Event Title

Bias-correcting global climate model data: Can we maintain physical consistency and reduce future climate projection uncertainty at the same time?

Presenter Information

Jonathan Meyer
Jiming Jin

Location

Eccles Conference Center

Event Website

http://water.usu.edu

Start Date

4-1-2014 5:25 PM

End Date

4-1-2014 5:30 PM

Description

In this study, we focus on bias-correction techniques of global climate model (GCM) data provided by the Community Climate System Model version 4 (CCSM4). Bias correction of GCM data is necessary when dynamically downscaling GCM data in order to reduce uncertainty in the boundary conditions provided to a regional climate model (RCM). Previous GCM bias-correction efforts have focused on removing systematic bias for specific variables independently from other variables. In doing so, a breakdown in the physical consistency amongst the atmospheric variables occurs. When projecting climate scenarios into the future, physically inconsistent boundary conditions act to compound the model uncertainties while reducing confidence in the model projection. To minimize uncertainties introduced through the boundary conditions, our focus is establishing a step-by-step process that reduces bias and therefore uncertainty while conserving the physical consistency amongst the climate variables to a large extent. To accomplish this, the key atmospheric variables and surface variables such as temperature and humidity are corrected using simple linear regression techniques and the global, observation-based Climate Forecast System Reanalysis (CFSR). Using physical equations, these corrected variables are then used to re-build the remaining variables such as geopotential height and wind needed to drive an RCM. In addition to presenting our step-by-step correction process, we also compare the CFSR and the CCSM4 datasets over the western United States to illustrate where biases exist. Lastly, we highlight the impact of biases between CCSM4 and CFSR datasets when used to drive an RCM over the western United States.

This document is currently not available here.

Share

COinS
 
Apr 1st, 5:25 PM Apr 1st, 5:30 PM

Bias-correcting global climate model data: Can we maintain physical consistency and reduce future climate projection uncertainty at the same time?

Eccles Conference Center

In this study, we focus on bias-correction techniques of global climate model (GCM) data provided by the Community Climate System Model version 4 (CCSM4). Bias correction of GCM data is necessary when dynamically downscaling GCM data in order to reduce uncertainty in the boundary conditions provided to a regional climate model (RCM). Previous GCM bias-correction efforts have focused on removing systematic bias for specific variables independently from other variables. In doing so, a breakdown in the physical consistency amongst the atmospheric variables occurs. When projecting climate scenarios into the future, physically inconsistent boundary conditions act to compound the model uncertainties while reducing confidence in the model projection. To minimize uncertainties introduced through the boundary conditions, our focus is establishing a step-by-step process that reduces bias and therefore uncertainty while conserving the physical consistency amongst the climate variables to a large extent. To accomplish this, the key atmospheric variables and surface variables such as temperature and humidity are corrected using simple linear regression techniques and the global, observation-based Climate Forecast System Reanalysis (CFSR). Using physical equations, these corrected variables are then used to re-build the remaining variables such as geopotential height and wind needed to drive an RCM. In addition to presenting our step-by-step correction process, we also compare the CFSR and the CCSM4 datasets over the western United States to illustrate where biases exist. Lastly, we highlight the impact of biases between CCSM4 and CFSR datasets when used to drive an RCM over the western United States.

https://digitalcommons.usu.edu/runoff/2014/2014Posters/20