Trend analysis of climate change compound indices in Iran

Climate change is considered as a global challenge whose effects differ by region and local characteristics (Leonard et al. 2014; Filho et al. 2016). During the past decades, large positive temperature increases from their mean values have become common in many parts of the world. In both hemispheres, the land surface temperature has increased at a faster rate than the oceans according to the Intergovernmental Panel on Climate Change (IPCC 2007). The Earth's surface temperature has maximized over the last three decades and has seen increasing trends since 1850. In this regard, the northern hemisphere has become much warmer from 1983 to 2012 than any 30-year period in the last 1,400 years. The linear trends of the average land and ocean surface combination temperature data during the period 1880 to 2012 reveal an increase of about 0.85 °C (IPCC 2013). The planning and performance of human community infrastructures in relation to food, water, energy, shelter, and transportation depend on climate extreme change values according to the World Meteorological Organization (WMO 2009). The economic changes and the effects of extreme climate change are among the most important challenges humans face regarding climate change, states the Climate Change Science Program (CCSP 2008). The latest outputs of meteorological models suggest that global warming has caused an increase in evaporation from land surface and surface water bodies, which is anticipated to have a serious impact over time on water resources management and the global population (Moazenzadeh et al. 2018). Meanwhile, global warming has caused increasing and several climate events such as drought, heat waves, floods, and fires in different parts of the world (Alexander et al. 2006; Aghakouchak et al. 2014; Leonard et al. 2014). Predicting climate extremes based on temperature and precipitation index in the Fifth Report of the IPCC highlights a significant growth in the number and magnitude of warm and dry periods (Hao et al. 2013; Filho et al. 2016). The daily index consists of temperature or precipitation extremes which have been used for assessing long-time variability and frequency of days above or below specific physically based thresholds (Zhang et al. 2011). Mainly, due to the interdependence and thermodynamics, relations between precipitation and temperature have been addressed in numerous studies (Liu et al. 2012). The main goal of analyzing extremes is to describe the balance and changes of climate to find the most desirable balance in macro plans for applying high safety standards and preventing great loss to communities and systems with regard to severe events. Sometimes, one variable may be in an extreme state, but more often, these events can be considered as a combination of variables, not all of which are necessarily extreme (Leonard et al. 2014). An extreme impact may be a combination of one or more variables leading to a severe change in climate, referred to as a compound event. According to copula theory (Miao et al. 2016), recent studies have been conducted based on describing the compound index of climate combinations as warm/dry, warm/wet, cool/dry, and cool/wet (Lopez-Moreno et al. 2011; Estrella & Menzel 2012; Arsenovic et al. 2013; Hao et al. 2013). These compound indices reveal the change at all spatial or temporal scales with significant trends in the frequency of the cool and warm modes.

Most parts of Iran are located in the desert belt, which is affected by extreme climate change and trends. Therefore, considering trends and climate change and their effects is essential and unavoidable for planning risk management in relation to communities, systems, and infrastructures through adaptation and the bad effects of climate change. Several national studies have investigated trends of climatic variables in Iran. In recent studies, several extreme precipitation and temperature indices have been analyzed (Rahimzadeh et al. 2009; Tabari et al. 2011; Parak et al. 2015), but the climate change was not concerned with the use of compound extreme indices. These studies provided important findings, although they have focused on only the individual elements of the climate and have used mean and extreme values of climate variables. This paper is one of the first national studies to be based on compound values. In this study, the composition of more different climatic elements is considered and several statistical tests were used for trend detection. The main purpose of this paper is to demonstrate the trend of compound indices in order to find the impact of climate change over the study area. In this regard, the trends of combined temperature, precipitation, humidity, wind speed, and sunshine statistics in a spatial domain in Iran are investigated as the main research objectives. The aim of the present research was to understand the variations in compound values of temperature and precipitation and other variables. Compared to previous studies, this can provide a comprehensive view of the behavior of the cool and warm modes of heat and moisture, of which analysis of the statistics of each variable is taken individually.

