* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Projection of Temperature and Precipitation in Southern Iran Using
Global warming controversy wikipedia , lookup
Fred Singer wikipedia , lookup
Economics of global warming wikipedia , lookup
Media coverage of global warming wikipedia , lookup
Climatic Research Unit email controversy wikipedia , lookup
Climate change in Tuvalu wikipedia , lookup
Climate change and agriculture wikipedia , lookup
Effects of global warming on human health wikipedia , lookup
Scientific opinion on climate change wikipedia , lookup
Solar radiation management wikipedia , lookup
Global warming hiatus wikipedia , lookup
Global warming wikipedia , lookup
Climate change feedback wikipedia , lookup
Public opinion on global warming wikipedia , lookup
Climate change and poverty wikipedia , lookup
Climate sensitivity wikipedia , lookup
Surveys of scientists' views on climate change wikipedia , lookup
Effects of global warming on humans wikipedia , lookup
Attribution of recent climate change wikipedia , lookup
Years of Living Dangerously wikipedia , lookup
Effects of global warming wikipedia , lookup
Physical impacts of climate change wikipedia , lookup
North Report wikipedia , lookup
Climate change, industry and society wikipedia , lookup
Climatic Research Unit documents wikipedia , lookup
General circulation model wikipedia , lookup
Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 DOI 10.1007/s40995-016-0009-8 RESEARCH PAPER Projection of Temperature and Precipitation in Southern Iran Using ECHAM5 Simulations M. J. Nazemosadat1 • V. Ravan1 • E. Kahya2 • H. Ghaedamini1 Received: 10 June 2015 / Accepted: 4 September 2015 / Published online: 13 April 2016 Ó Shiraz University 2016 Abstract This study applied a statistical approach to project temperature and precipitation condition in southern parts of Iran for the period 2016–2045. The applied methodology transformed the probability distribution of the observed historical series into the corresponding simulated time series of the ECHAM5 GCM model. Results showed that the mean annual temperature of the study area for the period 2016–2045 is greater than corresponding climatological mean by about 0.5–1.0 °C, if the emissions are within the range of the 20C3M scenario. These values, however, increased to about 2.0–2.4 °C when the high emission scenario of 1PTO2X was applied. We examined conformity between the trend lines of the historical temperature data and corresponding projected time series. The conformity was substantially improved when the outcomes of the 1PTO2X rather than 20C3M scenario were used. Under the 20C3M scenario, the mean values of the projected precipitation data were less than corresponding amounts of the historical time series by about 20%. Keywords Climate change GCM ECHAM5 Iran & M. J. Nazemosadat [email protected] 1 Atmospheric and Oceanic Research Center, Water Engineering Department, College of Agriculture, Shiraz University, Shiraz, Iran 2 Faculty of Civil Engineering, Istanbul Technical University, Istanbul, Turkey 1 Introduction Global warming and its associated causes and aftermaths are among the recent challenges in various scientific disciplines including meteorology, climatology, oceanography and environmental and socio-economic sciences (Ding and Houghton 2001; National Research Council 2001; Parry et al. 2007; Meehl et al. 2007). The primary greenhouse gases in the earth’s atmosphere are water vapor, carbon dioxide, methane, nitrous oxide, and ozone that absorb and emit radiation within the thermal infrared range. The net function of these gases on absorbing and emitting the terrestrial outgoing long-wave radiation has increased the biosphere’s theoretical temperature by about 33 °C and decreased the difference between day and night temperatures globally. It is worthwhile to note that, from theoretical point of view, average temperature of the earth’s surface is about -18 °C, although due to this radiative function of these gases, it is about ?15 °C in reality. In spite of the fact that water vapor is one of the main greenhouse gases which modulates global air temperature, abnormalities have not yet reported in its concentration in terrestrial scale. According to the recent observations, the current upward trend in the global temperature is mostly harmonized with the similar trend in the atmospheric concentrations of CO2, CH4, N2O and O3 that are mostly due to the human activities (Mestre-Sanchis and Feijoo-Bello 2009). It means that, in addition to possible natural sources, global warming is critically affected by those anthropogenic causes that are responsible for further release of these gases. In other words, the potential impact of the global warming on human activities such as agriculture, industry, urbanization and water supply depends on the future concentration and mixture of the greenhouse gases. 123 40 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 A number of climate forecast models with different temporal and spatial resolutions have been developed to project the future climate condition. Various scenarios of industrialization and gas emission are considered to make these models more compatible with the desires of decision makers and scientific communities. Projection of the future climate change and estimation of the potential consequences of this change on both human societies and natural system are the concern of the most well-known climate models (Meehl et al. 2007). Prediction accuracy of these models mostly depends on the adapted initial condition and the number of independent biospheric parameters considered for their development (U.S. National Assessment Synthesis Team 2000). Harmonized with advancement in computational technology, the emphasis of today’s exploration has mostly shifted toward the use of more complicated forecast schemes known as the general climate models (GCMs). The current