Modelling for Climate Variability
Conventional approaches to environmental Modelling are based on the coupling of a dynamical model and “delta” scenarios of projected changes in mean climate. This standard practice has the advantage of the dynamical simulation of the biophysical and hydrological processes, but the disadvantage of requiring large amounts of data to calibrate and validate the model, limiting the domain to a few decades in terms of range of variability and extremes. Therefore the model outputs generally are restricted to projections of changes in mean states between past and future decades. Reconstructions of the climate and hydrology of the past millennium reveal fluctuations at time scales (multi-decadal) that exceed the length of most instrumental records. This scale of variability is important for our understanding of the stationarity of the regional climate regime, and for natural resource planning and management for extreme events in terms of magnitude to duration. We model streamflow as a function of the ocean-atmosphere oscillations (teleconnection indices) that drive the natural variability of the regional hydroclimatic regime. We then drive these regression models using output from an ensemble of global climate models (GCMs) that simulate spectral and geographic characteristics of relevant teleconnection patterns – the El Niño Southern Oscillation and Pacific Decadal Oscillation. This Modelling approach captures the shift in climate variability that is forced by a warming oceans and atmosphere.
