97) were provided by Matt Barlow at the International Research Institute (IRI) for Climate Prediction, Columbia University, New York, New York.

Seasonal averages were computed from monthly data. Here, autumn is defined as October–December and winter as January–March. For each season, Nazemosadat and Cordery (2000a,b) defined the warm (cold) events of the SOI as those years during which the seasonal SOI remains in the upper (lower) 25% of all observed values (Table 1). This suggests that El Niño (La Niña) corresponds to the periods when the seasonal SOI is lower than −4.8 in autumn and −5.2 in winter (4.5 in autumn and 4.2 in winter; Table 1). Our preliminary investigation has shown that this methodology for distinguishing ENSO phases leads to more significant results than the list of El Niño and La Niña phases presented by Trenberth (1997). For warm, cold, and neutral phases, the ENSO precipitation composites were constructed for autumn and winter. For each station, the amounts of the 10th, 30th, 50th (median), 70th, and 90th percentiles of the composites were then computed.

Table 1.

The list of El Niño and La Niña phases for autumn and winter. The seasonal SOI data of each year are denoted in parentheses

The list of El Niño and La Niña phases for autumn and winter. The seasonal SOI data of each year are denoted in parentheses
The list of El Niño and La Niña phases for autumn and winter. The seasonal SOI data of each year are denoted in parentheses

We considered the 30th and 70th percentile levels of the neutral phases as the thresholds for distinguishing normal conditions from dry and wet anomalies, respectively. For each rainfall composite, the ratio of the 30th percentile during warm and cold ENSO phases (R30w and R30c) to the corresponding percentile during neutral phases (R30n) were then considered as a basis for the measure of the impact of ENSO on drought intensity. For instance, when the ratio of R30w/R30n is greater (less) than unity, it suggests that, during El Niño phases, the intensity of drought is less (more) disastrous than neutral phases. Similarly, the ratios of the 70th percentile of precipitation during extreme ENSO phases to the corresponding values during neutral phases [(R70w/ R70n) and (R70c/R70n)] were used for the assessment of ENSO extremes on the intensity of wet anomalies.

The SO-related shifts in the occurrence probability of the thresholds of drought and wet periods were also estimated by comparing the values of R30 and R70 during neutral phases with almost similar amounts of precipitation percentiles during El Niño and La Niña phases. For instance, if the amount of R30n of a particular station was equal to the 10th percentile value during warm ENSO phases, the probability that during warm phases rainfall is below R30n is 10%. This means that, for this station, warm ENSO events shift the probability of drought from 30% to only 10%. In the same way, if R70n was equal to R50c, the probability that during cold ENSO phases, rainfall is higher than the wet threshold (R70n) is 50% (instead of 30%). This methodology permits the estimation of the shift in occurrence probabilities of dry and wet thresholds during ENSO extreme phases. Parametric (two-tailed t test) and nonparametric (Mann–Whitney test) tests are used to investigate whether the mean and medians of the SO– precipitation composites during El Niño phases are statistically different from the composites of La Niña phases.

3. Results and discussion

a. Statistical tests

The results of both parametric and nonparametric tests have shown that, for autumn, the mean (and also median) of the SO–precipitation composites during warm ENSO phases are statistically (1% significance level) different from the composites mean (and also median) during cold phases for almost all stations. On the other hand, for winter, the difference between the mean (and also median) of the SO composites was generally found to be statistically insignificant for various ENSO phases, indicating less influence of ENSO during this season. In spite of this fact, as is shown later, the occurrence of ENSO does induce a considerable change in intensity and occurrence probability of dry and wet anomalies in winter precipitation. Furthermore, the interaction between ENSO and other factors including the Siberian high, and precipitation over both the IPX and tropical Bengal Gulf (TBG) regions induce a significant impact on winter precipitation in Iran that is discussed in the next sections.

b. Autumn analysis

1) Warm episode

The spatial distribution of the 30th and 70th percentile levels of precipitation during neutral phases (the drought and wet period thresholds, respectively) is depicted in Figs. 2a–d for both seasons. As indicated, the high and low precipitation levels are centered over the western and southeastern parts of the country, respectively. Moreover, with the exception of the Caspian Sea coasts, winter rainfall is consistently greater than autumn rainfall for the other parts. The measure of the shift in the magnitudes and occurrence probability of the depicted values associated with cold or warm ENSO phases was then investigated.

Fig. 2.

The spatial distribution of the dry and wet thresholds (mm season−1) for (a), (b) autumn and (c), (d) winter. For the stations over the Caspian Sea coasts, numbers show the threshold values. The probability that the precipitation is below the values shown in (a) and (c) and above the values shown in (b) and (d) is 30%

Fig. 2.

