The primary goal of dynamical climate models is to create future projections, reconstruct past conditions, and perform numerical experiments. These models, whether global or regional, are most accurate at large scales, but their accuracy diminishes at smaller scales. This decline in skill is partly because noise becomes increasingly significant at smaller scales, and also because the models must inevitably truncate the dynamics, ignoring states that aren’t captured by the numerical discretization. However, many studies focusing on the impacts of climate variations and changes require an understanding of these smaller scales. As a result, there’s a trend toward refining the models to capture smaller and smaller scales, though this approach significantly increases computational costs. A more economical alternative is to identify the statistical or dynamical relationships between large- and small-scale states, a process known as “downscaling.” In this chapter, we explore the development of both empirical and dynamical downscaling methods for estimating consistent small-scale dynamics and assessing regional and local impacts. One of the technical challenges in this process is effectively distinguishing between small and large scales through appropriate filtering techniques.