An assessment of observed and projected temperature changes in Armenia
Future changes in annual and seasonal temperature over Armenia based on the Community Climate System Model 4 (CCSM4) output data from newly developed dataset of phase 5 of the Coupled Model Intercomparison Project (CMIP5) have been analysed. The results of this study suggest that Armenia will experience significant temperature increase in the twenty-first century. Moreover, the greatest warming is expected in the summertime and can reach 4–6 °C under representative concentration pathways (RCP)8.5 for the middle and end of the century. However, observations show significant variability and seasonal amplitude in temperature regime which is peculiar to mountain regions located in mid-latitudes. It was shown that strong cold wave events can be directly followed by prolonged hot days within one year in Armenia. The CCSM4 model has substantial deficiencies in simulating the regional climate over Armenia. Validation results indicate the largest errors (root-mean-square error (RMSE)) in the representation of winter temperatures. There is also significant uncertainty in projected temperature change patterns in Armenia over the twenty-first century. However, the CCSM4 model becomes more confident in the prediction of temperature increase over Armenia toward the end of the century.
KeywordsClimate change Armenia Temperature CCSM4 model
Recent studies considered climate change and its impacts on hydrological balance in Armenia and Southern Caucasus, The Black and Caspian Seas basins (Gevorgyan 2014; Melkonyan et al. 2013; Melkonyan and Shindyan 2009; Regional Climate Change Impacts Study for the South Caucasus Region 2011; SNCCC 2010; Zhang et al. 2005; Elguindi et al. 2011; Melkonyan and Asadoorian 2013). The results showed significant increasing trends for mean annual temperature, mean daily minimum temperature and mean daily maximum temperature (Vardanyan et al. 2013; Regional Climate Change Impacts Study for the South Caucasus Region 2011). Mean annual temperature have increased by rate of 0–1.5 °C for 1935–2008 period. A very clear signal of warming is detected at most stations of Armenia and at 850 hPa level for summer and autumn seasons (Gevorgyan 2014). Furthermore, the accelerated warming can be seen in the recent past period (1979–2012) which leaves little chance that the warming trends in Armenia are due solely to natural variability.
It is worth noting that a significant increase in temperature and in hot extreme temperature events has been observed over the Middle East which neighbours the study region from the south (Almazroui et al. 2012; Almazroui et al. 2013; Darand et al. 2015). The observed annual maximum, mean and minimum temperatures have increased significantly at a rate of 0.71, 0.60 and 0.48 °C decade−1, respectively, in Saudi Arabia over the period 1978–2002. Furthermore, the Middle East region is projected to be a climate change ‘hot spot’ (Giorgi 2006; Lelieveld et al. 2012). Results from Sharif (2015) indicate that projected temperature departures in Saudi Arabia are likely to exceed 3.5 and 5.5 °C for the mid and end of the twenty-first century, respectively, under A2 scenario. On the other hand, the projected temperatures from CCSM4 model for the July–August months showed temperature increase relative to ERA-Interim reanalysis data over most of the study region by more than 2.0 and 2.5 °C for the mid and end of the twenty-first century, respectively, under representative concentration pathways(RCP)4.5 scenario (Gevorgyan and Melkonyan 2014).
The main purpose of this study is to assess future changes in temperature in Armenia based on a newly developed dataset from phase 5 of the Coupled Model Intercomparison Project (CMIP5). Relative to CMIP3, CMIP5 constitutes an unprecedented set of experiments that include higher spatial resolution models and improved model physics. The results of this study are expected to provide a framework for agricultural adaptation and water resource management for Armenia in the future.
