Assessment of climate change impacts on annual maximum daily runoff distribution via continuous stream-flow simulation

Recent human activities have raised average temperature of the Earth’s surface. This increase will also influence climate variables and will result in climate change. With changing climate variables such as temperature and precipitation, the hydrologic regime of the rivers, and consequently flood frequency and magnitude, will also change (IPCC, 2001; Khazaei at al., 2012). For active adaptation strategy, it is necessary to assess impacts of climate change on floods. It is frequently mentioned in the literature that one of the potential impacts of climate change is the change on floods, however limited studies aimed at investigation of the change of flood regime due to CC. Generally, climate change impact assessment on floods comprises two main steps: preparing future climate data and hydrological simulation of river-flow. For hydrological simulation of flood flows, hydrologic models can be classified in two main classes, including single-event models and continuous models. In some studies, single-event rainfall-runoff models have been used (e.g. Roy et al., 2001; Muzik, 2002). In such studies, the reason for using these models was simplicity. In spite of the ease in using single-event models, the initial soil moisture condition in the future changed condition is unknown. Due to high sensitivity of simulated floods to initial condition of soil moisture, application of single-event models deals with great uncertainty (Roy et al., 2001). So, for assessment of climate change impact on floods, continuous models are required (Prudhomme et al., 2002). For climate change impact assessment on floods using continuous hydrologic models, normally, watershed-scale time series with fine time step for future climate are needed (Prudhomme et al., 2003). The most common tools to simulate future climate scenarios are GCMs. However, since resolution of GCM outputs is course, it is necessary that GCM outputs be downscaled. At least, rainfall and temperature data should be downscaled for continuous hydrologic simulation, throughout it is important to preserve the correlation between the downscaled variables (Fowler et al., 2007). Among downscaling methods, Change Factors (CFs), Weather Generators (WGs), and RCMs preserve the correlation between downscaled variables. In general, RCMs and WGs cannot accurately reproduce extremes. So in most previous studies CFs method have been used (e.g. Reynard et al., 2001; Prudhomme et al., 2002, 2003; Mareuil et al, 2007; and Kay et al., 2009).
In this paper, the climate change impacts on floods were assessed in one of the main sub-basins of the Karun basin. Rainfall-runoff process of the basin was simulated using ARNO semi-distributed continuous rainfall-runoff model (Todini 1988). The main phenomena which represented in the ARNO model are: snow melt, water losses through evapotranspiration, soil moisture balance, groundwater flow, overland and channel flow routing. The rainfall-runoff model was calibrated and validated for the basin using eight years of high-quality daily data of stream-flow at the outlet, and daily precipitation, Tmax, and Tmin at Yasuj meteorological station, close to the centroid of the basin. For future climate scenarios, projections of the CGCM3 model under A2, A1B and B1 emission scenarios were used for both control (1974–2000) and future (2067–2093) periods. The climate scenarios were downscaled for the basin using the Change Factors (CFs) method. In CFs method, the differences between means of GCM outputs for control and future periods are applied to every baseline observed data series, either as summation (for temperature) or multiplication (for precipitation).
As result of rainfall- runoff modeling, the Efficiency Criterion (EC) of calibration and validation stages were turned out to be 0.87 and 0.83, respectively, also determination coefficients (R2) were 0.85 and 0.88, respectively; while the flood values were simulated closely. In comparison with other studies, Zhang and Savenije (2005) adopted acceptable calibration when EC is greater than 0.6. Meanwhile, Kamali et al. (2007) accepted EC values of greater than 0.7. Thus, the performance of the model in overall the hydrograph is very good. In order to produce future climate series, the CGCM3 projections for each of the A2, B1 and A1B scenarios were downscaled for the basin. With the future climate data and the rainfall-runoff model, daily stream-flow series at the future period were generated. Annual maxima daily floods were extracted from the daily stream-flow series and climate change impacts assessed on annual maximum floods distribution. Based on the results, the magnitude of the floods will increase in the changed future climate. For instance, under these scenarios, flood magnitude with a return period of 25 years will increase between 50% and 120% for period of 2067–2093, in comparison to 1974–2000 historic period. Also, flood magnitude with a return period of 100 years will increase between 54% and 126%, under B1, A1B, A2 scenarios. It is concluded that despite the magnitude of the change, which is upon the choice of the emission scenario, the climate change will impose considerable increase on floods under all the regarded scenarios.