dynamical GCMs are the outcome of mathematical analysis of the planetary ocean–atmosphere circulations using the Navier– Stokes equations on a rotating sphere (Roeckner et al. 2003). These equations are the basis of the complex computer programs commonly used to simulate the air-sea circulations and their interactions in various spatial and temporal scales. The atmospheric and oceanic GCMs (namely AGCM and OGCM) are the key parts of global climate models along with sea ice and land surface components. The GCM models are widely used for the weather forecasting, understanding current climate and projecting future climate. Some of the well-known GCM models include HADCM3, ECHAM5, CCSM3, PCM1, CNRM-CM3, CSIRO-MK3 MIROC-3.2, IPSL-CM4, CGCM-3.1, and GISS-ER. The GCM models were firstly being used for the climate forecasting at the global or continental coarse scale with an effective resolution of 200–300 km (Palmer and Anderson 1994; Hunt 1997; Mason et al. 1999; Gates et al. 1999; Goddard et al. 2001; Landman and Goddard 2002; Washington and Preston 2006). Although such spatial resolutions are appropriate for planetary scale analysis, they are hardly able to characterize local sub-grid features such as coastlines, steep topography, and vegetation gradients with an acceptable accuracy. Due to these deficiencies, biases were, frequently reported between GCMs typical outputs and observed meteorological parameters (Joubert and Hewitson 1997; Mason and Joubert 1997; Salathe et al. 2007). To obtain climate change information at local and regional levels (critical for impact assessments) and to account for land surface heterogeneity at the same time, GCM outputs need to be downscaled to finer resolutions (Christensen et al. 2007). Landman and Tennant (2000) and Landman et al. (2001), applied statistical approach to downscale the GCM outputs over southern Africa. Washington and Preston (2006) studied the role of the Indian 123 Ocean sea surface temperatures on the occurrence of wet years over southern parts of Africa by downscaling outputs of the HadAM3 GCM model. For the assessment of the effects of climate change on daily precipitation in India, Raje and Mujumdar (2009) used a statistical approach to downscale GCM simulations. Almazroui (2012) used regional climate model (RegCM4) nested in the European Community–Hamburg atmospheric model (ECHAM5) and the European Centre for Medium-Range Weather Forecasts’ (ECMWF) 40-year reanalysis (ERA40) data to analyze the effects of climate change on the Arabian Peninsula. While ECHAM5 and ERA40 exhibited drier and warmer biases over the Peninsula, RegCM4 predicted higher rainfall intensity and lower temperatures. ECHAM5 is one of the most recent versions in a series of ECHAM models, developed at the Max Planck Institute for Meteorology, Germany uses 1.8° 9 1.8° grid cells of a 31-layer atmosphere and a 40-layer ocean (Roeckner et al. 2003). Two scenario of this model, namely 20C3M and 1PTO2X, are used here for simulating historical data and projection of climate change in southern parts of Iran. The 20C3M scenario is an experiment with the assumption that the concentration of greenhouse gases will increase with the rate as was observed through the 20th century. This increasing rate is, however, steeper for the 1PTO2X scenario, since this experiment assumes that the increase rate of the greenhouse gases is 1 % per year from pre-industrial level until the concentration has doubled and held constant thereafter (1 % to double). Guo et al. (2012) have reproduced four sets of twenty year simulation that were respectively carried out by the ECHAM5 with three spatial resolutions and the RegCM3 nested in ECHAM5 in oneway mode (ECHAM5-RegCM3). Altitudinal and latitudinal dependence of future warming in Taiwan simulated by WRF nested with ECHAM5/ MPIOM has been evaluated by Lin et al. (2014). Simulation results showed close correlation between fine resolution downscaling by WRF nested with ECHAM5/MPIOM and the actual observation data for the period 1979–2003. Projection of future climate changes revealed both altitudinal and latitudinal variations in warming trend, with more significant temperature increase in mountain areas than in plain areas towards the end of the 21st century and more obvious warming in the north than in the south of Taiwan. Fei (2014) evaluated the possibility of simulating variability in the ENSO indicators using the ECHAM5/MPIOM model. They found that conformity between the observed and simulated data is quite realistic in terms of structure, strength and period. The annual cycle of SST and the phase of ENSO events were successfully reproduced by the ECHAM5/MPIOM. Reboita et al. (2014) studied climate projections of air temperature and precipitation over South America from the Regional Climate Model version 3 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 41 (RegCM3) nested in ECHAM5 and HadCM3 global models. According to their report the RegCM3 projections indicated general warming throughout all regions and seasons. The RegCM3 projections suggested a negative trend of precipitation over northern parts of South America. Mohammadi et al. (2015) assessed the impacts of future climatic change on the irrigated rice yield over southern coast of the Caspian Sea under three scenarios of climate change. In addition to local climate data, they utilized the outputs of some GCMs including HADCM3, ECHAM5, IPCM4, GFCM2, NCCCSM and INCM3. Their validation results suggested that for projecting future conditions of temperature and precipitation in this part of Iran, the ECHAM5 climate model has the highest correlation with the lowest error. Due to high temporal and spatial resolutions and also easier accessibility of the model output (from authors point of view), the ECHAM5 model and its two mentioned scenarios were used to project future status of temperature and precipitation in southern Iran. The