Iran is the spatial domain used in this study (latitude 25–40 and longitude 44–64). The total area is 1,648,000 km² and it includes a population of about 75 million, according to the latest National Census of Iran. The climate of Iran varies due to differences in latitude, altitude, and a range of geographic features, including mountains and deserts. Currently, daily data are available for more than 150 synoptic stations in Iran, but there are long-term records for only a few stations. Most of the stations, especially in the earlier years, contain inhomogeneities and uncertainties in their data sets. Our study was limited to only 45 stations, due to the problems mentioned above, or to the presence of wide data gaps, and also to the shortness of record length. On the other hand, the data after 2015 have either been unverified or are not available. The selected stations take into account the length and completeness of records. In order to interpret the climate change compound index, the data were collected from 45 synoptic stations in Iran for the period 1981–2015. These stations have long-time data and are well distributed in different climate regions and elevation levels. The data have been obtained from Iran Meteorological Organization (IRIMO) (Figure 1 and Table 1).

Due to the importance of climate change studies, the WMO focuses on quality control and data homogeneity (WMO 2009). One of the most important factors of data inhomogeneity is the change in observation, including time of observation and tools, change of station location, formula, and change in data processing (WMO 2011). Studies of extreme index in climate change have been carried out based on meteorological observation in Iran. As they have errors and uncertainty (Rahimzadeh et al. 2009), so the climatic series have inhomogeneity with sudden and abnormal spikes and deviations and unreal interpretation of climate change. The objective of quality control is to verify whether a reported data value is representative of what was intended to be measured and has not been contaminated by unrelated factors. In this study, ClimPACT software, available as Free Software, is undertaken as the main tool for data analysis. This software is written in R, a language and environment for statistical computing and graphics, and developed by the WMO Expert Team on Climate Change Detection and Indices. First, the ClimPACT quality control (QC) revealed unreasonable values of temperature and precipitation data, such as daily precipitation amounts less than zero and daily maximum temperature less than daily minimum temperature. In addition, the QC also identified outliers in daily maximum and minimum temperatures. It is worth mentioning that all potential data can be evaluated by information on the next and previous day of the event with specialized knowledge about local conditions. The QC test of other climatic variables such as humidity, sunshine, and wind speed was done using simple statistical methods in SPSS software.

There are several methods for the assessment of homogeneity in time series data including relative and absolute methods. The relative tests are standard normal homogeneity test and, more reliably, it is recommended that the test is conducted by correlating the test data series with the homogeneous data series of a neighboring station (Tayanç et al. 1998; Wijngaard et al. 2003; Tsidu 2012). The research uses the R homogeneity test (RH-test) V5 software (Wang & Feng 2013) by using the penalized maximal F-test (PMF) methods to check the homogeneity of the monthly temperature and precipitation and other variables' meteorological data. The software method is based on the PMF or F-test and can identify, and adjust for, multiple change points in a time series (Wang 2008). The F-test is a relative test for structural breaks with reference series, namely multiple change point detection method, and the RH-test has the best detection rate (Yozgatligil & Yazici 2015). In the case of a lack of reference station, the F-test is used for inhomogeneity detection in time series (Wang et al. 2007; WMO 2011; Vincent et al. 2012; Fan & Chen 2016). PMF requires the use of reference stations for the homogeneity analysis, but PMF can be used as an absolute method (i.e., when there are no neighbouring stations to use for comparison). According to the diagrams, if any heterogeneous factors are identified, the time series is assumed to be inhomogeneous, otherwise the time series is homogeneous.

The aim of the indicators is to illustrate the temporal and spatial distribution of climate change. It is important to develop a set of compound indices that are statistically robust to cover a wide range of climates and detect changes in climate extremes. Combined indices and heat wave indices are a set of statistical indices that can cover a wide range of climate characteristics and detect variability and changes in climate. In other words, most extreme weather events are the result of combining climate variables. For example, high temperatures coupled with reduced rainfall can cause heat waves and droughts (Klok & Klein Tank 2008). Therefore, following annual study, a specified compound extreme index consisting of seven indices was considered as specified in Table 2.

