Methodological approach to parameter perturbations in GEM-LAM simulations Leo Separovic, Ramon de Elia and Rene Laprise
MOTIVATION Sub-grid parameterization schemes describe: - well-known processes that can be exactly represented (e.g. radiation transfer) but need to be approximated so that they do not take excessive computational time; - less-well understood processes (e.g. turbulent energy transfer) that are situation dependent; parameters in such parameterizations rely on mixture of theoretical understanding and empirical fitting; Parameters’ values are uncertain due to measurement errors and problems with their “representativity”; hence the tuning cannot completely eliminate the model error. We develop methodological approach to quantify the parametric uncertainty in RCM seasonal simulations. To this end we study the response of GEM-LAM seasonal climate to multiple perturbations of parameters. The perturbations’ size are within a range of uncertainty specified by the experts that participated in the model tuning. GEM-LAM (0.5deg, ERA40) is ran over 1 year and the response is studied for each of the 4 seasons. For each combination of parameters a small ensemble with perturbed initial conditions is run in order to estimate the statistical significance of the response. - the following slide shows the changes in 2m temperature (signal) induced by two singleton perturbations of the threshold vertical velocity in Kain-Fritch convection; - the second slide shows the rejection level of the 0 th hypothesis that the difference is only due to sensitivity to initial conditions.
2m-Temperature (signal) DJFMAM JJASON DJF MAM JJASON KFCTRIG=0.020 (-)KFCTRIG=0.048 (+) K KFCTRIG(ref)=0.034
2m TEMP: rejection level DJFMAM JJASON DJF MAM JJASON KFCTRIG=0.020 (-)KFCTRIG=0.048 (+) % KFCTRIG(ref)=0.034