Providing Regional Climate Services to British Columbia

You are here

You want how many climate variables?! Towards multivariate bias correction and downscaling to support climate-impact modellers

Dr. Alex Cannon
February 27, 2019 - 3:00pm to 4:00pm

Room 002, University House 1, UVic
2489 Sinclair Rd.
Victoria , BC
University map 
Google Map

Climate change impacts and adaptation studies often involve running environmental models using meteorological variables obtained from climate model simulations. For example, a distributed hydrological model may require inputs of surface air temperature, pressure, wind, humidity, precipitation, and incoming shortwave and longwave radiation. Climate models are not perfect and their historical simulations typically exhibit systematic biases relative to the observational datasets used to calibrate an impact model. In practice, outputs from climate models are bias corrected – systematic errors in statistical properties are removed – prior to their use in subsequent modelling. Most bias correction algorithms, for example quantile mapping methods, are applied to univariate time series. They neglect the multivariate dependence that exists between different variables. Recent studies have demonstrated that this neglect can have substantial consequences on future projections of environmental processes that depend on multiple, inter-dependent variables.

This talk describes how an image processing technique designed to transfer colour information from one image to another – the N-dimensional probability density function transform – has been adapted for use as a multivariate bias correction algorithm (MBCn) for climate model projections of multiple climate variables. The method is demonstrated on three case studies. First, MBCn is applied to an image processing example with characteristics that mimic a climate projection problem. Second, MBCn used to correct a suite of surface meteorological variables from the Canadian Centre for Climate Modelling and Analysis Regional Climate Model. Components of the Canadian Forest Fire Weather Index System, a set of multivariate indices that characterizes the risk of wildfire, are then calculated and verified against observed values. Third, MBCn is used to correct biases in RCM temperatures and precipitation amounts used as inputs to a hydrological model for alpine watersheds. In all cases, the multivariate approach provides demonstrable benefits over standard univariate methods. A discussion of potential pitfalls of bias correction, whether univariate or multivariate, will be interwoven throughout the talk.