Downscaling

Downscaling is any procedure to infer high-resolution information from low-resolution variables. This technique is based on dynamical or statistical approaches commonly used in several disciplines, especially meteorology, climatology and remote sensing. The term downscaling usually refers to an increase in spatial resolution, but it is often also used for temporal resolution. This is not to be confused with image downscaling which is a process of reducing an image from a higher resolution to a lower resolution. Global Climate Models (GCMs) used for climate studies and climate projections are typically run at spatial resolutions of the order of 150 to 200 km and are limited in their ability to resolve important sub-grid scale features such as convection clouds and topography. As a result, GCM based projections may not be robust for local impact studies. To overcome this problem, downscaling methods are developed to obtain local-scale weather and climate, particularly at the surface level, from regional-scale atmospheric variables that are provided by GCMs. Two main forms of downscaling technique exist. One form is dynamical downscaling, where output from the GCM is used to drive a regional, numerical model in higher spatial resolution, which therefore is able to simulate local conditions in greater detail. The other form is statistical downscaling, where a statistical relationship is established from observations between large scale variables, like atmospheric surface pressure, and a local variable, like the wind speed at a particular site. The relationship is then subsequently used on the GCM data to obtain the local variables from the GCM output. Wilby and Wigley placed meteorological downscaling techniques into four categories: regression methods, weather pattern-based approaches, stochastic weather generators, which are all statistical downscaling methods, and limited-area modeling (which corresponds to dynamical downscaling methods). Among these approaches regression methods are preferred because of their relative ease of implementation and low computation requirements.