Climate change has a non-consistent effect on surface runoff and streamflow among different regions of the world. In the Midwestern, northwestern, and North Eastern USA, an increase in streamflow was observed as a response to climate change, whereas a reduction in streamflow was observed in the southern states of the USA (Romero-Lankao et al., 2014; Talib and Randhir, 2017). For instance, in the Texas High Plains, prolonged high temperature and low rainfall caused severe drought events in 2011 and 2012 (Hoerling et al., 2013; Ray et al., 2018, 2017). Climate change also triggers a change in the seasonal magnitude and timing of streamflow. In snow-dominated watersheds, an increase in temperature has resulted in a shift in the magnitude and timing of hydrological events (Bates et al., 2008; Kim et al., 2017; Ray et al., 2016; Romero-Lankao et al., 2014). In the snow-dominated rivers of western North America, snowmelt attributed to climate change trigger an early peak flow of runoff (Barnett et al., 2005; Das et al., 2011). These designated region and watershed-specific studies of climate change's effect on hydrological characteristics are essential.
Hydrological climate change impact assessment studies are prone to have considerable uncertainties attributed to greenhouse gas emission scenarios, climate models, downscaling and bias correction techniques, and hydrological models (IPCC, 2013; Kundzewicz et al., 2018). For this, it is non-trivial to use the output of robust climate change scenarios for climate change impact studies. Climate change scenarios are consistent and plausible representations of future climate conditions through blending emission scenarios, climate model simulations, and downscaling techniques (IPCC-TGICA, 2007; IPCC, 2013; Moss et al., 2010). In climate change scenarios, GCMs have limitations in simulating regional and local scale precipitation and cloud cover of the mountain and coastal regions due to low spatial resolutions and inadequate parametrization of regional scale drivers of climate (Flato et al., 2013; Randall et al., 2007). GCMs downscaling is a commonly used technique in climate research to improve the horizontal resolution of climate models and better parametrize the effect of factors such as topography on local and regional scale climate (Flato et al., 2013; Fowler et al., 2007; IPCC, 2013). On the other hand, downscaling of GCM simulations to RCMs is also an important source of uncertainty, and a caveat is needed before using the outputs of RCM simulations. It is commendable to use the ensemble of RCM simulations than the single RCM simulation to capture possible uncertainties stem from multiple sources (Flato et al., 2013; IPCC-TGICA, 2007; Teutschbein and Seibert, 2010).
Statistical bias correction techniques are essential in climate change scenario development to adjust biases in the downscaled climate model simulations and add value to hydrological impact assessment. For instance, surface runoff simulated using bias-corrected RCM simulations is more reliable than surface runoff simulated using raw RCMs simulation (Hagemann et al., 2011; Muerth et al., 2013). There are bias correction techniques such as linear scaling (Lenderink et al., 2007), which correct only mean values, while there are bias correction methods such as distribution mapping (Piani et al., 2010; Teutschbein and Seibert, 2012), which adjust the distribution of values at all quantiles. Using multiple statistical bias correction techniques captures the biases from RCMs parametrization schemes and bias correction algorithms (Teutschbein and Seibert, 2012; Pourmokhtarian et al., 2016).
Besides, robust hydrological models are essential to reduce uncertainties and develop climate change impact assessment (Baldassarre et al., 2011; Kundzewicz et al., 2018). Identifying sensitive hydrological parameters, optimization algorithms, input data, and best performance measures are essential to reduce uncertainties in hydrological models used for climate impact assessment (Bárdossy and Singh, 2008; Gan et al., 2018). The hydrological models calibrated using observed data, are used for future hydrological climate change impact studies. For this, calibration and validation using the Differential Split-Sample Test (DSST) approach are recommended to test hydrological models' performance under changing and even contrasted climate conditions (Daggupati et al., 2015; Guilpart et al., 2020; Huang et al., 2020; Klemeš, 1986). Further, calibration and validation at different temporal scales and areas of the basin and considering hydrological signatures in calibration and validation are important to reduce the uncertainty of hydrological projections (Huang et al., 2020; Melišová et al., 2020). The conceptual and parameterization structure of hydrological models could also trigger uncertainty on projected hydrological components (Poulin et al., 2011). Thus, caution is needed in selecting, calibrating, and validating hydrological models before using them for climate change impact assessment.
The effect of climate change on hydrological extremes also warrants special focus since extreme values are sensitive to climate change and could be impacted by uncertainties in climate modeling. Climate change amplifies low and high-flow change signals mainly attributed to changes in precipitation and temperature (De Girolamo et al., 2022; Kay et al., 2021; Marx et al., 2018). For instance, a greater decrease in low flows and an increase in high flows were projected across Great Britain until the end of the 21st century (Kay et al., 2021). Extreme hydrological events such as high flow frequency, extreme peak flow quantile, and extreme low flow quantile are characterized by uncertainty due to the uncertainties derived from climate data (Kay et al., 2021; Meresa et al., 2021). High uncertainty in precipitation of climate model simulation has triggered uncertainty in hydrological and extreme hydrological values in the Southeast Asian basins (Shrestha et al., 2021). It is apparent that changes in hydrological extremes events trigger more profound effects on the natural and anthropogenic ecosystems than changes in annual, seasonal, and monthly scales (Arnell, 2004; Taye and Willems, 2012).
The focus of this study is to evaluate the impact of climate change on hydrology and hydrological extremes in the Bosque watershed of Brazos River Basin, Northcentral Texas, where climate change already poses a negative impact on the natural ecosystem and water availability (Hoerling et al., 2013; Shafer et al., 2014). To reduce uncertainty, this study blends multiple emission scenarios, GCM simulations, statistical downscaling, and bias correction techniques and uses a robustly calibrated and validated hydrological model. This study can be essential to develop optimal water management and agriculture systems that help stakeholders to ensure sustainable water development and agricultural production.