Meteorological Drought Analysis Using Global Climate Model and Drought Indicators in West of Iran

Climate change and global warming impact the frequency of droughts and supply systems. Therefore, it is necessary to conduct appropriate studies to evaluate the impact of climate change on weather patterns and drought. For this purpose, data from 6 synoptic stations located in the wet and temperate areas in the Zagros region in western Iran were used to construct four general atmospheric models including BCC-CSM1, CANESM2, HADGEM2-ES, NORESM1-M under representative concentration pathways (RCPs) 2.6, 4.5, and 8.5, for three future periods (2010-2039), (2040-2069) and (2070-2099). Then, spatio-temporal variations of drought severity and frequency were studied in the study area using SPI and SPEI indices in different periods up to 2100. The results showed the spatial extent of areas classied as extremely dry will increase by 47.9% in the rst period compared to the base period. In the second and third periods, however, the severely dry class covers more area. Analysis of SPEI showed that drought will be more severe in all future periods. According to SPEI, drought frequency will increase by 2% according to the rst period (2010-2039) relative to the base period (1984-2013), and by 0.3% in the second and third periods by 2099. The results of this study indicate that the severity, frequency, and impacts of drought will increase in the study area until the end of the century. Therefore, appropriate measures should be taken to control and reduce its potential effects in the future.


Introduction
Climate change is one of the most important challenges facing humans and natural ecosystems, bringing about consequences such as the elevated occurrence of atmospheric-climatic hazards, increased number of storm events, reduced yield of crops and orchards, and lowered food security, as well as increased migration from areas facing more severe climate change (Chanapathi 2020; Memarian and Akbari 2021). Drought in uences different parts of an ecosystem on different scales. Studies show that in some regions of the world, climate change has triggered droughts, intensi ed its damaging effects, or created conditions for subsequent droughts. Droughts occur naturally, but climate change has generally accelerated the contributing hydrological processes, making them less gradual and more intense, with many consequences, not the least of which is increased wild re risk. Different types of drought are being studied, such as meteorological, agricultural, hydrological, and socioeconomic droughts; however, a lack of unanimous de nition complicates the study of droughts (Tang 2020;Mukherjee et al. 2018;Haile et al. 2020). According to the Hydrological de nition, droughts are extreme hydrological phenomena that are characterized by a long-term lack of precipitation in a vast area, in any climatic condition (Taxak et al. 2014; Akbari et al. 2016).
Various scenarios for the increase in greenhouse gases have been de ned in terms of general circulation patterns of the atmosphere to predict climate change. Since the spatial accuracy of general circulation models is very low and the effects of variations have limited applicability, these models need to be converted for smaller spatial scales, which are referred to as the "scalar" model. In general, there are two methods for the conversion of general circulation models for ner scales: dynamic and statistical methods, in addition to combinations of both. Many different approaches have been proposed to carry out these conversions, the most reliable of which is the use of coupled AOGCM simulations, integrating (coupling) global climate models (GCMs) and regional climate models (RCMs) to enable the study of climate change over time in different parts of the world (Semenov 2009 Drought monitoring has helped many users and organizations by capturing the characteristics of droughts using droughts indexes. More than 100 drought indexes have been suggested so far, some of which have been used to describe drought characteristics in grid maps at regional and national scales (Zargar et Lweendo et al. 2017). The Zagros region (a large region in the west of Iran), has been subjected to tensions due to lack of precipitation, which has led to the depletion of the region and serious socioeconomic problems. Although some studies have utilized different models to assess climate change and drought in the region, few have combined different models with climate change scenarios for meteorological analysis of droughts. Therefore, we assessed the impact of climate change on drought in the Zagros region of Iran using four general circulation models, under three climate change scenarios. Also, the Standardized Precipitation Index (SPI) and the Standardized Precipitation Evapotranspiration Index (SPEI) were calculated for future periods. In this work, we 1-examine the variation in SPI and SPEI to describe variation in drought characteristics in the Zagros region, 2evaluate the impact of climate change on drought characteristics, 3-predict drought severity by 2099, and 4investigate the spatial distribution of changes in drought patterns.

Study area
The Zagros climatic zone covers Iran from the northwest to the southwest (Fig. 1). The region has a temperate and humid climate characterized by seasonal precipitation. According to the climatic classi cation of (Alijani 1999), the regions consist of three climate classes: sub-humid subtropical, cold, and semi-mountainous areas. The average annual precipitation (P) varies between 250 and 800 mm in different parts of the region. Rainfall historically occurs during winter and spring. Summers are hot and dry (mean temperature: 35 ºC) and winters are cool (mean temperature: 7 ºC) (Fathizadeh et al. 2013). Past studies show that only six synoptic stations in the area have data with adequate quality for this research. Table 1 shows the characteristics of these stations.

Dataset
Minimum temperature, maximum temperature, irradiation, and precipitation data were obtained daily from six synoptic stations in the Zagros region. The common statistical basis was considered from 1984 to 2013, and preprocessing of data took place.