As illustrated in Table 2, the combination of the precipitation and temperature quantities represent the four climate combinations: cool/dry (low temperature and low precipitation), cool/wet (low temperature and high precipitation), warm/dry (high temperature and low precipitation), and warm/wet (high temperature and high precipitation). The 10 and 90 percent quantities define an extreme event, according to an IPCC report (2007) and the 25% and 75% values of precipitation and temperature are used as threshold levels for defining the compound extremes of temperature and precipitation in order to capture a large number of events.

Another compound index, which was used in this study, is the UTCI. The scale of the index is able to express even slight differences in the intensity of meteorological stimulus. The assessment of the thermo-physiological effects of the atmospheric environment is one of the key issues in human biometeorology. To quantify these effects, the UTCI is developed in Cooperation in Scientific and Technical Research (COST) action 730. The main aim of the index is to present an environmental-physiological evaluation model to increase the plans related to health and welfare in public climate service, public health systems, prevention designs, and climate effects research (Broede et al. 2012).

In order to calculate UTCI, the data of average wind speed, relative humidity, and sunshine duration and mean radiant temperature (MRT) were used. Sunshine duration was used to estimate direct solar radiation (R_s) (which is required as input to the mean radiant temperature) and relative humidity was used to estimate water vapor pressure based on daily maximum and minimum temperature values (Lemke & Kjellstrom 2012). MRTs were calculated from air temperature, global temperature, and wind speed.

Finally, UTCI is computed by the regression equation found at www.UTCI.org. The rating categories of the UTCI are shown in Table 3.

A positive value of Z indicates an increasing trend and a negative value indicates a decreasing trend. The null hypothesis, H, that there were no trends in the records is either accepted or rejected depending on whether the computed Z statistic is less than or more than the critical value of Z statistics obtained from the normal distribution table at the 5% significance level (Kampata et al. 2008).

Most extreme climate events result from combined climate variables. For instance, high temperature with low rainfall may cause heat waves and drought. The changes in threshold ranges for four precipitations (CD, CW, WD, and WW) at Ramsar meteorological station during 1981–2015 show that the possibility of occurrence is different and dry modes have the highest possibilities since 1998 (Figure 2). There is a clear correspondence between temperature and precipitation compound index. There are long-term changes in temperature and rainfall annual series based on field observations in Figures 3–7.

The annual trends of CD compound index based on temperature and precipitation data in Iran are presented in Figure 3(left). This index has a decreasing trend in most stations. The rate of change in the study period based on location between 1 and 10 days in each decade and 1 and 3 days in each decade is decreasing. Except for 15 cases with significant negative trends (p < 0.05) that are mainly located in the northwest, we cannot find significant negative trends for the annual CD over the whole country, but the decreasing trends are dominant. The changes depend on the location over the course of 1981–2015 in CD mode, and decrease between a large spread of 1–10 days per decade, to a much tighter spread of 1–3 days per decade.

The largest and the smallest significant negative trends of CD days occurred in Tabriz station, about (−) 6.5 day per decade and in Abadan station about (−) 3.5 day per decade (Figure 4). Negative trends in CD were larger than those in CW at most stations. The positive trend in this index was observed in some dispersed small areas (Shahrekord, Sabzevar, Gorgan, Rasht, and Torbat e-heydareyeh stations). The magnitudes of the increasing trends in annual CD ranged between (+) 3 days per decade at Shahrekord station and (+) 0.3 days per decade at Sabzevar station.

In some regions, such as Khoramabad and Bojnourd, there are very weak and indiscernible trends. Also, coastal areas of the Caspian Sea and coastal strips of the Persian Gulf clearly show higher annual CD days.

Distinguishable negative trends of annual CD were found to be in good agreement with the results obtained for other territories in Europe (Beniston 2009), where a tendency towards colder years has been identified.