Data and method
To assess potential future changes in temperature over Armenia, monthly outputs from the Community Climate System Model 4 (CCSM4, National Center for Atmospheric Research) on a 1.25° × 0.9° grid are used in this study. The CCSM4 is considered as high-resolution model from CMIP5 GCMs including new physical parameterizations (cumulus convection scheme, a high-accuracy radiation scheme), new land models and aerosol effects on clouds (Gent et al. 2011; Chen et al. 2013). The CCSM4 was used in several recent studies (Wang and Chen 2013; Meleshko and Govorkova 2013; Sporyshev and Govorkova 2013; Pavlova and Kattsov 2013). Meleshko and Govorkova (2013) and Sporyshev and Govorkova (2013) employed 34 models out of CMIP5 to investigate model-performance error over the Northern Hemisphere and Russia quantified by the root-mean-square error (RMSE) and correlation coefficient, and CCSM4 was among the ten successful models in terms of temperature and precipitation representation over the mentioned areas. Changes in mean monthly 2-m maximum temperature from the CCSM4 model for July–August months have been analysed over the Armenian Highland and neighbouring regions by Gevorgyan and Melkonyan (2014) recently to assess the possible influence of regional temperature contrasts on heat-induced circulations under future climate conditions.
We also used the RMSE in its classical sense to validate the CCSM4 model ability in reproducing the historical climate. This statistical measure has been used by different authors (Meleshko and Govorkova 2013; Elguindi et al. 2013) to evaluate simulated temperature against observations both at the global and regional scales.
The two twenty-first century scenarios for future greenhouse gas emissions used in this study are RCP4.5 and RCP8.5 as defined in Moss et al. (2010). where RCP8.5 is a high-emission path and RCP4.5 assumes lower emissions. Only one ensemble member is chosen for this study identified as r1i1p1 for CCSM4 though multiple ensemble members are accessible. For temperature projection reasons, we consider the reference period 1961–1990 for present climate, while three additional time intervals of 30 years, 2011–2040 (the beginning of the century), 2041–2070 (the mid-century) and 2071–2100 (the end of the century) for future climate. Future changes in temperature in Armenia were estimated for the four climatological seasons, namely winter (DJF), spring (MAM), summer (JJA) and autumn (SON), as well as annual changes were analysed.
Observed temperature trends in Armenia
August of 2014 was ranked the second warmest in Armenia since 1961 (with a monthly temperature anomaly of 3.5 °C). However, Yerevan and southwestern low elevated parts of Armenia experienced prolonged series of hot days. Since the 16th till the 31st of August, maximum temperatures in Yerevan exceeded 36 °C with positive anomalies higher than 4 °C and reaching up to 7–10 °C for several days (Fig. 4b). Figure 4d shows that most part of Armenia experienced significant positive temperature anomalies in August of 2014 exceeding 3 °C. Monthly temperature anomalies at several stations reached as high as 4–6 °C.
Validation of historical climate
Model confidence is usually based on the evaluation of their performance at reproducing observed features of current climate. In this section, we evaluate the ability of the CCSM4 model to reproduce the historical climate over Armenia in terms of temperature representation. Seasonal and annual temperatures are constructed in Armenia for the historical period 1961–1990 derived from the observed data and the CCSM4 data (Fig. 2).
The observed data show the largest variability of temperature during winter (Fig. 5) with the temperature range around the observed mean equaling ±4.3 °C, while CCSM4-simulated temperature range is about 70 % of the observed one (±3.0 °C). The increased interannual variations in winter temperatures in Armenia are due to the great variability in atmospheric circulation regime during winter in the mid and high latitudes of the Northern Hemisphere (Sporyshev and Govorkova 2013). However, the CCSM4 historical data still show maximum variability of temperature during winter for 1961–1990. On the other hand, the CCSM4 model overestimates the temperature range for spring (±2.5 °C) relative to the observed one (±1.8 °C). The estimated confidence intervals of temperature derived from the observed and the CCSM4 model are in better agreement for summer and autumn, with the observed temperature range ±1.5 and ±1.7 °C, respectively, and with CCSM4 temperature range equalling ±1.9 °C for both of the seasons. Finally, the estimated ranges of annual temperature from observed data and the CCSM4 model consist of ±1.6 and ±1.3 °C, respectively.
