• Two dynamical downscaling approaches were applied to model the atmosphere. • The reinitialized approach significantly shortens computation time. • Both the continuous and reinitialized methods yield comparable outcomes. • Neither technique accurately captures pressure and wind speed in high-altitude regions. General Circulation Models provide comprehensive climate projections but are limited by their coarse spatial resolution. Regional Climate Models are therefore used to obtain higher-resolution simulations for assessing regional climate change impacts through dynamical downscaling. This is typically performed using continuous simulations over a selected period. Alternatively, short, frequently reinitialized simulations can be applied to reduce error accumulation and computational cost via parallelization, although they may limit the development of some atmospheric processes. In this study, the WRF model was applied using both continuous and daily reinitialized downscaling approaches. Simulations were driven by ERA5 and CMIP6 data at spatial resolutions of 1° and 1.25°, respectively, over a domain spanning 115°W–40°E and 20°N–60°N, and downscaled to 20 km. Model outputs were evaluated against ERA5 data at 0.25° resolution. The analysis considered wind speed, air temperature, humidity, precipitation, surface pressure, and solar radiation. Both downscaling techniques showed good to excellent agreement with the reference data, though neither reliably reproduced wind speed or surface pressure in mountainous regions. In ERA5-driven simulations, the reinitialized approach performed better for air temperature and humidity in coastal areas, while the continuous method slightly improved solar radiation estimates. For CMIP6-driven simulations, results were largely similar, except for marginally better solar radiation performance over land with the continuous approach. The reinitialized method required approximately 30 times less computational time while maintaining comparable accuracy, strongly supporting its recommendation as the preferred approach when both performance and efficiency are considered. High-resolution climate information is a key component of climate services that support policy making, risk management, and sectoral planning. Regional climate simulations are widely used by public agencies and practitioners to assess climate variability, extremes, and future change at spatial scales relevant for decision making. However, the computational cost associated with producing such datasets often constrains their operational use, limiting spatial coverage, ensemble size, update frequency, or the range of variables that can be delivered to users. This study provides practical guidance on how these limitations can be reduced by comparing two dynamical downscaling strategies commonly used in climate services: continuous and daily reinitialized simulations. The results show that daily reinitialized dynamical downscaling can generate climate information with accuracy comparable to that of continuous simulations, while reducing computational time by approximately a factor of 30. This efficiency gain has direct implications for climate service providers, including national meteorological services, climate data centers, and publicly funded research infrastructures. By adopting a reinitialized approach, these institutions can substantially expand their modeling capabilities without increasing computational resources, enabling larger ensembles, higher spatial resolution, broader domains, or more frequent updates of climate products. For most atmospheric variables relevant to climate services—such as air temperature, humidity, precipitation, solar radiation, and wind speed—both downscaling approaches reproduce large-scale climatological patterns and variability well. Importantly for practitioners, the reinitialized simulations do not introduce systematic discontinuities in time series when an adequate spin-up period is used. This ensures that the resulting datasets are suitable for applications requiring temporal consistency, including the calculation of climate indicators, assessment of trends, and analysis of extreme or compound events. The study also highlights limitations that are relevant for practical use. Both downscaling strategies show reduced reliability for near-surface wind speed and surface pressure in mountainous regions. Users applying downscaled climate data in complex terrain, for example in infrastructure design, hazard assessment, or renewable energy planning, should therefore interpret these variables with caution and consider complementary methods such as bias adjustment, ensemble interpretation, or the use of additional data sources. Overall, this study supports the use of daily reinitialized dynamical downscaling as a practical and efficient approach for many climate service applications. By maintaining comparable accuracy while dramatically reducing computational cost, this method can enhance the scalability, accessibility, and sustainability of climate services, ultimately improving the delivery of actionable climate information to policy makers and practitioners.
Thomas et al. (Wed,) studied this question.