The CW index is defined as the number of days where T25/P75. Here, T25 indicates temperature occurrences below the 25% quantile, while P75 denotes occurrences above the 75% quantile. The outputs of the statistical tests in the annual CW series over the period of 1981–2015 are shown in Figure 3 (right). Except in parts of the south-west country in the south of the Zagros Mountain range and some dispersed stations, other parts of the country showed negative trends for the CW index. However, the MK test indicated statistically significant trends at 0.05 levels for only the two stations of Ramsar and Sannandaj in this index. In other words, the number of cold/wet days at the eight stations (17%) had an increase over the study period while the other stations (83%) showed a decreasing trend.

The largest negative trends of the CW index can be seen in the northern half of the country and the southern part, especially the south-east of the country. The largest negative trends in CW have been found in Sanandaj station, showing a decrease of approximately (−) 1.68 days per decade and Hamedan (Airport) station with the magnitude of approximately (−) 1.12 days per decade (Figure 5).

In some stations, there are very weak positive trends and fluctuations such that there is no significance. In other words, there are no significant trends and regular dispersal of CW in these areas.

Generally, Iran experienced a negative trend for CD and CW indices. It is reasonable to assume that CD and CW extreme indices are related to general circulation patterns over Iran. The results are consistent with the results of the study by Hao et al. (2013) for the Middle East region.

Based on our results, the index trend of warm/dry extremes has increased all over the country. The results of the WD index (80% of the stations) show the increase at more than 80% of stations (Figure 6(left)). Significant positive trends of WD for mountainous ranges, dry and semi-dry climatic regions were recognized. In addition, the significant increasing trends in annual WD varied between (+) 3.7 days per decade at Chabahar station and (+) 14.5 days per decade at the Kish Island station (Figure 7). In recent years, in many regions such as central Africa, eastern Australia, and some parts of Russia, the WD has increased too (Hao et al. 2013). Except for Shahrekord station located in the south-west of the country, the other areas of the country experienced a positive trend in WD. This may be due to the cooling trends found for the southern half of the Zagros Mountains and Shahrekord station (Parak et al. 2015). Similar behavior in fluctuations in observational extreme precipitation and temperature data sets for 1961–2010 can also be seen in Parak et al. (2015). To answer this anomaly needs more research and more precise demands. It should be noted that in some regions, such as southern Zagros (west) and eastern Alborz (north-east), positive trends were non-significant for the number of the WD index.

Figure 7 depicts the time series of warm/dry (WD) for Kish Island and Chabahar stations. The trend line shows an increase in the annual WD since 1981 for Kish Island and Chabahar stations. These stations are located in coastal and marine climates and extreme conditions were modified by marine conditions. Although in these regions temperatures have been affected by the general trend of global warming over the country, local trends can be due to urbanization in these areas. This pattern is important in occurrences of weather phenomena like heatwaves and droughts that cause social and environmental damage (Hao et al. 2013). In addition, factors such as local and regional air pollution may influence the nature and magnitude of the climatic trends in different ways (De Sario et al. 2013).

Positive trends for the WD were found across most of the country which is in parallel with IPCC findings. Generally, the period from 1983 to 2012 is much warmer than any 30-year period in the last 1,400 years in the northern hemisphere.

The warm/wet (WW) combination trends are shown in Figure 6 (left). The results of linear regression indicate that about 42% of the studied stations show a positive trend in extreme warm/wet index during the past 35 years. However, at the 95% confidence level, few of the observed incremental trends are significant and are mostly located in the north and north-west of the country. Based on the above, most positive trends are relatively weak and non-significant; however, the strongest is at Noushahr station (north of the country) with a trend of (+) 0.56 per decade.

The result also indicated no trends of warm/wet in coastal strips of the Persian Gulf and the south-west and some scattered areas in the country. If the country is divided into two parts with an artificial northeast–southwest line, most of the positive trends are seen on the north side of the line (Figure 6 (right)). Generally, the warm/wet (WW) index has incremental trends over the high-latitude regions, whereas the warm/dry indicator has increased in most areas. Our result is consistent with the trend observed in different countries and at the global scale (Beniston 2009; Hao et al. 2013).