Projected changes in temperature in Armenia
It is worth noting that there is a large interseasonal variation in terms of future warming. The magnitude of the temperature increase peaks in summertime in any period both under RCP4.5 and RCP8.5. This is consistent with the recent warming in Armenia obtained from observations (Fig. 3a–e). The latter suggests even stronger amplitude of the annual cycle of temperature in Armenia under future climate conditions leading to a stronger continentality of climate relative to current levels. This is more pronounced under RCP8.5 scenario showing that a temperature increase in summer can reach 6.0 °C for the end of century (Fig. 7c). Mean temperatures in winter, spring and autumn are expected to increase from 3.9 to 4.4 °C over the twenty-first century (Fig. 7c).
The estimated values of temperature ranges are of the same magnitude and even exceed respective temperature changes in most cases (Fig. 7a–c). The largest temperature range for period 2011–2040 (Fig. 7a) was obtained for winter (±2.7 °C), while those for the periods 2041–2070 (Fig. 7b) and 2071–2100 (Fig. 7c) were found during summer (±2.7 °C) and spring (±3.1 °C), respectively. In other words, there is great uncertainty in projected temperature change patterns in Armenia which can be partly due to the natural variability of seasonal temperatures discussed in Section 3.2. However, the projected temperature ranges are overestimated relative to those for present climate (Fig. 5) in most cases, except for winter and annual temperature ranges. Furthermore, the CCSM4 data do not show well-defined maximum variability in winter season as was obtained for present climate from observations (Fig. 5). It is worth noting that the temperature ranges for annual temperature changes are significantly reduced (from ±1.0 to ±1.7 °C) indicating less uncertainty in projected annual temperature changes relative those for the seasons. On average, the values of estimated temperature ranges for seasonal and annual mean temperatures are of about the same magnitude for all of the three considered periods of the twenty-first century (Fig. 7a–c) both under RCP4.5 scenario (±2.1 °C) and under RCP8.5 scenario (from ±2.2 to ±2.4 °C). With the continuous increasing temperature in the future, the latter suggests that the CCSM4 model is becoming more confident in the prediction of temperature increase over Armenia toward the end of the century. However, we should be careful with the temperature change estimates for winter and summer seasons, since these seasons are characterized by higher RMSE values (Fig. 6).
Conclusions and discussions
In this study, changes in one of the main climate variable, i.e. temperature, in Armenia up to the end of the century are analysed. The manuscript is the first to present climate forecasting over a little studied part of the world. It should be noted that the results of future temperature changes from the CCSM4 model are in line with previous results based on regional climate model PRECIS developed by the Hadley Centre of the UK Meteorological Office (Regional Climate Change Impacts Study for the South Caucasus Region 2011; SNCCC 2010). The results of this study suggest that Armenia will experience significant temperature increase in the twenty-first century. Both recent temperature changes obtained from observed data and temperature change projections in the future obtained from the CCSM4 data show the greatest warming in Armenia during the summer. It is worth noting that the expected temperature increase in the summertime can reach 4–6 °C under RCP8.5 for the middle and end of the century. The above would lead to substantial negative impacts on agricultural, hydrological and socioeconomic sectors and would present significant adaptation challenges. It is expected that higher air temperatures in the warm season would result in an increase in evaporation and aridity. The latter will have negative impacts on the economy of Armenia which is highly dependent on the water sector.
Obviously, the CCSM4 model has substantial deficiencies in simulating the regional climate over Armenia. Validation results showed higher errors (RMSE) in the representation of winter and summer temperatures. These errors are probably due to the inadequacies in the representation of complex topography and the associated circulation over the study region. Observations show significant variability and seasonal amplitude in temperature regime which is peculiar to mountain regions located in mid-latitudes. It was shown that strong cold wave events can be directly followed by prolonged hot days within one year in Armenia.
