Assessment of Climate Change Effects on Shahcheraghi Reservoir Inflow

Introduction: Forecasting the inflow to the reservoir is important issues due to the limited water resources and the importance of optimal utilization of reservoirs to meet the need for drinking, industry and agriculture in future time periods. In the meantime, ignoring the effects of climate change on meteorological and hydrological parameters and water resources in long-term planning of water resources cause inaccuracy. It is essential to assess the impact of climate change on reservoir operation in arid regions. In this research, climate change impact on hydrological and meteorological variables of the Shahcheragh dam basin, in Semnan Province, was studied using an integrated model of climate change assessment. Materials and Methods: The case study area of this study was located in Damghan Township, Semnan Province, Iran. It is an arid zone. The case study area is a part of the Iran Central Desert. The basin is in 12 km north of the Damghan City and between 53° E to 54° 30’ E longitude and 36° N to 36° 30’ N latitude. The area of the basin is 1,373 km2 with average annual inflow around 17.9 MCM. Total actual evaporation and average annual rainfall are 1,986 mm and 137 mm, respectively. This case study is chosen to test proposed framework for assessment of climate change impact hydrological and meteorological variables of the basin. In the proposed model, LARS-WG and ANN sub-models (7 sub models with a combination of different inputs such as temperature, precipitation and also solar radiation) were used for downscaling daily outputs of CGCM3 model under 3 emission scenarios, A2, B1 and A1B and reservoir inflow simulation, respectively. LARS-WG was tested in 99% confidence level before using it as downscaling model and feed-forward neural network was used as raifall-runoff model. Moreover, the base period data (BPD), 1990-2008, were used for calibration. Finally, reservoir inflow was simulated for future period data (FPD) of 2015-2044 and compared to BPD. The best ANN sub-model has minimum Mean Absolute Relative Error (MARE) index (0.27 in test phases) and maximum correlation coefficient (ρ) (0.82 in test phases). Results and Discussion: The tested climate change scenarios revealed that climate change has more impact on rainfall and temperature than solar radiation. The utmost growth of monthly rainfall occurred in May under all the three tested climate change scenarios. But, rainfall under A1B scenario had the maximum growth (52%) whereas the most decrease occurred (–21.5%) during January under the A2 climate change scenario. Rainfall dropped over the period of June to October under the three tested climate change scenarios. Furthermore, in all three scenarios, the maximum temperature increased about 2.2 to 2.6°C in May but the lowest increase of temperature occurred in January under A2 and B1 scenarios as 0.3 and 0.5°C, respectively. The maximum temperature usually increased in all months compared to the baseline period. Minimum and maximum temperatures enlarged likewise in all months, with 2.05°C in September under A2 climate change scenario. Conversely, solar radiation change was comparatively low and the most decreases occurred in February under A1B and A2 climate change scenarios as –4.2% and –4.3% , respectively, and in August under the B1 scenario as –4.2%. The greatest increase of solar radiation occurs in April, November, and March by 3.1%, 3.2%, and 4.9% for A1B, A2, and B1 scenarios, respectively. The impact of climate change on rainfall and temperature can origin changes on reservoir inflow and need new strategies to adapt reservoir operation for change inflows. Therefore, first, reservoir inflow in future period (after climate change impact) should be anticipated for the adaptation of the reservoir. A Feed-Forward (FF) Multilayer-Perceptron (MLP) Artificial Neural Network (ANN) model was nominated for the seven tested ANN models based on minimization of error function. The selected model had 12 neurons in the hidden layer, and two delays. The comparison of forecasted flow hydrograph by selecting an ANN model and observed one proved that forecasted flow hydrograph can follow observed one closely. By comparison with the IHACRES model, this model displayed a 54% and 46% lower error functions for validation data. The selected model was used to forecast flow for the climate change scenarios of the future period. Conclusions: The results show a reduction of monthly flow in most months and annual flow in all studied scenarios. The following main points can be concluded: • By climate change, flow growths in dry years and it declines in wet and normal years. • The studied climate change scenarios showed that climate change has more impact on rainfall and temperature than solar radiation.