coupled ocean–atmosphere GCMs do not provide predictions of the distant future in the sense that we try to predict the weather one day or a few days ahead. The good GCMs will generate realistic sequences of main global climate phenomenon such as El Niño, hurricanes, or monsoons, but not the same sequence as has been observed historically (Dettinger et al. 2004). In other words, only the statistics (mean rates, frequencies of occurrence, etc.) would ever be expected to be reproduced in the GCMs. This study has, therefore, been prompted to project changes in the magnitudes of future temperature and precipitation in southern parts of Iran for the period 2016–2045 by downscaling ECHAM5 model outputs. The outcome of the study is, hopefully, beneficial for long-term planning of socio-economical activities in Iran, where the shortage of water resources and frequent of climatic hazards cost lives and billions of dollars annually. Fig. 1 The geographical location of the considered synoptic stations on the global 1.8° 9 1.8° network of the ECHAM5 model. While the black circles represent the position of the ground stations, rectangles signify the selected nodes whose data were used for downscaling procedure as indicated in Table 2 2 Materials and Methods 2.1 Data Monthly mean temperature and precipitation data for 12 synoptic stations spread over various parts of southern Iran were gratefully obtained from the website of the I.R Iran Meteorological Organization (http://www.weather.ir). Geographical locations and record length of these stations are shown in Fig. 1 and Table 1. Since about 85–90 % of the annual precipitation of the considered stations occurs during the October–March period, precipitation data were merely analyzed for these 6 months. Temperature data were, however, studied for all months of the year. The record lengths of the considered time series vary from 55 to 16 years as indicated in Table 1. 123 42 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 Table 1 The geographical position and record length of the selected stations Station Latitude (°N) Longitude (°E) Period Abadan 30.22 48.15 1951–2005 Ahwaz 31.20 48.40 1957–2005 Bandar Abass 27. 13 56.22 1957–2005 Bandar Daier 27.50 51.56 1994–2005 Bandar Mahshahr 30.33 49.90 1988–2005 Boushehr 28.59 50.50 1951–2005 Fasa Jask 28.58 25.38 53.41 57.46 1967–2005 1985–2005 Abomoosa Island 25.50 54.50 1984–2005 Kangan Jam 27.49 52.22 1990–2005 Lar 27.41 54.17 1990–2005 Shiraz 29.36 52.32 1951–2005 Monthly simulations of temperature and precipitation data for both 20C3M and 1PTO2X scenarios were appreciatively extracted from the http://cera-www.dkrz. de/wdcc/ui/index.jsp webpage for the period 1951–2045. These data were collected for 20 nodes that surround the ground-based stations so that each of these stations is encircled by four nodes (Fig. 1). As it will be shown later, these four sets of simulated data were then examined to select the most proper nodes for final downscaling procedure. Due to an intrinsic structure of the GCM models, no yearto-year correspondence between the historical observations and simulations is expected. However, ideally, one would hope that the probability distributions of the observed and projected series are comparable. Statistical downscaling has an important advantage over the regional models since it is computationally efficient and allows the consideration of a large set of climate scenarios (Dettinger et al. 2004). The applied procedure consisted of the following steps: A: B: C: 2.2 Statistical Downscaling Correlation analysis was performed to investigate the linkage between the station data and four sets of the model output around each station (Fig. 1). The linkage was found to be poor for all stations which implicates that the GCM model could not directly reproduce the historical data in a fine resolution. The simple deterministic downscaling function as described by Dettinger et al. (2004) was, therefore, applied to capture the probability distribution of the historical series and transfer it into the GCM projected dataset. According to this methodology, the historical GCM’s output data are firstly modified so that their probability distributions approximate the corresponding distribution of the weather station data. The same deterministic transformations were then applied to produce climate projection for the period of 2016–2045 (O’Brien et al. 2001). Pryor et al. (2005) has successfully used a similar approach for downscaling wind data. Their results were generally found to be comparable with the outcomes of the sophisticated regional dynamical models. We adapt the inherent probability distribution of the observed data as the main basis for downscaling and subsequent estimation of the projected climate change. 123 Monthly time series of temperature and precipitation data were sorted for each individual weather station. Likewise, the sorted historical data were also obtained for the GCM simulated data. Since each weather station was surrounded by four 1.8° 9 1.8° nodes, corresponding to one set of the observed data, four sets of the simulated values with similar record length were arranged for each of the GCM’s scenarios. Exceedance probability was computed for each of the sorted data using the Weibull probability function m (i.e., P ¼ Nþ1 , where m and N denote rank of the considered data and sample size, respectively). Since observed and simulated data have identical sample size, the assigned probability was also equal for the observed and simulated data laying in an identical row. A simple regression analysis was applied to derive the regression equation between the sorted values of observed and simulated data around each station. Among four surrounding nodes, the one with the highest value of the correlation determination (R2) and least of mean square error (RMSE) was selected for further analysis, and the other three were discarded (Table 2; Fig. 1). The applied equations for R2 and RMSE are: PK 2 m¼1 Xm Ym R ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1Þ PK P 2ffi 2 Y X m m¼1 m vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u K uP 2 u tm¼1 ðXm Ym Þ ð2Þ RMSE ¼ K where Xm and Ym denote the observed and simulated values, respectively. K is the sample size of the series. The application of this form of bias correction (i.e., rank regression) is recommended when the GCM outputs are used to reproduce the specific patterns of historical events rather than an attempt to ‘‘correct’’ day-by-day GCM outputs (Dettinger et al. 2004). Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 D: E: By plugging observed values into the developed regression equation (i.e., between observed and selected node), a new set of data, namely, historical downscaled values were generated. With this generation, corresponding to each of the proposed exceedance probabilities, three sets of data consisting of observed, model simulated and downscaled values were obtained. A proper regression analysis was performed between the observed and the downscaled data. The obtained regression equation was then used to project the temperature and precipitation values for the period 2016–2045. The linear or polynomial regression approach generally leads to more appropriate good fit for temperature or precipitation data, respectively. The better the regression fit, the more preservation of statistical moments through the future projection. 3 Results and Discussion 3.1 The Probability Distributions (Temperature vs. Precipitation) The probability distribution of precipitation was found somewhat more complex than the essentially Gaussian temperature distribution. This is because precipitation data includes many months without precipitation and because precipitation seldom exhibits a Gaussian distribution. Moreover, due to the different mechanisms of rainfall production, the probability distributions might be dissimilar even for the stations not far from each other. For instance, according to Nazemosadat and Ghaedamini (2010), the measure of the effects of the Madden Julian Oscillation (MJO), on Iran’s precipitation is different between southwestern and southeastern districts. Such discrepancies were also found when the influences of the El Niño Southern Oscillation phenomenon on Iran’s precipitation were analyzed (Nazemosadat and Ghasemi 2004). Overall, for those stations located in the arid zones of subtropical high, the precipitation data generally have a complicated probability distribution, and downscaling procedure of these data is more tedious than temperature time series. Figure 2 illustrates probability distributions for the observed, simulated and downscaled temperature and precipitation data (in January) at Shiraz station, the capital city of Fars province in Iran as an example. A simple comparison between the Fig. 2a, b implicates that the probability distribution of precipitation is more complicated than that for temperature data. For instance, if probability distribution of the simulated temperature data shifted to the left by about 5 °C, it would be roughly overlaid with the 43 corresponding distribution of the observed records. In other words, for this particular month and station and if the extreme values are ignored, by subtracting model outputs by 5 °C, they will be transformed into the observed data. Such a simple relationship is, however, not evident between the observed and simulated precipitation data. As indicated in Fig. 2b, for the period 1951–2005, January precipitation was above 130 mm for 8 times and even it reached to 324.5 mm in 1965. The simulated data are, however, essentially below 130 mm and the applied model failed to reproduce data greater than this value. Incompatibility between the observed and simulated data is not restricted to wet years, but also evident during the years with low or even moderate precipitation. This inconsistency between the observed and simulated time series implies that, without performing a proper downscaling procedure, the GCM outputs cannot directly project future climate of an arid zone such as southern Iran. Using the GCM simulations without careful examination of historical climate behavior might also lead to a wrong conclusion. Table 3 shows the mean monthly values (i.e., average values of 12 months) of R2 and RMSE for the correlations between observed and simulated temperature data before and after downscaling. As indicated, the applied downscaling procedure has dramatically improved the earlier poor linkage between these two datasets so that all correlations become strongly significant after downscaling. Similar enhancement in the correlations was also found for Fig. 2 The probability distribution of observed, simulated and downscaled data for Shiraz station during January 1951–2005: a temperature and b precipitation data 123 44 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 Table 2 The nodes whose data were used for downscaling the observed data at the mentioned stations (see Fig. 1) Station Boushehr Mahshahr Bandar Daier Bandar Abass Ahwaz Abadan Node N9 N4 N14 N16 N2 N3 Station Shiraz Lar Kangan Jam Abomoosa Island Jask Fasa Node N6 N15 N14 N18 N20 N10 Table 3 The mean monthly values of R2 and RMSE between observed and simulated temperature data before and after downscaling for the scenarios of 20C3M and 1PTO2X 20C3M scenario Obs-Sim R 2 a 1PTO2X scenario Obs-Ds RMSE R 2 Obs-Sim RMSE R 2 Obs-Ds RMSE R2 RMSE Abadan 0.02 2.28 0.92 0.69 0.01 2.49 0.88 0.67 Ahwaz 0.03 2.39 0.99 0.14 0.02 2.90 0.98 0.47 Bandar Abass 0.01 3.11 0.95 0.35 0.01 2.91 0.94 0.83 Bandar Daier Mahshahr 0.01 0.06 3.23 3.20 0.92 0.95 0.41 0.19 0.01 0.03 3.32 3.41 0.83 0.94 0.58 0.25 Boushehr 0.01 3.17 0.95 0.26 0.01 3.19 0.92 0.35 Fasa 0.03 2.73 0.96 0.43 0.02 2.85 0.96 0.62 Jask 0.08 3.09 0.98 0.17 0.06 3.11 0.96 0.19 Abomoosa Island 0.02 2.65 