20.04.2013 Views

Documentation of the Evaluation of CALPUFF and Other Long ...

Documentation of the Evaluation of CALPUFF and Other Long ...

Documentation of the Evaluation of CALPUFF and Other Long ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Figure C‐ ‐15. Global model perfoormance<br />

staatistics<br />

for thhe<br />

CAMx sennsitivity<br />

testts<br />

without uusing<br />

<strong>the</strong> PiG ssubgrid‐scale<br />

e puff moduule<br />

for CAPTTEX<br />

Release 5 (NoPiG).<br />

C.3.7 SPATIAL<br />

PERFO ORMANCE FFOR<br />

CTEX5 PPIG<br />

EXPERIMMENTS<br />

As with t<strong>the</strong><br />

CTEX5 No oPiG experimments,<br />

interpretation<br />

<strong>of</strong>f<br />

<strong>the</strong> spatial performancce<br />

for CTEX5 for<br />

<strong>the</strong> PiG CCAMx<br />

sensiti ivities posedd<br />

a slightly mmore<br />

difficultt<br />

challenge tto<br />

interpret ddue<br />

to<br />

similarities<br />

amongst <strong>the</strong> Kz/adveection<br />

solverr<br />

options forr<br />

<strong>the</strong> FMS meetric<br />

(Figure C‐16). The<br />

range <strong>of</strong> difference for f <strong>the</strong> FMS bbetween<br />

<strong>the</strong>e<br />

minimum a<strong>and</strong><br />

maximuum<br />

for all <strong>of</strong> <strong>the</strong> eight<br />

combinations<br />

was less<br />

than 2%, with all in thhe<br />

range <strong>of</strong> 221.5%<br />

‐ 23.33%,<br />

noting a slight<br />

degradattion<br />

across <strong>the</strong><br />

board froom<br />

<strong>the</strong> correesponding<br />

NNoPiG<br />

experiments<br />

(0.2% % ‐ 3.1%).<br />

Examinattion<br />

<strong>of</strong> <strong>the</strong> extended e spaatial<br />

statisticcs<br />

reveals a ssimilar<br />

patteern<br />

in perforrmance<br />

compareed<br />

to <strong>the</strong> NoP PiG equivaleent<br />

tests. Grreater<br />

differences<br />

in perrformance<br />

are<br />

observed<br />

across thhe<br />

various Kz z/advection solver combbinations,<br />

especially<br />

for t<strong>the</strong><br />

POD <strong>and</strong>d<br />

TS metrics.<br />

For<br />

both <strong>of</strong> <strong>the</strong>se<br />

metrics,<br />

<strong>the</strong> CMAQQ<br />

Kz combinations<br />

clearlly<br />

yield betteer<br />

spatial peerformance<br />

tthan<br />

<strong>the</strong> o<strong>the</strong>r<br />

Kz options (5% ‐ 10% bbetter<br />

for POOD<br />

<strong>and</strong> 2% ‐ 5% for TS thhan<br />

<strong>the</strong> secoond<br />

best Kz<br />

option (TTKE)).<br />

Consistennt<br />

with results<br />

from <strong>the</strong> NoPiG scenarios,<br />

<strong>the</strong> CMMAQ<br />

Kz option<br />

appears to perform best<br />

across thhe<br />

majority <strong>of</strong> o <strong>the</strong> spatial<br />

metrics. The<br />

largest diifference<br />

in spatial perfoormance<br />

is<br />

determinned<br />

by <strong>the</strong> user’s<br />

selection<br />

<strong>of</strong> <strong>the</strong> Kz option thann<br />

ei<strong>the</strong>r <strong>the</strong> choice <strong>of</strong> addvection<br />

solvver<br />

or<br />

use <strong>of</strong> <strong>the</strong><br />

subgrid sc cale PiG moddule<br />

in CAMxx.<br />

19

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!