Although there is a similar behavior between WW and WD, magnitude and spatial coherence of the earlier index is very weak. Our results indicate that the stations located in the Caspian Sea region such as Ramsar, Noushahr, Rasht have experienced warm/wet days in most years of the period, whereas other stations have experienced some years with no warm/wet days. In the stations located in the south of the country, the WW index has been zero in most years. In the warm months, especially in summer, the anticyclonic circulation is strengthened in the west of the Arabian Peninsula, and its extension over this area blocks moisture advection into the south of Iran from the moisture source (Parak et al. 2015). This anticyclonic circulation prevents the air from rising and causing precipitation in the area. Such patterns may lead to differences in local climate that may be affected by global warming which tends to make wet areas wetter and dry areas drier (Romm 2015). But the effects differ in different parts of the country. The comparison of results for the indices of WD and WW days showed a similar behavior between the two indices, but the magnitude and spatial consistency of WW days were much less than WD days.

The results of this study indicate that most of the country generally experienced a reduction in the number of days with cold stress. On the other hand, the increase in terms of warm stress also affects climate comfort conditions in the country. In other words, decreasing trends were found in the annual occurrence of the no thermal stress class, at 45 stations (95%), which was statistically significant (p < 0.05) (Figure 8(left)). Negative significant trend rates in UTCI in the whole of the country lay in the range of (−) 4 days per decade (Ahvaz station) to −20 days per decade at (Gorgan station).

In a region like the Zagros Mountain range, the negative trends obtained are not so strong. From the results, it is clear that the decreasing trends of the no thermal stress class in the north, northwest, and southeast regions of Iran were stronger than those in the other areas during the last decades.

Our results revealed that positive trends of UTCI were only experienced at Ardebil and Shiraz stations in the range of 9–26 °C. The significant positive trend for this index was found in the western regions of the Caspian Sea at Ardebil station by (+) 6.9 days per decade (Figure 9).

Based on the results presented in Figure 8 (left), annual occurrence of strong heat stress (32–38 °C) increased approximately all over the country. Furthermore, the value of the significant increasing trend in this index varied from (+) 0.5 to (+) 17.3 days per decade at Zanjan and Chabahar stations, respectively. Less than one-fifth of the country, including regions in the south and southwest, experienced negative trends or no trends in this index, but stations such as Ahvaz, Abadan, Boushehr, Fasa, Kish Island, Bandar e-lengeh, and Bandarabas showed an increasing trend in UTCI events in the annual occurrence of the very strong heat stress (38–46 °C) class.

Generally, compared with other compound quantities of temperature and precipitation index, UTCI is consistent with warming and increases of the warm modes' frequency and the probability of the intensity of the thermal stress is not diminished and has shifted to moderate and high levels. Global warming raises the average temperature which causes heat waves to become more intense and more frequent. For the same reason, heat waves will last longer and cover a large region. In recent years, the severity and frequency of drought has increased due to global warming in many parts of the world (Romm 2015). The results show factors such as topography and latitude play an important role in UTCI distribution. In other words, the annual UTCI increases with decreasing latitude throughout most of Iran.

Trends in TCI, including the number of days with TCI ≥ 60, and the number of days with TCI ≥ 80 are shown in Figure 10. The results indicated that the majority of the trends in the annual TCI60 and TCI80 series were positive. The majority of the country showed a positive significant trend in TCI60. Significant positive trend rates in the TCI60 series were in the range of (+) 1.6 days per decade at Rasht station to (+) 16 days per decade at Ardebil station.

The greatest values of this index can be observed in the provinces located in the northwest of the country.

On the contrary, the magnitude of significant negative trends at Chabahar and Gorgan stations located in the southeast and north of the country, respectively, showed a significant negative trend. The steepest negative trend was seen at Chabahar station (southeast), showing a decrease of (−) 8 days per decade.

Comparison between TCI≥80 and TC≥I60 indicated that the TCI≥80 index has a more irregular distribution than positive and negative trends in the country. Over 75% of the stations studied in the country showed an increase in the annual occurrence of TCI≥80 and 17% of these have significant increasing trends (Figure 10(right)). The largest positive significant trend in TCI≥80 was found at Ardebil station, which shows an increase of 10 days per decade.