0.98 0.25 0.01 3.01 0.96 0.60 Kangan Jam 0.04 3.13 0.97 0.20 0.04 3.16 0.93 0.23 Lar 0.01 2.49 0.98 0.11 0.01 2.67 0.97 0.13 Shiraz 0.06 2.07 0.94 0.37 0.05 2.21 0.92 0.39 The presented values are the arithmetic mean of the given statistics for monthly data during the January–December period. All the correlation coefficients are statistically significant after downscaling a The abbreviations Obs, Sim and Ds signify the observed, simulated and downscaled data, respectively the precipitation data as indicated in Table 4. Greater values of the error index in this Table (compare to Table 3) are mostly due to the bigger absolute values as well as the complexity and non-linearity in the precipitation data as they compared with temperature series. indicated in this Table, measure of the increase or decrease in precipitation data is site specific and varies from about (-21.28 %) to (?42.80 %) of the base period. The next two sections are devoted for further discussion about the projected series. 3.2 Future Projection 3.3 Temperature Analysis By plugging the 2016–2045 simulated data into the already developed regression equations (i.e., between observed and downscaled values), the projected values of temperature and precipitation were obtained for this period. Table 5 and 6 summarizes the average values of temperature and precipitation data after downscaling. In these Tables, monthly values were averaged into annual or 6 monthly data for temperature or precipitation, respectively. According to the given statistics, regardless of the applied scenarios, a widespread warmer condition is anticipated for southern Iran during the next three decades. Precipitation, however, showed suppression or enhancement according to either 20C3M or 1PTO2X scenario is used, respectively (Table 6). Furthermore, as According to the low emission scenario of 20C3M, the mean annual temperature of the considered station for the period 2016–2045 will be warmer than the corresponding values of observed records by about 0.5–1.0 °C in different stations. This temperature rise, however, jumps to about 2–2.4 °C if simulation is conducted under the high emission scenario of 1PTO2X. Since these statistics are the mean values for three decades, some years will, therefore, be warmer than the presented mean values. As an example, Fig. 3a, b illustrate the historical as well as projected fluctuations of January temperature data for Shiraz station according to the 20C3M and 1PTO2X scenarios, respectively. While the historical data are identical in both figures, the projected parts (i.e., data after 2005) are 123 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 45 Table 4 Like Table 3, but for precipitation data during the October–March period Station name 20C3M 1PTO2X Raw data R2 DS data RMSE R2 Raw data R2 RMSE DS data RMSE R2 RMSE Abadan 0.07 3.19 0.96 3.02 0.01 4.60 0.92 4.57 Ahwaz 0.07 7.69 0.95 1.72 0.01 13.70 0.92 3.26 Bandar Abass Bandar Daier 0.05 0.14 4.19 4.32 0.93 0.90 3.95 3.82 0.26 0.32 6.52 11.26 0.96 0.91 4.35 4.49 Mahshahr 0.01 11.43 0.94 6.87 0.001 15.44 0.91 8.05 Boushehr 0.04 7.81 0.91 5.78 0.22 10.05 0.81 6.32 Fasa 0.05 9.95 0.98 0.66 0.01 7.99 0.92 1.49 Jask 0.03 25.60 0.91 7.99 0.09 22.02 0.71 13.56 Abomoosa Island 0.10 11.82 0.95 11.01 0.02 19.64 0.89 12.73 Kangan Jam 0.09 22.41 0.92 10.88 0.08 31.02 0.94 9.45 Lar 0.09 34.96 0.92 14.88 0.22 51.69 0.86 19.37 Shiraz 0.12 11.66 0.97 9.02 0.11 22.92 0.92 4.54 Table 5 Mean annual air temperature data for the observed records (Obs) and projected (20C3M or 1PTO2X) scenarios Station Annual mean temperature Obs 20C3M (°C) Diff. 1PTO2X (°C) Diff. Abadan 25.32 25.90 0.58 27.37 2.05 Ahwaz 26.12 26.76 0.64 28.10 1.98 Bandar Abass 26.99 27.68 0.69 28.99 2.00 Bandar Daier 27.09 27.60 0.51 29.04 1.95 Mahshahr 25.89 26.59 0.70 28.31 2.42 Boushehr 24.53 25.39 0.86 26.64 2.11 Fasa Jask 20.28 27.51 20.84 28.47 0.56 0.96 22.51 29.83 2.23 2.32 Abomoosa Island 27.98 28.56 0.58 30.14 2.05 Kangan Jam 20.46 21.07 0.61 22.55 2.16 Lar 25.77 26.34 0.57 27.88 2.11 Shiraz 17.90 18.55 0.65 20.25 2.35 The positive value of differences (Diff) indicate that the projected air temperatures are warmer than historical data in all stations the downscaled records. The slope of trend lines for the projected part (2016–2045) was found to be around zero and 0.016 °C per year for these two scenarios, respectively (not shown). These statistics are less than the slope if the 55 years of historical data are used that is around 0.45 °C. Therefore, in spite of the fact that the mean global projected values are consistently greater than the mean historical temperature, their trends are less for the projected data particularly for 20C3M scenario. This suggests the weakness of the GCMs for precise prediction of meteorological variables in chronological sequence as was discussed earlier. Our examination indicated that the warming trend in historical data was consistently increased for recent periods. A sequential regression analysis for the period 1951–2005 with 25 years of window width was performed to detect successive slopes of the relevant trend lines (Fig. 4). This figure exhibits continued trend for the periods of 1951–1975, 1952–1976 and eventually 1981–2005 (-0.026, -0.025, and eventually ?0.06 °C per year, respectively). As indicated, the warming trend is positive and much steeper for the recent periods supporting the hypothesis that the current increasing trend could be continued during the projected period. In other words, it is reasonable if someone assumes that the slope of trend line during the 25 years period of 2021–2045 is greater than 0.06 °C per year. A shift from a negative to a steep positive trend was found for all stations with more than 40 years of records. It is worthwhile to note 123 46 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 Table 6 Mean monthly values of the observed precipitation records (Obs) and projected