The spatial distribution of the dry and wet thresholds (mm season−1) for (a), (b) autumn and (c), (d) winter. For the stations over the Caspian Sea coasts, numbers show the threshold values. The probability that the precipitation is below the values shown in (a) and (c) and above the values shown in (b) and (d) is 30%

As indicated in Fig. 3a, for most parts of the country, the ratios of R30w/R30n are greater than unity and vary from about 1 to 3. This means that, compared to the neutral periods, the intensity of drought events is generally much smaller for warm phases. Furthermore, during such phases, autumn precipitation is substantially above the drought threshold, particularly for the southern districts. Although the SOI–rainfall correlations are generally weak for these districts (Nazemosadat and Cordery 2000), the impact of the warm ENSO phase on both drought avoidance and the total volume of the precipitated water is essential for the southern regions. Prediction of rainfall amount is, however, more reliable for the northwestern areas, where the SOI–rainfall relationships are strongly significant. The ratio is less than unity for the southwestern parts of Caspian Sea shores, indicating greater likelihood of drought conditions for this area during El Niño events.

Fig. 3.

The spatial distribution of autumnal (a) R30w/R30n and (b) R70w/R70n

Fig. 3.

The spatial distribution of autumnal (a) R30w/R30n and (b) R70w/R70n

As indicated in Table 2a for all stations in the northwestern region, the 30th percentile precipitation value during neutral phases is very close to the 10th percentile amount during warm phases. Therefore, during El Niño phases, the probability that this region encounters deficient (below the 30th percentile of neutral phases) rainfall is only 10%. Although in Tables 2a,b the percentile values are shown for some specific stations (that are referred to in the text for supporting discussion), similar computations were carried out for all the stations and are presented in spatial maps.

Table 2.

The rainfall values of percentiles for SO–precipitation composites (mm season−1 ) during warm, neutral, and cold phases for (a) autumn and (b) winter for stations that are discussed in the text

The rainfall values of percentiles for SO–precipitation composites (mm season−1 ) during warm, neutral, and cold phases for (a) autumn and (b) winter for stations that are discussed in the text
The rainfall values of percentiles for SO–precipitation composites (mm season−1 ) during warm, neutral, and cold phases for (a) autumn and (b) winter for stations that are discussed in the text

Figure 4a indicates that the occurrence probability of a drought event is less than 15% for most parts of Iran, suggesting that, during warm ENSO phases, it is unlikely that rainfall is below the drought threshold. It is worthwhile to note that, in the presented probability maps, the measure of the deviation of contour lines from 30% shows the degree of the influence of ENSO extremes on shifting the neutral phases probabilities. The highest probability for drought (about 40%) was observed on the southwestern corner of the Caspian Sea coastal strip indicating a maximum risk of drought for this part of Iran during El Niño events.

Fig. 4.

The spatial distribution of the probability that during warm ENSO phases, autumn precipitation is (a) below the drought threshold and (b) above the wet threshold

Fig. 4.

The spatial distribution of the probability that during warm ENSO phases, autumn precipitation is (a) below the drought threshold and (b) above the wet threshold

The unique geographical characteristics of the Caspian Sea narrow coastal strip (bounded by the Caspian Sea waters in the north and the Alborz Ranges in the south) facilitates the lifting of the oversea moisture-laden easterly winds over the Alborz highlands for producing abundant rainfall. In contrast to the width of the strip, rainfall amounts show a decreasing trend from the west to the east resulting from the wind direction. It is more likely that during late autumn and winter, the strength of the Siberian high (which is situated near the northern parts of the Caspian Sea) affects the intensity of these easterly winds (heading to the southwestern coasts, near Bandar-Anzali). Zhang et al. (1996) reported that, during El Niño phases, the surface pressure of the Siberian high and its associated 500-hPa trough and 200-hPa jet stream are weaker than normal. They emphasized that the interannual variation of this high is in general agreement with the SOI fluctuation. These findings imply that the rainfall of the Caspian Sea coasts is highly influenced by the characteristics of the Siberian high, ENSO phases, and the sea surface temperature (SST; Nazemosadat and Ghasemi 2002). As is shown later, for winter, when the Siberian high is at its highest strength and ENSO is less influential in Iran, the aforementioned mechanism is consistent with these results regarding the fluctuation of rainfall over the southwestern Caspian Sea coasts. This means that for El Niño and La Niña phases, rainfall in this region is mostly below and above values of neutral phases, respectively. However, for autumn, when the Siberian high is not yet completely developed and ENSO is highly influential on rainfall in Iran, the shifts in precipitation percentiles during ENSO extreme phases are in general agreement with other parts of the country. In contrast to other parts of Iran, the autumn ratio of R30w/R30n is less than unity for the southwestern Caspian Sea. However, as is shown later, the other ratios associated with this region (for autumn) are generally consistent with the other parts of Iran.

