Climate predictions made by Global Circulation Models (GCMs) are excellent tools for assessing the possible impacts of climate change on water resources. However, their coarse spatial resolution is insufficient to support basin scale hydro-logic applications. Downscaling methods for GCM outputs have been proposed which can be classified as: statistical and dynamical downscaling methods. The studies focused on climate predictions over Turkey utilized the latter ones, in which regional climate models (RCMs) are nested within a GCM.
GCMs provide information on the response of the atmosphere to different scenarios of global GHG emissions. Among the standard GHG emission scenarios defined by IPCC, A2 and B2 scenarios are the most commonly used by the climate change studies focusing on Turkey. The A2 scenario reflects a somewhat pessimistic view of future with a very heterogeneous world, continuously increasing population and slow technological change. The B2 scenario, foresees moderate global population growth and intermediate levels of economic development and technological change. These scenarios possibly reflect the upper limits of human-induced global warming.
Until now, only a few studies utilized RCMs to investigate the potential role of global warming on the future climate of Turkey. Among others, Onol and Semazzi  and Dalfes et al.  used the regional climate model RegCM3 (30 km horizontal resolution) forced by the general circulation model NASA-fvGCM to simulate the future climate change projections (2071-2100) over the Eastern Mediterranean (EM) based on A2 emission scenario. Their projections have shown general warming over EM (2-8°C). Over Turkey, the increase in summer temperatures is predicted to be more pronounced (5-7°C) in western Turkey with a prolonged summer season, whereas the increase in winter temperatures is predicted to be more pronounced in the eastern parts. Their simulations predicted a significant increase (10-50%) in winter precipitation along the east coast of the Black Sea whereas significant decrease (30%) in winter precipitation was predicted across western and southern Turkey. Note that these predictions are in line with the past observed trends (Sect. 12.2).
Fujihara et al.  investigated the potential impacts of climate change on the Seyhan River Basin (southern Turkey). They dynamically downscaled the hydrome-teorologic variables to 8.3 km using the TERC-RAMS regional model forced by two GCMs utilizing A2 emission scenario. The RCM output was also used to drive a hydrology model after bias correction. They concluded that the Seyhan Basin will be significantly impacted by 2070s, with the drastic decrease in annual runoff (52-61%) due to an increase in temperature (2.0-2.7°C) and a decrease in precipitation (2529%). They however concluded that water scarcity will not be an issue in the future, given that the water demand stays constant. In a parallel study, Tezcan et al.  used the same downscaled RCM output used by Fujihara et al.  as input to the models simulating surface flow, groundwater flow and salt water intrusion processes. Their predictions indicated significant decrease in precipitation (~30%), streamflow (~40%), groundwater recharge (~25%) and snow storage and deterioration in water quality due to sea water intrusion along the coastal zone by 2070s.
Focusing on western Turkey, Ozkul  assessed possible impacts of global warming on Gediz and Buyuk Menderes River basins. The assessment was based on the model for the Assessment of GHG Induced Climate Change (MAGICC; v.4.1) coupled with a regional climate change scenario generator (SCENGEN). Using this model setup, Ozkul  simulated temperature and precipitation changes in 2030, 2050, and 2100 based on A2-ASF and B2-MESSAGE emission scenarios. The predictions indicated a mean annual temperature increase of 1.2°C, 2°C, 4.4°C for years 2030, 2050 and 2100. For the same years, mean annual precipitation was predicted to decrease by 5%, 10% and 24% respectively. Ozkul  also investigated the impact of these changes on runoff using a simple downscaling method (alpha method) and a simple monthly water balance model. The runoff predictions indicated that, by the years 2030, 2050 and 2100, the surface waters in the basin will be reduced by 20%, 35% and 50% respectively. Aksoy et al.  studied climate variability and change in the European part of Turkey for the twenty-first century. In the procedure they used MAGICC/SCENGEN software to generate outputs from global climate models HadCM2 and ECHAM4 based on A2 and B2 emission scenarios and a simple downscaling method to estimate localized estimates of future climate. Their analyses indicated that the study area will be under the effect of increasing temperature and decreasing precipitation during the twenty-first century with higher variability and wider range reflecting vulnerability of the region to extreme events, floods, and droughts in the twenty-first century.
To summarize, climate modeling studies performed at the regional scale over Eastern Mediterranean and at selected basins projected an increase in temperatures and a decrease in precipitation and runoff over much of the country, especially in the western and southern parts. It should be noted that all projections made by climate models depend on various assumptions, including the model structure and trajectory of future global greenhouse gases. Hence the uncertainty in model projections of future climate should be explicitly incorporated in climate change studies. The most common uncertainty assessment method has been the ensemble approach which requires a collection of simulations from different climate models, with varying assumptions of the future atmospheric composition and with different initial conditions .
Was this article helpful?