There is also a great uncertainty in projected temperature change patterns in Armenia over the twenty-first century. However, the CCSM4 model becomes more confident in the prediction of temperature increase over Armenia toward the end of the century. To produce the reliable projections required for regional climate assessment in Armenia, we need to bias-correct the raw model outputs and to apply different downscaling techniques (e.g. nesting of regional models performed within the CORDEX activity). The consideration of different GCMs and the results for the multi-model ensembles is expected to be important to better understand temperature change in Armenia in the future. In particular, the selected GCMs should cover as much as possible the range of different responses of greenhouse gas (GHG) forcing by the full ensemble of available GCMs. The spatial resolution of models is considered as a very important issue for the simulation of temperature, precipitation, wind and other meteorological elements over the study region characterized by mountain topography as was shown both in this and previous works (Gevorgyan 2012; Gevorgyan and Melkonyan 2014; Elguindi et al. 2011). Apart from annual and seasonal temperature changes, it is also important to focus on the change in precipitation, extreme temperatures and precipitation on a daily scale. The above mentioned issues are associated with high computational costs and will clearly be a subject of interest for future study.
- Gevorgyan A (2012) Verification of daily precipitation amount forecasts in Armenia by ERA-Interim model. Int J Climatol 33:2706–2712Google Scholar
- Intergovernmental Panel on Climate Change (2007) Climate change 2007: the physical basis. In: Solomon S (ed) Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Cambridge Univ. Press, New YorkGoogle Scholar
- Lelieveld J, Hadjinicolaou P, Kostopoulou E, Chenoweth J, Giannakopoulos C, Hannides C, Lange MA, El Maayar M, Tanarthe M, Tyrlis E, Xoplaki E (2012) Climate change and impacts in the eastern Mediterranean and the Middle East. Clim Chang 114:667–687. doi: 10.1007/s10584-012-0418-4 CrossRefGoogle Scholar
- Meleshko VP, Govorkova VA (2013) Performance of CMIP3 and CMIP5 models in simulation of current climate. Trans Voeykov Main Geophys Obs 568:26–51Google Scholar
- Melkonyan H, Shindyan S (2009) Natural and anthropogenic climate change in Armenia. Trans Yerevan State Univ 3:31–37Google Scholar
- Melkonyan H, Ovsepyan A, Iritsyan A, Ye K, Gevorgyan A (2013) Climate change assessment in Armenia. Trans Inst Hydrometeorol Georgian Tech Univ 119:33–37Google Scholar
- Moss RH, Edmonds JA, Hibbard KA, Manning MR, Rose SK, van Vuuren DP, Carter TR, Emori S, Kainuma M, Kram T, Meehl GA, Mitchell JFB, Nakicenovic N, Riahi K, Smith SJ, Stouffer RJ, Thomson AM, Weyant JP, Wilbanks TJ (2010) The next generation of scenarios for climate change research and assessment. Nature 463:747–756. doi: 10.1038/nature08823 CrossRefGoogle Scholar
- Pavlova TV, Kattsov VM (2013) World ocean ice cover as simulated with CMIP5 models. Trans Voeykov Main Geophys Obs 568:7–26Google Scholar
- Regional climate change impacts study for the South Caucasus Region. 2011. TbilisiGoogle Scholar
- Second National Communication on Climate Change (SNCCC) (2010) Ministry of nature protection of Armenia. p.132Google Scholar
- Sporyshev PV, Govorkova VA (2013) Temperature changes in Russia according to observations and model simulations with a separate account of anthropogenic and natural external impacts. Trans Voeykov Main Geophys Obs 568:51–80Google Scholar
- Vardanyan L, Melkonyan H, Hovsepyan A (2013) Current status and perspectives for development of climate services in Armenia. Ministry of Emergency Situations of Republic of Armenia; 40Google Scholar
- Zhang X, Aguilar E, Sensoy S, Melkonyan H et al. (2005) Trends in Middle East climate extreme indices from 1950 to 2003. Geophys Res Lett 110. doi: 10.1029/2005JD006181