The ratios of R70w/R70n (Fig. 3b) are also greater than unity for most parts of the country, with the highest value (2.5) observed in Abadeh. Therefore, most parts of Iran tend to experience wet conditions in autumn associated with warm ENSO phases. The ratio was found to be near unity in some stations, including Zahedan, Zabol, Noushahr, Astara, Khoy, Sanandaj, Zanjan, and Gorgan, for which the occurrence of very wet conditions appears to be hampered by strong El Niño events.

Table 2a implies that the 70th percentile of neutral phases is marginally greater than the 30th percentile of Tabriz rainfall during warm episodes. Therefore, Tabriz has about a 65% probability for receiving the wet threshold or greater precipitation during the strong negative SO phases. As indicated in Fig. 4b, for most parts of the country, the probability that during warm ENSO phases rainfall is above the wet threshold is generally greater than 60%. This means that, during warm ENSO phases, seed germination and vegetative growth of wheat and other planted cereals are at a low risk. For these phases, more cloudiness and rainy days are expected and the occurrence of flash floods is more likely. These results indicate the El Niño events are generally coincident with more-than-usual rainfall over most parts of Iran, and severe autumn drought is hence not likely for such periods.

2) Cold episode

The ratio of R30c/R30n (Fig. 5a) is less than unity over most parts of Iran, with a minimum in Abadeh, indicating that severe autumn drought is expected during cold ENSO phases. Exceptions are found for Biabanak, Birjand, Abadan, Fasa, Bandare-Abbas, Dezful, and Gorgan, for which the ratios are equal to or slightly greater than unity. Severe drought is less likely for these stations during La Niña phases.

Fig. 5.

The spatial distribution of autumnal (a) R30c/R30n and (b) R70c/R70n

Fig. 5.

The spatial distribution of autumnal (a) R30c/R30n and (b) R70c/R70n

The 30th percentile for Tabriz precipitation during neutral phases is between the 30th and 50th percentiles of cold phases (Table 2a). As a result, in association with cold phases, this station has about a 40% probability that rainfall is equal or less than the drought threshold. As indicated in Fig. 6a, during intense La Niña phases, the probability of drought in Iran varies from 40% to 65%. The probability is generally high for arid regions of southeastern Iran, indicating a higher risk of severe drought for these areas.

Fig. 6.

Same as in Fig. 4, but for cold ENSO phases

Fig. 6.

Same as in Fig. 4, but for cold ENSO phases

The ratio of R70c/R70n is generally equal to or less than unity for most parts of Iran (more than 90% of the stations) and greater than unity in a few stations including Biabanak, Kashan, Yazd, Birjand, Mashhad, and Broujerd (Fig. 5b). This result reaffirms that, during cold ENSO events, the shortage of precipitation prevails within the country. For instance, in the northwestern parts of Iran including Tabriz, Oromieh, and Khoy, the 70th percentile during neutral phases is approximately equal to the 90th percentile during cold phases, suggesting that these stations have only a 10% probability of receiving the 70th neutral year percentile (Fig. 6b). As is shown, this probability is below 25% for most parts of the country, indicating a small chance that autumn rainfall exceeds the considered threshold of the wet period during La Niña phases. Decision makers and other community sectors should consider drought mitigation and preparedness policies during such phases.