series for the period 2016–2045 (mm/month) Station Mean precipitation Obs 20C3M % Diff. 1PTO2X % Diff. Abadan 22.99 22.98 -0.04 23.07 Ahwaz 34.55 33.21 -3.87 35.12 1.66 Bandar Abass 27.95 20.35 -27.19 31.86 13.96 Bandar Daier 40.27 32.40 -19.53 44.07 9.44 Mahshahr 33.28 23.11 -30.56 43.90 31.92 Boushehr 44.48 39.50 -11.19 46.55 4.65 Fasa Jask 46.87 20.42 44.01 16.07 -6.10 -21.28 48.12 29.15 2.67 42.80 Abomoosa Island 19.16 15.10 -21.20 23.63 23.29 Kangan Jam 63.37 52.76 -16.74 67.66 6.76 Lar 31.89 25.91 -18.75 36.18 13.45 Shiraz 51.61 44.47 -13.83 62.08 20.30 0.35 0.0800 21 19 y = 0.0448x + 16.728 R² = 0.7917 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 2011 2016 2021 2026 2031 2036 Mean annual temperature (c) 23 15 (b) Fig. 3 Fluctuations of mean annual temperature in past and future period from: a 20C3M scenario and b 1PTO2X scenario for Shiraz. For the 1951–2005 time period, the slope of trend line is 0.035. When the data of 2016–2045 added to the applied time series, the slope changes into 0.016 or 0.0448 under the 20C3M or IPTO2X scenario, respectively. The difference between 0.035 and these two slopes are smaller for the latter rather than former scenario that the increasing trend during the recent 20 years of data in Shiraz is also about 0.06 °C/year. This suggests much steeper trend for this station during the 2012–2045 period. In this case, the mean temperature for the predicted data (by trend analysis) is about 2.5–3 °C greater than the corresponding value for the historical series that was projected by the 123 0.0400 0.0200 1975 1973 1971 1969 1965 1967 1963 1961 1959 1957 1955 -0.0400 1953 0.0000 -0.0200 Start year of window width Fig. 4 Sequential values of the slopes of the trend lines for the regression analysis between time and Shiraz temperature for the period 1951–2005 using 25 years of window width. The given values in horizontal axis delineate the start year of the 25 year period considered in the regression analysis (a) 17 0.0600 1951 y = 0.0164x + 17.365 R² = 0.33 Trend-line's slope 21 20 19 18 17 16 15 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 2011 2016 2021 2026 2031 2036 Mean annual temperature (c) Precipitation analysis was conducted for October–March period. The differential (Diff) values are computed by this formula [100 9 (a-b)/a], where a and b are the mean values of the observed and projected precipitation series, respectively. 20C3M and 1PTO2X are the names of climate scenarios 1PTO2X scenario. In spite of some differences, most of the slope values in the considered stations were greater than 0.035 °C per year for all of 25 years window widths after 1964–1989 periods. As indicated in Fig. 3, if the projected values of the 1PTO2X scenario are the continuation of the historical data, the correlation determination is much stronger than if the values are from the 20C3M scenario. The slope of the trend line is also more realistic if this high emission scenario is used. Based on the provided evidence and for the purpose of future planning and decision making, accepting the outcomes of 1PTO2X scenario or steep warming trend is a reasonable choice. Figure 5 depicts the concurrent variations of the observed and projected monthly temperature data for all station according to 20C3M and 1PTO2X scenarios. As indicated, greater or lower temperature rise are mostly associated to the warmer or colder months of the year, Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 47 Ahwaz Month 20.00 Month 40.00 DEC NOV SEP JULY Month Bushehr Bandar Mahshahr Bandar Daier JUNE APR MAY FEB JAN NOV DEC SEP 0.00 OCT Observed 20C3M scenario 1PTO2X scenario 10.00 OCT JUNE APR MAY FEB MAR DEC NOV SEP 0.00 30.00 AUG Observed 20C3M scenario 1PTO2X scenario 10.00 OCT JUNE APR MAY FEB MAR JAN 0.00 AUG 10.00 20.00 AUG Observed 20C3M scenario 1PTO2X scenario JAN 20.00 30.00 JULY 30.00 MAR 40.00 Bandar Abass 40.00 Temperature (C) 40.00 Temperature (C) 50.00 JULY Temperature (C) Abadan 50.00 40.00 50.00 DEC NOV OCT DEC NOV OCT SEP Month Shiraz Lar 40.00 JUNE APR MAY FEB JAN DEC NOV OCT SEP 0.00 AUG Observed 20C3M scenario 1PTO2X scenario 10.00 Month Kangan Jam JUNE APR 20.00 MAR Month AUG APR MAY MAR FEB JAN DEC NOV 0.00 JULY Observed 20C3M scenario 1PTO2X scenario JULY 20.00 10.00 OCT SEP JUNE APR MAY MAR AUG Observed 20C3M scenario 1PTO2X scenario JUNE 20.00 Temperature (C) 30.00 Temperature (C) 30.00 FEB MAY Jazireh Abomoosa 30.00 JAN MAR Jask 40.00 0.00 FEB JAN DEC NOV OCT SEP AUG Month 40.00 JULY 40.00 50.00 Month 20.00 DEC NOV OCT SEP AUG JUNE APR MAY MAR FEB DEC NOV SEP 0.00 JULY Observed 20C3M scenario 1PTO2X scenario 10.00 OCT APR MAY FEB 0.00 JAN DEC NOV OCT 10.00 AUG Observed 20C3M scenario 1PTO2X scenario 30.00 JAN 20.00 MAR Month SEP JUNE APR MAY MAR FEB JAN 0.00 AUG 10.00 JULY Observed 20C3M scenario 1PTO2X scenario 30.00 JULY Temperature (C) 20.00 Temperature (C) 40.00 30.00 JUNE Temperature (C) JULY JUNE APR MAY MAR FEB JAN DEC NOV Month Fassa Temperature (C) 0.00 40.00 10.00 Observed 20C3M scenario 1PTO2X scenario 10.00 0.00 Month 20.00 SEP Observed 20C3M scenario 1PTO2X scenario AUG 20.00 30.00 JULY 30.00 10.00 OCT JUNE APR MAY MAR FEB JAN 0.00 SEP 10.00 AUG Observed 20C3M scenario 1PTO2X scenario Temperature (C) Temperature (C) 20.00 JULY Temperature (C) 40.00 30.00 Month Fig. 5 Comparison between the observed and projected monthly temperature data for all stations. The GCM scenario’s names are depicted inside the figures respectively. In other words, the rate of increase in summer (July–September) temperature will be steeper than that for cold seasons. An acute increase in the regional potential evapotranspiration and water consumption is, therefore, predicted during dry and hot months of the years. In addition to monthly analysis for each station (i.e., Fig. 5), the mean monthly values of the projected temperature of all twelve stations were also computed. These monthly values were then transformed to seasonal series by averaging three monthly values. The difference between the projected and historical values was then calculated to estimate the measure of regional climate change in seasonal scale. According to the given results, based on the low emission scenario of 20C3M, the study area will be warmer than the historical mean by about 0.9, 0.29, 0.57 and 0.44 °C during winter, spring, summer and autumn, respectively. These values, correspondingly, will increase to 1.7, 2.64, 3.51 and 1.54 °C, if the high emission rate scenario of 1PTO2X is under consideration. As mentioned earlier, these values are more sensible than the outcome of 20C3M scenario. 