Pipeline to identify dominant features in spatial data

Dominant-feature identification decomposes spatial data into several additive components to make different features apparent on each component. It recognizes their dominant features credibly and assesses feature attributes. This paper describes the pipeline to apply this method to regular and irregular lattice data as well as geostatistical data. These implementations are all openly available and templates for each case are provided in an associated git repository. As geostatistical data is typically large, we propose several efficient approximations suitable for such data. Emphasizing the use of these approximations in the context of dominant-feature identification, we apply them to data from a climate model describing the monthly mean diurnal range for the period between the years 2081 and 2100.