c. Winter analysis

1) Warm episode

The ratio of R30w/R30n was generally found to be equal to or marginally greater than unity, with maximum and minimum values of about 1.6 and 0.6 in Iranshahr and Khoy, respectively (Fig. 7a). The given results imply that, during El Niño events, a severe decline in winter precipitation is not expected for most of the stations. The ratio was less than unity for eastern regions (Kerman, Sarakhs, Biabanak, Bam, Birjand, Khash, Zahedan, and Zabol), the northwest extremity of Iran (Khoy), and Bandar-Anzali in the southwest of the Caspian Sea coasts. For these stations, decline in winter precipitation and meteorological drought are anticipated as a result of warm ENSO periods. Strikingly, the minimum value of the ratio is found in Khoy, where the ratio is greater than or equal to unity for the nearby stations. According to Table 2b, winter rainfall in Khoy is below the drought threshold (30th percentile of neutral phases) during warm spells, with about a 55% probability. This probability is, however, decreased to about 20% and 30% in Oromieh and Tabriz, situated in surrounding areas, respectively (Fig. 8a). In order to understand this finding, we have performed percentile analysis for two rainfall stations [Kars, Turkey (40°37′N, 43°06′E) and Dir, Turkey (39°55′N, 44°03′E)] near Khoy in Turkey. The results indicate that during warm ENSO phases, for 46% and 54% probabilities, the winter rainfall of these two stations is also lower than their drought thresholds, respectively. The consistency of the given results for Khoy and two nearby stations in Turkey reaffirm that, for the northwestern extremity of Iran, El Niño increases the risk of severe winter drought. It should be noted that the general elevation of the northwestern extremity of Iran shows a sharp decreasing trend toward the north so that it changes from about 1360 m above mean sea level at Oromieh to about 600 m near the northern extremity of the Iran– Turkey borderline. This change in elevation causes a considerable variation in the climate characteristics of this area, including higher temperatures and less precipitation.

Fig. 7.

Same as in Fig. 3, but for winter

Fig. 7.

Same as in Fig. 3, but for winter

Fig. 8.

The spatial distribution of the probability that during warm ENSO phases, winter precipitation is (a) below the drought threshold and (b) above the wet threshold

Fig. 8.

The spatial distribution of the probability that during warm ENSO phases, winter precipitation is (a) below the drought threshold and (b) above the wet threshold

Comparing Figs. 7a and 8a suggests that drought is highly probable for the regions for which the ratios of R30w/R30n are smaller than unity. On the other hand, it is unlikely that an area for which the ratio is well above unity (such as Iranshahr, with a ratio of 1.6) faces serious shortage of winter precipitation during El Niño phases. Due to the scarcity of rainfall stations with long records of reliable data in the southeastern region, it is not easy to understand why Iranshahr precipitation is highly sensitive to El Niño while such sensitivity was not observed for the other stations (such as Zahedan and Zabol) in this region. Since the pressure system over the northwestern Indian Ocean is generally below normal during warm ENSO phases (Nazemosadat 1996), it is more likely that the dominance of such a low supplies moisture for the relatively highland areas near Iranshahr.

Although the ratio of R30w/R30n is generally equal or greater than unity for the Caspian Sea coastal strip, it drops to 0.8 for Bandar-Anzali (to the southwest), indicating an intense winter drought during warm ENSO phases. Compared to other parts of Iran, the probability that rainfall is less than the drought threshold is also high and reaches about 50% (Fig. 8a). As indicated earlier, during warm ENSO phases, the Siberian high and its associated oversea easterly wind is weaker than usual, enhancing the intensity of drought over southwestern coasts.

In contrast to the southwest, the ratio was found to be around 1.2 for Gorgan and Gonbadeghabus in the southeastern Caspian Sea shores, the areas that are less affected by the easterly winds. The 30th percentile of these stations during neutral phases is near the 10th percentile during warm phases (Table 2b). This suggests that, during an El Niño event, the probability that these stations receive below–drought threshold precipitation is only 10% (Fig. 8a). During such events, the north– south pressure gradient over the Caspian Sea becomes weaker than usual, and less deflection (to the west due to Coriolis effect) of the prevailed wind is, therefore, expected. As a result, due to the weakening of the Siberian high and its associated easterly winds, southeastern (southwestern) Caspian Sea coasts receive more-than-usual (less than usual) rainfall during El Niño events.

Figure 7b shows that the ratios of R70w/R70n tend to be equal to or marginally less than unity for most parts of the country. Exceptions are found for a few stations including Gorgan, Nishabour, Borujred, Kashan, and Arak. In addition to the mechanism of such exception that is already presented for Gorgan, local synoptic situations and sporadic showers over these regions could be the cause of such exemption. Comparing Figs. 7a and 7b suggests that, although warm ENSO phases prevent severe shortages of precipitation over most of the country, well-above-normal rainfall is also not anticipated during such a phase. The probability that rainfall exceeds the 70th percentile of neutral phases varies from 20% to 30% with maximum values (near 50%) for the region on the southeastern Caspian Sea coasts (Fig. 8b). In contrast to the southeastern Caspian Sea coasts, for southeastern Iran, the probability of above-normal precipitation (drought) is very low (high) during warm ENSO phases. Therefore, this arid part of the country could face a serious shortage of winter rainfall for such phases.

