Evaluation of RegCM4 Regional Climate Model simulations for the land surface water budget components (A case study in Toyserkan plain, Hamedan province)

Background and objectives: In regional and local climate studies usually coarse-resolution outputs of global climate models are downscaled to produce necessary fine scale data. Regional climate modeling is a dynamical downscaling method. In this study HadGEM2 general circulation model outputs have been downscaled by RegCM4 Regional Climate Model coupled CLM4.5 land surface scheme. The model was run over Toyserkan plain for a reference period (1999-2005) for model evaluation and a projection period (2015-2025). In the projection period, regarding to the importance of water resources in the study area, water budget components and surface water balance equation have been evaluated.
Materials and methods: For the reference period, daily modeled precipitation, temperature and runoff were compared with observed values at available meteorological stations in the region. Modeling efficiency, correlation coefficient, bias, mean absolute error and root mean squared error statistical indices were used to evaluate model’s simulations. Modeled precipitation was compared with observations of 6 available stations in the region, Observed temperatures of the station of Hamedan- Airport were also used for modeled temperature verification; For runoff verification, the only river station at the outlet of the catchment was used. Projections are based on RCP4.5 scenario of the Fifth Assessment Report of Intergovernmental panel on climate change. For the projection period spatiotemporal variations of surface water budget components including precipitation, evapotranspiration and runoff have been studied and water balance equation in the catchment has been evaluated.
Results: Model evaluation results showed that the model has its worst performance for runoff, because of low modeling efficiency and relatively large errors while has its best performance in simulating precipitation (especially for the first five stations) and temperature. Finally, the model shows its best performance for temperature and precipitation respectively, regarding to more positive efficiency, higher correlation and smaller errors. For the first year of projection period (2015) the highest values of precipitation occur in eastern and central parts while the lowest values occur in southwestern part of the catchment. Through the next 10 years (2016-2025), precipitation decreases in most parts except small parts in east and south of the catchment. The highest values of precipitation decrease (about %12) also occur in northeastern and central parts. Since the amount of precipitation determines available moisture, spatial distribution and variation of evapotranspiration are same as those for precipitation. The highest values of runoff occur in eastern and northeastern (high elevation) parts and during 2016-2025 a %30-50 increase in eastern part and a %10-20 decrease in central part will occur. Annual water budget evaluation shows that for 7 years there is a water balance, but for the remaining 3 years (2017, 2023 and 2024) differences between two sides of the water balance equation are large.
Conclusion: Considering small area of the catchment and short time of projection period that does not allow to detect impacts of climate change, small amount of mean 10-year difference between two sides of the equation (1.3 mm) shows that the model performance in estimating water balance for the catchment is acceptable. Generally, although the water balance estimation has been improved, the land surface scheme has shortcomings in water budget parametrization especially for runoff. But regarding to improvement of mean temperature simulation, model has an appropriate performance in simulating the energy budget.