Top Banner
1 First results and methodological approach to parameter perturbations in GEM-LAM simulations PART II Leo Separovic, Ramon de Elia and Rene Laprise
12

Leo Separovic, Ramon de Elia and Rene Laprise

Dec 30, 2015

Download

Documents

nadine-snyder

First results and methodological approach to parameter perturbations in GEM-LAM simulations PART II. Leo Separovic, Ramon de Elia and Rene Laprise. Reduction of domain size. Integral. 58. 140x140. 55. 15. 70x92. - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Leo Separovic, Ramon de Elia and Rene Laprise

1

First results and methodological approach to parameter perturbations

in GEM-LAM simulationsPART II

Leo Separovic, Ramon de Elia and Rene Laprise

Page 2: Leo Separovic, Ramon de Elia and Rene Laprise

2

Reduction of domain size

140x140

70x92

100x100

~5500km 30x52

• Integral

• Core only

55 15

58

Core of the small domain and the common region used to compute grid statistics in the following

Page 3: Leo Separovic, Ramon de Elia and Rene Laprise

3

GZ850

DJF MMA JJA SON DJF MMA JJA SON

P -P +

• Statistical significance: RMS (t) over the small-domain core area(-) wklcl=0.020, (+) wklcl=0.048

Page 4: Leo Separovic, Ramon de Elia and Rene Laprise

4

GZ850

Error

P -P +

DJF MMA JJA SON DJF MMA JJA SON

• Error magnitude: 2xRMS standard deviation of the ensemble average

• Signal magnitude: RMSD between reference and perturbed-parameter ensemble averages; (-) wklcl=0.020, (+) wklcl=0.048

Page 5: Leo Separovic, Ramon de Elia and Rene Laprise

5

90.0

95.0

97.5

99.0

99.5

99.9

99.9

99.5

99.0

97.5

95.0

90.0

GZ850: WKLCL(-), JJAInternal variability

signal

SL

SL

SL

Rejection level for H0

%

Page 6: Leo Separovic, Ramon de Elia and Rene Laprise

6

PCP

DJF MMA JJA SON DJF MMA JJA SON

P -P +

• Statistical significance: RMS (t) over the small-domain core area(-) wklcl=0.020, (+) wklcl=0.048

Page 7: Leo Separovic, Ramon de Elia and Rene Laprise

7

PCP

DJF MMA JJA SON DJF MMA JJA SON

Error

P -P +

• Error magnitude: 2 x RMS standard deviation of the ensemble average• Signal magnitude: RMSD between reference and perturbed-parameter

ensemble averages; (-) wklcl=0.020, (+) wklcl=0.048

Page 8: Leo Separovic, Ramon de Elia and Rene Laprise

8

90.0

95.0

97.5

99.0

99.5

99.9

99.9

99.5

99.0

97.5

95.0

90.0

PCP: WKLCL(+), JJAInternal variability

signal

SL

SL

SL

Rejection level for H0

%

Page 9: Leo Separovic, Ramon de Elia and Rene Laprise

9

TTscn

DJF MMA JJA SON DJF MMA JJA SON

P -P +

• Statistical significance: RMS (t) over the small-domain core area(-) wklcl=0.020, (+) wklcl=0.048

Page 10: Leo Separovic, Ramon de Elia and Rene Laprise

10

TTscn

DJF MMA JJA SON DJF MMA JJA SON

Error

P -P +

• Error magnitude: 2 x RMS standard deviation of the ensemble average

• Signal magnitude: RMSD between reference and perturbed-parameter ensemble averages; (-) wklcl=0.020, (+) wklcl=0.048

Page 11: Leo Separovic, Ramon de Elia and Rene Laprise

11

90.0

95.0

97.5

99.0

99.5

99.9

99.9

99.5

99.0

97.5

95.0

90.0

TTscn: WKLCL(-), JJAInternal variability

signal

L

SL

SL

SL

Rejection level for H0

%

Page 12: Leo Separovic, Ramon de Elia and Rene Laprise

12

Future work

• Add an intermediate domain;

• Repeat the study for singleton perturbations of the threshold relative humidity for the onset of precipitation;

• Study synchronous perturbations of the two parameters;

• (Non)linearity of the model response to multiple perturbations and domain size.

Alexandru et al. (2007)