2) Cold episode

In Fig. 9a, the spatial distribution of the R30c/R30n ratio is plotted. The ratios are smaller than unity for about 30% of stations, mostly located over the coastal area of the Persian Gulf and Oman Sea and the northeastern regions. This implies that rainfall deficits are likely for these areas during years when the SOI is strongly positive. As indicated in Fig. 10a, the probability that winter rainfall in these areas is below the drought threshold is generally higher than in the other parts of Iran. For such periods, decision makers, farmers, environmentalists, and other community sectors living in these areas should consider drought mitigation policies. On the other hand, for the stations situated over western and northwestern districts as well as the Caspian Sea coastal areas, it is unlikely (only a 15% to 25% probability) that during cold ENSO phases, winter precipitation is below the drought threshold (Fig. 10a).

Fig. 9.

Same as in Fig. 5, but for winter

Fig. 9.

Same as in Fig. 5, but for winter

Fig. 10.

Same as in Fig. 8, but for cold ENSO phases

Fig. 10.

Same as in Fig. 8, but for cold ENSO phases

The ratio of R70c/R70n is equal to or greater than unity for all considered stations with the exception of the northeastern region as well as Chabahar in the southeastern corner of the country (Fig. 9b). This implies that, in association with cold phases, winter rainfall is marginally shifted toward above normal over most parts of the country. The maximum (minimum) value of this ratio is 2 (0.6) for Kashan and Ghom (Chabahar). As suggested by Table 2b, for Kashan and Chabahar, the amounts of R70n are near R30c and R90c, respectively. This means that during cold events, these stations have about a 70% and 10% probability of receiving the wet threshold or greater precipitation, respectively. The probability that during the cold ENSO phase, winter precipitation in Iran is above the 70th percentile of neutral phases is generally high and greater than 40% (Fig. 10b). This means that, in winter, due to La Niña events, the occurrence probability of above-normal rainfall in most parts of Iran increases by more than 10%.

For southwestern parts of the Caspian Sea shores, the 30th and 70th percentile levels of precipitation during La Niña phases is greater than corresponding values for neutral phases (Figs. 9a,b). For such phases, the probability that this region faces drought is low, and well-above-normal rainfall is anticipated. The given results reaffirm the earlier discussion indicating that, in winter, the prevalence of cold (warm) ENSO phases is generally coincident with the strengthening (diminishing) of the Siberian high and its associated oversea easterly winds. As much as these winds are stronger (weaker), more precipitation is expected over the southwestern (southeastern) Caspian Sea coasts.

The characteristics of the contour lines in Figs. 9a,b illustrate the complicated role of cold ENSO phases in the precipitation of the southeastern corner of the country. While cold ENSO phases are associated with shortages of rainfall in Chabahar (on the Oman Sea coast), they favor wetter-than-usual conditions over the interior parts of this area. This implies that the source regions of rainfall are different for coastal and interior parts of the southeastern district. According to these results and the authors' experience, winter precipitation in Chabahar is mainly supplied from the nearby waters in the Indian Ocean that is generally warmer and colder than usual during El Niño and La Niña phases. Therefore, it is not surprising if the increase and decrease of winter rainfall in Chabahar is associated with the variation of ENSO phases. However, the interior region mostly benefits from the southwest–northeast cloud bands passing through the Persian Gulf and propagating to the east. It seems that the rainfall of this region is mainly influenced by the interaction of ENSO and rainfall in the IPX region rather than the northwestern Indian Ocean SSTs.

3) Cold ENSO phases and rainfall in southern Iran and the TBG region

Comparing Figs. 9a and 9b suggests that the occurrences of cold ENSO phases have a dual character over the Persian Gulf coastal region. Although it can cause a serious shortage of rainfall during some years, abundant rainfall in these regions has also occurred during La Niña events. For instance, out of 12 cold ENSO phases during the studied period (Table 1), a substantial shortage of rainfall was evident for the winters of 1955, 1962, 1963, 1967, 1971, and 1989. For the other six years, winter rainfall was found to be above normal. This dual character of the rainfall suggests that the impact of cold ENSO events on the rainfall of this region could be modulated by other factors.

Barlow et al. (2002) have recently used data after 1979 and shown that the dry (wet) periods in southwest Asia are generally coincident with positive (negative) anomalies of rainfall over the IPX region. The IPX index is simply the average value of November–April precipitation over the region bounded between 15°N–10°S and 90°–100°E. They have concluded that due to the enhanced convective activity and heating over the IPX region, waves are generated in the upper atmosphere over southwest Asia (to the west of the heating); these waves interact with the regional jet stream and lead to drought over this region. In order to examine this mechanism further, the simultaneous influence of both IPX and La Niña phenomenon on the winter rainfall of southern Iran was investigated. Such an examination was also conducted for El Niño phases, but the results were observed to be insignificant. The correlation analysis between IPX data and the winter rainfall in Iran (1979– 97) was also carried out, but the results were found to be insignificant.