Downscaling statistical data
For this purpose, past and future climates at the six stations were simulated. The basic requirement of the model at the calibration stage is a le that identi es past climate at the stations ( Semenov and Stratonovitch 2010). This le was compiled using daily precipitation data, minimum temperature, maximum temperature, and irradiance at synoptic stations, taking 1984-2013 as the base period. According to the latest IPCC recommendation, the use of multiple models in climatic simulations, instead of running the models individually, can be effective in reducing the uncertainties in the model. To that end, we used four GCMs (BCC-CSM1, CANESM2, HADGEM2-ES, NORESM1-M, Table 2), and three representative concentration pathways (RCP 2.6, 4.5, and 8.5), and calculated the arithmetic mean for each group in R. Finally, precipitation, minimum temperature, maximum temperature, and irradiance were projected for future periods (2010-2039) and (2040-2069) and (2070-2099).

Standardized Precipitation Index (SPI)
The standardized precipitation index is one of the most important indices in drought monitoring. This index is one of the few time-sensitive drought indices. SPI allows for determining the time scale, depending on which aspects of the impact of droughts are more important (e.g., agricultural resources, hydrological features, etc. year. Only the precipitation parameter is used in the calculation of SPI. The precipitation of each station is calculated on the timescale. Table 3 shows the classi cation of SPI and SPEI indices according to (McKee 1995).
Where T is the average monthly temperature in Celsius, m is the coe cient of dependence on I, I is the sum of 12month heat indexes, k is the correction factor in terms of month and latitude, NDM is the number of days in a month, and N is the maximum number of hours of sunlight. Thus, with a value for PET, the difference between the precipitation (P) and PET for the month i is calculated: The calculation of SPEI requires a three-parameter distribution. The probability distribution function of series D is based on the following equation: Where, α, β and are the parameters of scale, shape and position for D values (Singh et al. 1993 Kernel smoothing can be a good option given noisy measurements x(t i ) of processes at irregularly spaced times t i .
The smoother is a weighted average (Eqs 8 and 9).
3 Results LARS-WG is one of the most commonly applied weather generators, in which the meteorological data for the future period are generated by considering temperature and precipitation alterations under the climate scenarios of RCP 2.6, 4.5 and 8.5. According to the obtained SPI values, droughts will increase in the eastern and northern parts of the regions, while the opposite will take place in the western part of the region. However, SPEI shows that drought will increase from the center of the region to the west. Overall, drought will increase in the entire region by about 11-15% compared to the base period.
SPI and SPEI were used to study temporal and spatial changes of drought severity and frequency in the Zagros region in different periods until 2099. SPI shows that although the spatial extent of drought increases signi cantly in the rst period compared to the base period (47.9%), in the second and third periods of the drought compared with the base period, dry classes have a much larger area. However, SPEI indicates the spatial dominance of drought with very high intensity in all periods. A general review of the results indicates an increase in the frequency and severity of drought as well as the development of drought-affected areas by the end of the century. Therefore, it is necessary to take appropriate measures to reduce the potential effects of climate change on different ecosystems in the region.

Impact of climate change on drought based on SPI
The standardized precipitation index (SPI) depends on precipitation as a single variable, while the standardized precipitation evapotranspiration index (SPEI) is obtained from precipitation and temperature in the form of a simple ( ) water balance. Both indicators detected changes in the frequency of drought in the future compared to the base period. SPI and SPEI agreed on the direction of change, but showed different effects for climate change on drought conditions. SPI is useful because it only needs rainfall as input. However, the use of SPI to describe drought should be done with caution. According to Table 4, normal dry and normal wet classes are the most frequent. The extremely wet class has the lowest frequency (17.7 month). On the whole, more than 7% of the month in the base period (1984-2013) were categorized in the severely dry and extremely dry classes.

Changes in the rst period compared with the base period
Our analysis showed that the frequency of drought in the rst period will increase by 28% compared to the base period, while the frequency of wet classes will increase by 0.35%. Normal classes (both wet and dry) were unchanged, and the largest difference was observed for the dry classes with an increase of about 9.23% (As shown in Tables 4 and 5).

Changes in the second period compared with the base period
The result showed the frequency of drought will increase by 4.78% compared to the base period, and the wet period will decrease by 7.7% the percentage of changes in the normal classes has decreased by 1.41%. Among the stations, Sanandaj shows the largest change in the occurrence of dry periods with an increase of 84.88% compared to the base period (Tables 4 and 6).

Spatial changes in SPI
The frequency of three levels of the drought was calculated and their spatial variations were plotted using kernel smoothing. (Fig. 3) shows the percentage of the study area covered by each class. In the base period, the northeastern and central parts of the study area experienced the most severe droughts, with all three classes having roughly the same area ( Fig. 2 and Fig. 3). As we move from the northeast to the northwest and from north to south, In the rst period (2010-2039), the most severe droughts are expected in the central and northeastern parts of the region and the least severe droughts in the southeastern parts. In this period, the severity of droughts will increase, so that about half of the region will be exposed to very severe drought. In the second period (2040-2069), drought is projected to occur in the north. On the other hand, the southeastern and southwestern parts are less likely to experience severe drought. The severely dry class will cover more than half of the region in this period. In the third period (2070-2099), drought is projected mainly in the northeast. The extent of areas in the severely dry class will increase compared to the base period, but coverage will decrease for the moderately dry class.