About 20% of the stations, mostly in the northwest and southeast of the country, had negative trends. The largest negative trend occurred at Iranshahr station, showing a decrease of (−) 2.2 days per decade. Generally, the relationships that have been observed may be explained by several physically different mechanisms. Such correlations may occur because of changes in cloud cover or changes in precipitation and the consequent changes in the heat balance of the regions under consideration (Zhao & Khalil 1993). Thus, it is worth mentioning the issue of whether the observed trends were related to a long-term process like global warming or if they were part of a multi-decadal natural oscillation. Further studies would be interesting to compare the trends found in our study with the results of other tests to detect trends.

This paper has attempted to provide an appropriate consideration of extreme climate process by using compound extreme index with a primary focus on extreme events. There were long-term records for a few stations before 1981. There are some limitations for most stations, such as inhomogeneities and uncertainties in their data sets caused by changes like location, exposure, instrumentation, and observation practice, and also from missing data. We used synoptic stations whose data were reliable, continuous, and long. These stations cover most of the climate types and are evenly geographically distributed throughout the country. These indices were calculated and analyzed for 47 sites during 1981–2015 to provide a general overview of climate change in Iran. Climatic variables such as temperature, precipitation, humidity, sunshine, and wind speed were analyzed. The results of non-parametric tests highlighted the statistically significant spatially coherent trends in compound values of temperature and precipitation index (using the 25% and 75% value levels) corresponding to a warming trend in the country. It was concluded that in more than 80% of Iran the frequency of the warm modes has increased while the frequency of cold modes has decreased but with smaller magnitudes. Generally, Iran experienced a negative trend for the CD and CW index and a positive trend in CW was observed in some dispersed small areas.

More than 97% of the stations exhibited a positive trend for the annual WD index. Positive trends for the WD were found all over the country which is in parallel with IPCC findings. The results of linear regression indicate that about 42% of the studied stations show a positive trend in extreme warm/wet index that are seen in the northern half of the country. The rate of the significant increasing trends in annual WD varied from (+) 3.7 days per decade at Chabahar station to (+) 14.5 days per decade at Kish Island station. Based on the results of the analysis, apart from a few stations and, more specifically, in Shahrekord station, the overall patterns of compound values of temperature and precipitation show the same trend, and responding to these differences definitely requires further investigation. Both WD and WW indices have similar behavior, but the magnitude and spatial consistency of WW days were much less than WD days. The results show that most of the country generally experienced a reduction in the number of days with cold stress and an increase in warm stress. These factors affect climate comfort conditions in the country.

Based on the results, UTCI is consistent with warming and increases of the warm mode frequency and the probability of the intensity of the thermal stress is not diminished and has shifted to moderate and high levels. Global warming raises the average temperature that causes heatwaves to become more intense and more frequent. At the same time, the occurrence of TCI and UTCI in strong heat stress (32–38 °C) events has increased over most of the country, consistent with increases in the warm modes' frequency. The majority of the country showed a positive significant trend in TCI≥60, and TCI≥80 index has a more irregular distribution than positive and negative trends in the country. The distribution of the annual compound index trends indicated that the negative and positive significant trends mainly occurred in the northwest of Iran. The results also suggest the need for further investigation on the intervention of local factors in the environment, which could be one of the major causes of climate change. The findings of this study confirmed the results of previous studies (e.g., IPCC 2007). Although our results are consistent with the findings of other studies such as a Middle East regional study (Zhang et al. 2005, 2011) and trends of climate extremes over Iran (Parak et al. 2015), the results also suggest the need for further investigation by using gridded daily data and remote sensing data on local anthropogenic factors, which could be one of the major causes of climate change in Iran.

The authors would like to thank the anonymous reviewers for their comments and suggestions which have contributed to improving the manuscript. We also thank the Islamic Republic of Iran Meteorological Organization (IRIMO) for providing the data.