Since most of the La Niña events accompanying drought in southern Iran had occurred before 1979, the IPX data used by Barlow et al. (2002) needs to be extended for the period before this year. The composite maps of the precipitation anomaly [provided by the National Oceanic and Atmospheric Administration/Climate Prediction Center (NOAA/CDC) Web site] were, therefore, applied as the proxy of IPX data. For deriving these maps, the La Niña phases are first categorized into the years with and without drought in southern Iran. For each category, the precipitation anomaly map was then derived for both the IPX region and southern Iran (Fig. 11). Similar figures were also provided (but not shown) for each of 12 La Niña phases to compare rainfall variability in these two distinct regions more precisely. As indicated (Figs. 11a,b), for wet years in southern Iran (1951, 1956, 1972, 1974, 1976, and 1996), precipitation is below normal for the proposed IPX region. However, such an inverse relationship was not found for dry years (1955, 1962, 1963, 1967, 1971, and 1989) in Iran (Figs. 11c,d). This means that drought in the southern parts of Iran is not meaningfully associated with well-above-normal precipitation over the IPX region and its associated heating.

Fig. 11.

The composite maps of the winter precipitation anomaly in (a), (c) southern Iran and (b), (d) the IPX region during La Niña phases. Panels (a) and (b) are related to the years with well-above-normal rainfall in Iran; (c) and (d) are related to the years with below-normal rainfall

Fig. 11.

The composite maps of the winter precipitation anomaly in (a), (c) southern Iran and (b), (d) the IPX region during La Niña phases. Panels (a) and (b) are related to the years with well-above-normal rainfall in Iran; (c) and (d) are related to the years with below-normal rainfall

Further investigation was therefore carried out to determine if rainfall in southern Iran reveals a strong inverse association with the rainfall of the region near the convective activity. Stronger dependence was evident when precipitation in southern Iran was compared with precipitation over the TBG region, between 6°–10°N and 80°–95°E. Figures 12a,b depict the fluctuation of the precipitation over the TBG region for La Niña phases with (Fig. 12a) and without (Fig. 12b) drought in southern Iran. As indicated, dry and wet periods in southern Iran are generally coincident with positive and negative anomalies of rainfall in the TBG area. Comparing Figs. 11d and 12b delineates the superiority of TBG to IPX, at least regarding rainfall variability over the southern part of Iran (not all southwest Asia).

Fig. 12.

The anomaly maps of the winter precipitation of the TBG region for the cold ENSO phases (a) with and (b) without drought in southern Iran

Fig. 12.

The anomaly maps of the winter precipitation of the TBG region for the cold ENSO phases (a) with and (b) without drought in southern Iran

The associations between the SST of the TBG region and rainfall in southern Iran were also examined for the two categories of La Niña phases. Despite the fact that the tropical Indian Ocean SSTs are generally below normal during these phases, above-normal (below normal) winter rainfall in southern Iran was found to be coincident with the period in which the TBG SSTs are extremely (mildly) cold (Figs. 13a,b). In other words, when the TBG SST is mildly (extremely) cold, above-normal precipitation is anticipated for the TBG region (southern Iran). Drought periods for the TBG region (southern Iran) were also found to coincide with the extremely (mildly) cold SSTs. Besides winter SST, the strong (weak) negative autumnal SSTs of the TBG region were also found to be associated with above-normal (below normal) rainfall in southern Iran during the following winter (Figs. 13c,d). The presented results have indicated that SOI data in conjunction with anomalous data of either precipitation or TBG SST could potentially be used as indices for the prediction of wet and dry conditions in southern Iran.

Fig. 13.

The SST anomaly for (a), (b) winter and (c), (d) the preceding autumn over the TBG region of the cold ENSO phases (a), (b) with and (b), (d) without drought in southern Iran

Fig. 13.