Impact of climate change on drought based on SPEI
Extremely dry and severely dry droughts identi ed by SPEI are generally more severe than those identi ed by SPI in terms of affected area and duration.

Changes in the rst period compared with the base period
Projected SPEI showed that the frequency of droughts will increase by 11.9% in the rst period compared to the base period, and that the frequency of wet months will decrease by 6%. The frequency of normal dry and normal wet months showed a 1% decrease. Shiraz station will experience the greatest increase in dry periods (52%) compared to baseline (Tables 8 and 9).

Changes in the second period compared with the base period
Results showed that the frequency of drought in the second period will be 12.44% higher than the base period, which is similar to the changes in the rst period. Wet periods are also reduced by 1.3%, and the percentage change in the normally dry and normal wet months is expected to be -2.47% (Tables 8 and 10).

Changes in the third period compared with the base period
The frequency of drought will increase by 22% in the third period compared to the base period and wet periods will decrease by 7%. Also, normal dry and normal wet periods will undergo a decrease of 61.3%. At Khorramabad station, signi cantly compared to the base period, and Sanandaj station will most frequently experience dry periods among the stations, with a large change (23%) in drought months compared to the base period (Tables 8 and 11).

Spatial changes in SPEI
Investigating the spatial extent of changes in SPEI in the base period showed droughts were most frequent in the central and northeastern parts of the region and least frequent in the south of the region (Fig. 4). In the rst period, extreme droughts are predicted to occur to the south of the central areas, and the northern parts will experience less severe droughts. The central part of the area is affected by severely dry conditions. The severely dry class will cover nearly half of the region (47.6%) ( Fig. 4 and Fig. 5). In the second period, extremely dry conditions are expected in the north and south of the region, and severely dry conditions are predicted from the center towards the north. The severely dry condition will dominate the region with a coverage of 55.4% ( Fig. 4 and Fig. 5). In the third period, the northwest will have moderately dry conditions, and the severely dry class will be the most prevalent at 45.25% ( Fig.   4 and Fig. 5). predictions made using SPI and SPEI. From a time series perspective, the changes in SPI and SPEI at each time scale had some similarities. In short timescales, these two indices had the most uctuation and the difference between them was the largest. There was a 30% difference between the frequency of drought intensity classes predicted based on SPI and SPEI. Over long timescales, the uctuations in SPI and SPEI were mild and the differences between them decreased. However, these small changes still led to different classi cation outputs. Also kernel smoothing Compared to the classical kriging approach, it shows better performance in certain cases of data sets and data with unstable behavior.

Discussion And Conclusion
In this study, standardized precipitation index (SPI) and standardized precipitation evapotranspiration index (SPEI) were used to study droughts in the Zagros region of Iran in the past and future. SPI showed that normal classes are predominant in different periods. However, in the SPEI index, this frequency will increase by 2% relative to the rst period relative to the base period, and by 0.3% in the second and third periods by 2099. Using SPI, we found that the extremely dry class covers the central and southern parts of the region, and the mildly arid class occupies the northwestern and northeastern parts, with a smaller area. Using SPEI however, it was observed that the extremely dry class will gradually extend to the northern and southern parts of Zagros, which is consistent with the results of Evaluating the performance of GCMs and simulating future rainfall is important to understand current climate change and its impact on hydrology, water resources, agriculture, and ecology. We found that droughts will become more frequent in the Zagros region by 2100, which aligns with past studies ( One of the most prominent features of this research is the joint implementation of models to predict the geographical and temporal extent of droughts in a semi-humid area. The use of other drought indicators along with other GCMs can be considered in future research. Also, the development of a network of stations measuring climate parameters could improve the accuracy, reliability, and coherence of predictions. Based on the results, it is expected that the range of areas affected by drought in the developed region will be necessary, therefore, it is necessary to make the arrangements and exibility of the community-based in this region. These predictions can be considered for the formulation of adaptation strategies in the face of drought. The results of this research can be used to manage drought risk in the region and to coherently develop water and water supply development projects. These results are strongly dependent on simulations from only one set of climate change data. There are therefore limitations due to the uncertainty in various climate change scenarios. However, this study showed that the Zagros region will be exposed to signi cant drought due to climate change. Figure 1 The location of the study area and the synoptic stations.

Figure 2
The spatial distribution of three classes of drought severity based on SPI during the base period (A), rst period (B), the second period (C), and third period (D) Figure 4 The spatial distribution of three classes of drought severity based on SPEI during the base period (A), rst period (B), the second period (C), and third period (D)