123 48 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 3.4 Precipitation Analysis Table 6 summarizes the mean values of historical and projected precipitation data for all stations. As indicated, the mean values of the projected precipitation are consistently less than their corresponding historical statistics from about zero to 35 % if the 20C3M scenario is used. The projected deficiencies are, however, mostly less than 20 %. On other hand, mean precipitation increases from about 0–43 % if climate change is within the range of 1PTO2X scenario. However, out of twelve considered stations, only four sites precipitation enhancement is above 15 %. This implicates no substantial development in water resources even precipitation according to the high emission scenario. Overall, the projected increase in precipitation can hardly compensate the ever-increasing of water demand over the dry zones of southern Iran which also faces a substantial warming trend. Since the study area is vulnerable to prolong drought and precipitation abnormality, the projected temperature and precipitation data suggest a considerable amount of temperature rise and serious water shortage over the study area. While both scenarios predict warmer condition for the next few decades, a significant increase in regional precipitation is not anticipated. Moreover, the probable precipitation deficiency (particularly under the low emission scenario) will severely affect the regional climate condition. Nazemosadat et al. (2006) have analyzed concurrent variations of the southern oscillation index (SOI) and Iran’s precipitation for the period 1951–1999 seeking changepoint years in the considered series. They reported that the mid 1970s are the most probable changepoint years in the SOI data. The frequency and intensity of El Niño/La Niña events have increased/decreased during 1975–1999 as it compared with the former 25 years of 1951–1974 period. Consistent with SOI data, precipitation records in Iran have also shown significant change around the year 1975. Compared to the 1951–1975 period, a significant precipitation enhancement was evident for the period of 1975–1999; the era with more negative SOI data. This means that, in addition to local geographical attributes, increase or decrease in the considered precipitation data is significantly associated to the climate condition over the ocean tropics. Due to such dependencies and because of inconsistency between the results of the applied scenarios, significant enhancement or suppression in mean annual precipitation for the period 2016–2045 is in question. Although climate condition during the next few decades depends critically on temperature and precipitation status at that time, current policies on soil, water and air management are also important on the upcoming climate conditions. For instance, agricultural activities over the study area are essentially dependent on the groundwater 123 condition. The recent rapid depletion in the water table restricts these activities over thousands of hectares of farmlands. Reduction in cultivated and pasture areas, in turn, will increase the local temperature due to the lack of evapotranspiration. In other words, in addition to radiative climate forcing, the abuse of local water and soil resources exacerbate the impact of climate change in regional scale that might not be fully considered in dynamical climate modeling. 4 Conclusions This study has been prompted to project future climate change in southwestern parts Iran for the period of 2016–2045. Monthly, seasonal and annual time series of temperature and precipitation were considered as the climate indicators. Two scenarios of the ECHAM5 GCM model, namely 20C3M and 1PTO2X were used for projecting temperature and precipitation condition during the next three decades. The adapted statistical downscaling procedure transformed the probability distribution of the historical time series into the projected data sets. The poor correlations between the observed and simulated data were dramatically improved as the model outputs were downscaled. The correlations were, however, stronger for the Gaussian distributed temperature data than for the precipitation series. The given results suggest that the annual temperature in southern parts of Iran will increase by about 2.0–2.4 °C during the period 2016–2045, if climate changes according to the high emission scenario of 1PTO2X condition. This range of temperature rise was found to be more realistic than the 0.5–1.0 °C warmer condition that was projected by the low emission scenario of 20C3M. Higher or lower increase rate is anticipated for warm or cold months of the year, respectively. While the low emission scenario of 20C3M predicted less than climatological mean precipitation for southern Iran, precipitation enhancement is anticipated according to the 1PTO2X scenario. This enhancement is, however, mostly less than 15 % of the mean precipitation, suggesting no substantial improvement in accessible water resources even on the basis of this scenario. Due to a sharp increase in temperature trend, a higher value of water demand is expected for domestic use and agricultural activities. References Almazroui M (2012) Dynamical downscaling of rainfall and temperature over the Arabian Peninsula using RegCM4. Clim Res 52:49–62 Iran. J. Sci. Technol. Trans. Sci. (2016) 40:39–49 Christensen JH, Hewitson B, Busuioc A, Chen A (2007) Regional climate projections in climate change 2007: the physical science basis. Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, New York, p 996 Dettinger MD, Cayan DR, Meyer MK, Anne EJ (2004) Simulated hydrological responses to climate variations and change in the Merced, Carson, and American river basin, Sierra Nevada, California, 1900–2099. Clim Change 62:283–317 Ding Y, Houghton JT (2001) Climate change 2001: the scientific basis. Contribution of working group 1 to the third assessment report of the intergovernmental panel on climate change. Cambridge University Press. Cambridge, New York, p 881 Fei Z (2014) ENSO variability simulated by a coupled general circulation model: ECHAM5/MPI-OM. Atmos Ocean Sci Lett 7(5):471–475 Gates WL, Boyle JS, Covey C, Dease CG, Doutriaux CM, Drach RS, Fiorino M, Gleckler PJ, Hnilo JJ, Marlais SM, Phillips TJ, Potter GL, Santer BD, Sperber KR, Taylor KE, Williams DN (1999) An overview of the results of the atmospheric model intercomparison project (AMIP I). Bull Am Meteorol Soc 80:29–55 Goddard L, Mason SJ, Zebiak SE, Ropelewski CF, Basher R, Cane MA (2001) Current approaches to seasonal-to-interannual climate predictions. Int J Climatol 21:1111–1152 Guo Y, Jie C, Wei H (2012) Simulation of the modern summer climate over greater mekong sub-region (GMS) by ECHAM5RegCM3. Proc Eng 31:807–816 Hunt BG (1997) Prospects and problems for multi-seasonalpredictions: some issues arising from a study of 1992. Int J Climatol 17:134-154 Joubert AM, Hewitson BC (1997) Simulating present and future climate changes of southern Africa using general circulation models. Prog Phys Geogr 21:51–78 Landman WA, Goddard L (2002) Statistical recalibration of GCM forecasts over southern Africa using model output statistics. J Clim 15:2038–2055 Landman WA, Tennant WJ (2000) Statistical downscaling of monthly forecasts. Int J Climatol 20:1521–1532 Landman WA, Mason SJ, Tyson PD, Tennant WJ (2001) Retroactive skill of multi-tiered forecasts of summer rainfall over southern Africa. Int J Climatol 21:1–19 Lin CY, Chua YJ, Sheng YF, Hsu HH, Cheng CT, Lin YY (2014) Altitudinal and latitudinal dependence of future warming in Taiwan simulated by WRF nested with ECHAM5/MPIOM. Int J Climatol. doi:10.1002/joc.4118 Mason SJ, Joubert AM (1997) Simulated changes in extreme rainfall over southern Africa. Int J Climatol 17:291–301 Mason SJ, Goddard L, Graham NE, Yelaeva E, Sun L, Arkin PA (1999) The IRI seasonal climate prediction system and the 1997/98 El Niño event. Bull Am Meteorol Soc 80:1853–1873 Meehl GA, Covey C, Delworth T, Latif M (2007) The WCRP CMIP3 multimodel dataset: a new era in climate change research. Bull Am Meteorol Soc 88:1383–1394 Mestre-Sanchis F, Feijoo-Bello ML (2009) Climate change and its marginalizing effect on agriculture. Ecol Econ 68(3):896–904 Mohammadi H, Azizi GH, Mazaheri D, Rabbani F (2015) Simulation of rice production under climate change scenarios in the Southern coasts of Caspian Sea. Desert 20(2):197–206 49 National Research Council (2001) Climate change: an analysis of some key questions. National Academy Press, Washington, D.C., p 42 Nazemosadat MJ, Ghaedamini H (2010) On the relationships between the Madden–Julian oscillation and precipitation variability in Southern Iran and the Arabian Peninsula: atmospheric circulation analysis. J Clim 23:887–904 Nazemosadat MJ, Ghasemi AR (2004) Quantifying the ENSO-related shifts in the intensity and probability of drought and wet periods in Iran. J Clim 17:4005–4018 Nazemosadat MJ, Samani N, Barry DA, Molaii Niko M (2006) ENSO forcing on climate change in Iran: precipitation analyses. Iran J Sci Technol 30:47–61 O’Brien TP, Sornette D, McPherron RL (2001) Statistical asynchronous regression: determining the relationship between two quantities that are not measured simultaneously. J Geophys Res 106:13247–13259 Palmer TN, Anderson DLT (1994) The prospects for seasonal forecasting—a review paper. Q J R Meteorol Soc 120:755–793 Parry ML, Canziani OF, Palutikof JP, Van der Linden PJ, Hanson CE (2007) IPCC fourth assessment report in climate change 2007: impacts, adaptation and vulnerability. Contribution of working group II to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, p 976 Pryor SC, Schoof JT, Barthelmie RJ (2005) Empirical downscaling of wind speed probability distributions. J Geophys Res. doi:10. 1029/2005JD005899 Raje D, Mujumdar PP (2009) A conditional random field–based downscaling method for assessment of climate change impact on multisite daily precipitation in the Mahanadi basin. Water Resour Res. doi:10.1029/2008WR007487 Reboita MS, Rocha RP, Dias CG, Ynoue RY (2014) Climate projections for South America: RegCM3 driven by HadCM3 and ECHAM5. Adv Meteorol. doi:10.1155/2014/376738 Roeckner E, Bäuml G, Bonaventura L, Brokopf R, Esch M, Giorgetta M, Hagemann S, Kirchner I, Kornblueh L, Manzini E, Rhodin A, Schlese U, Schulzweida U, Tompkins A (2003) The atmospheric general circulation model ECHAM5. Part I: Simulated climatology and comparison with observations. Report No. 349, MaxPlanck-Institute for Meteorology, Hamburg Salathe EPJ, Mote PW, Wiley MW (2007) Review of scenario selection and downscaling methods for the assessment of climate change impacts on hydrology in the United States Pacific Northwes. Int J Climatol 27:1611–1621 U.S. National Assessment Synthesis Team (2000) Climate change impacts on the United States: the potential consequences of climate variability and change. Overview Report of the National Assessment Synthesis Team, US Global Change Research Program, Washington, DC Washington R, Preston A (2006) Extreme wet years over southern Africa: role of Indian Ocean sea surface temperatures. J Geophys Res. doi:10.1029/2005JD006724 123