The SST anomaly for (a), (b) winter and (c), (d) the preceding autumn over the TBG region of the cold ENSO phases (a), (b) with and (b), (d) without drought in southern Iran

4. Conclusions

This study presents statistical relationships between the Southern Oscillation phenomenon and precipitation during autumn (October–November–December) and winter (January–February–March) in Iran. Based on SOI data, warm, cold, and neutral ENSO phases were selected and the associated SO–precipitation composites were constructed for each phase. The 10th, 30th, 50th (median), 70th, and 90th percentiles (R10, R30, R50, R70, and R90) of the precipitation composites were then computed for 51 meteorological stations. For each station, the 30th and 70th percentile values of neutral phase composites were then considered as dry and wet thresholds, respectively. The ratio of the 30th and 70th percentiles of precipitation during warm and cold ENSO phases to the corresponding value of neutral phases provides the basis for the evaluation of ENSO effects on the intensity of drought and excess rainfall. The shifts in the occurrence probability of the dry and wet thresholds were also investigated for warm and cold ENSO events.

The results have shown that, in autumn, the ratios of R30w/R30n and R70w/R30n are greater than unity over most parts of the country. This indicates that compared to neutral phases, drought intensity is generally much smaller and most parts of the country tend to experience wet conditions when warm ENSO prevails. The probabilities that rainfall is below-drought thresholds and above-wet thresholds are generally less than 15% and above 60%, respectively. Furthermore, although SOI– rainfall correlations are not significant for southern regions, a warm ENSO phase induces a substantial impact on both drought avoidance and total precipitation amounts over these regions.

In contrast to a warm ENSO phase, for La Niña phases, the autumnal ratios of R30w/R30n and R70w/R30n are usually less than unity, and the probability for serious drought in Iran deviates from 40% to 65%. This probability is particularly high for arid regions of southern Iran, indicating a higher risk of drought hardships for these areas. For such phases, the occurrence probability of autumnal wet periods is generally below 25%, suggesting a small chance of excess rainfall during these phases.

In winter, the response of rainfall to ENSO extreme phases is generally weaker and less wide ranging than the corresponding response during autumn. It is concluded that, since the impact of ENSO on winter precipitation is modulated by other factors including the Siberian high and rainfall in the TBG region, its spatial variation is more complicated than in autumn. It is shown that although warm ENSO phases prevent severe drought over most of the country, wet epochs are also not anticipated during such phases. The probability that during warm events winter rainfall exceeds the considered wet threshold is low and mostly varies from 20% to 30% with maximum values (near 50%) for the region on the southeastern side of the Caspian Sea coasts. In contrast to the eastern coasts, it is more likely that, during warm ENSO phases, western Caspian Sea coasts and eastern and southeastern districts, as well as the northwestern extremity of the country, are seriously affected by winter drought. The more probable mechanism of the interaction between the Siberian high and ENSO phases for producing rainfall over southwestern and southwestern Caspian Sea is explained. A low probability (only 15%–25%) was found for winter drought over the Caspian Sea coastal areas and the western and northwestern parts of the country during La Niña events. On the other hand, the probability that winter precipitation in Iran is above the wet threshold was estimated to be greater than 40%.

During cold ENSO phases, the coastal area of the Persian Gulf and Oman Sea, as well as northeastern districts, could face both a serious shortage and abundant rainfall. To justify the mechanism, the study has found that during cold ENSO events, SOI data in conjunction with the precipitation amount (and also SST anomalies) over the TBG region could be used as the indices for winter drought in southern Iran. During cold ENSO phases for which more than usual winter rainfall had occurred in the TBG region (periods with mild negative SST anomalies in this region), drought was generally widespread over the southern part of Iran. Conversely, the prevalence of dry conditions in the TBG region (extreme negative SST anomalies) generally favors wet conditions over the Iranian southern districts. Moreover, the autumnal SST in the TBG region could be used for the prediction of winter precipitation in southern Iran.

Acknowledgments

The authors gratefully thank Anji Sath of the International Research Institute (IRI) for Climate Prediction, Columbia University, for her kind support regarding technical comments and English editing. The grant provided by the Agricultural Products Insurance Fund in Iran is highly appreciated. The authors are supported by the Research Council of Shiraz University (interuniversities research project). We also thank the anonymous reviewers for providing helpful comments.

REFERENCES

REFERENCES
Allan
,
R. J.
,
J.
Lindesay
, and
C.
Reason
,
1995
:
Multidecadal variability in the climate system over the Indian Ocean.
J. Climate
,
8
,
1853
1873
.
Allan
,
R. J.
,
G. S.
Bread
,
A.
Close
,
A. L.
Herczeg
,
P. D.
Jones
, and
H. J.
Simpson
,
1996
:
Mean sea level pressure indices of the El Niño– Southern Oscillation: Relevance to stream discharge in south-eastern Australia.
CSIRO Divisional Rep. 96/1, Division of Water Resources, 23 pp
.
Barlow
,
M.
,
H.
Cullen
, and
B.
Lyon
,
2002
:
Drought in central and southwest Asia: La Niña, warm pool, and Indian Ocean precipitation.
J. Climate
,
15
,
697
700
.
Cook
,
K. H.
,
2001
:
A Southern Hemisphere wave response to ENSO with implications for southern Africa precipitation.
J. Atmos. Sci
,
58
,
2146
2162
.
Curtis
,
S.
,
A.
Neltin
,
J.
Huffman
,
R.
Adler
, and
D.
Bolvin
,
2001
:
Evolution of tropical and extratropical precipitation anomalies during the 1997–1999 ENSO cycle.
Int. J. Climatol
,
21
,
961
971
.
Drosdowsky
,
W.
,
1995
:
Analogue (non-linear) forecasts of the Southern Oscillation index time series.
NOAA Experimental Long-Lead Forecast Bulletin, Vol. 4, 28–31
.
Ghasemi
,
A. R.
,
2003
:
Meteorological drought in Iran and its association with the El Niño–Southern Oscillation and the Caspian Sea surface temperature.
M.S. thesis, Dept. of Arid Region Management, University of Shiraz, 120 pp
.
Kane
,
R. P.
,
1999
:
El Niño timing and rainfall extremes in India, East Asia and China.
Int. J. Climatol
,
19
,
653
672
.
Kane
,
R. P.
,
2002
:
Precipitation anomalies in southern South America associated with a finer classification of El Niño and La Niña events.
Int. J. Climatol
,
22
,
357
373
.
Lough
,
J. M.
,
1997
:
Regional indices of climate variation: Temperature and rainfall in Queensland, Australia.
Int. J. Climatol
,
17
,
55
66
.
Nazemosadat
,
M. J.
,
1996
:
The impact of oceanic and atmospheric indices on rainfall variability.
Ph.D. thesis, University of New South Wales, 265 pp
.
Nazemosadat
,
M. J.
,
2001
:
Winter rainfall in Iran: ENSO and aloft wind interactions.
Iranian J. Sci. Technol
,
25
,
611
624
.
Nazemosadat
,
M. J.
, and
I.
Cordery
,
1997
:
The influence of geopotential heights on New South Wales rainfalls.
Meteor. Atmos. Phys
,
63
,
179
193
.
Nazemosadat
,
M. J.
, and
I.
Cordery
,
2000a
:
On the relationships between ENSO and autumn rainfall in Iran.
Int. J. Climatol
,
20
,
47
61
.
Nazemosadat
,
M. J.
, and
I.
Cordery
,
2000b
:
The impact of ENSO on winter rainfall in Iran.
Proc. 26th National and 3rd Int. Hydrology and Water Resources Symp., Perth, Australia, Institute of Engineers, 538– 543
.
Nazemosadat
,
M. J.
, and
A. R.
Ghasemi
,
2002
:
The influence of the Caspian Sea SST on the winter and spring rainfalls over northern parts of Iran.
Proc. Int. Conf. on Hydrology and Watershed Management, Hyderbad, India, Jawaharlal Nehru Technological University, 297–304
.
Nazemosadat
,
M. J.
,
A. R.
Ghasemi
,
I.
Cordery
, and
A.
Sharma
,
2003
:
Is ENSO more influential on precipitation in NSW or Northwestern Iran?
Proc. Int. Hydrology and Water Resources Symp., Vol. 2, Wollongong, New South Wales, Australia, Institute of Engineers, 3–10
.
Richard
,
Y.
,
S.
Trzaska
,
P.
Roucou
, and
M.
Rouault
,
2000
:
Modification of the southern African rainfall variability/ENSO relationship since the late 1960s.
Climate Dyn
,
12
,
883
895
.
Ropelewski
,
C. F.
, and
M. S.
Halpert
,
1996
:
Quantifying Southern Oscillation–precipitation relationships.
J. Climate
,
9
,
1043
1059
.
Trenberth
,
K. E.
,
1997
:
The definition of El Niño.
Bull. Amer. Meteor. Soc
,
78
,
2771
2777
.
Zhang
,
Y.
,
K. R.
Sperber
, and
J. S.
Boyle
,
1996
:
Climatology of east Asian winter monsoon and cold surges: Results from the 1979– 1995 NCEP/NCAR reanalysis.
PCMDI Rep. 38, 29 pp
.

Footnotes

Corresponding author address: M. J. Nazemosadat, Irrigation Department, Climate Research Center, College of Agriculture, Shiraz University, Bajgah, Isfahan Freeway, Shiraz, Iran. Email: jafar@shirazu.ac.ir