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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Using <strong>the</strong> NMSE statistical performance metric, <strong>the</strong> CMAQ vertical diffusion scheme performs<br />

best with OB70 <strong>and</strong> TKE producing very similar results next, with <strong>the</strong> ACM2 exhibiting <strong>the</strong> worst<br />

NMSE performance results (Figure 6‐18a, top right). The PPM horizontal advection scheme is<br />

performing slightly better than <strong>the</strong> Bott algorithm based on <strong>the</strong> NMSE metric.<br />

The CMAQ vertical diffusion scheme is also <strong>the</strong> best performing method according to <strong>the</strong> FB<br />

metrics followed by <strong>the</strong> TKE <strong>the</strong>n ACM2 <strong>and</strong> <strong>the</strong>n OB70 in last. According to <strong>the</strong> FB metrics,<br />

PPM performs slightly better than Bott.<br />

For <strong>the</strong> KS parameter, <strong>the</strong> OB70 is <strong>the</strong> best vertical mixing method with CMAQ barely beating<br />

out ACM2 in second <strong>and</strong> TKE slightly worse. The PPM horizontal advection solver is performing<br />

slightly better than Bott for <strong>the</strong> KS parameter.<br />

For <strong>the</strong> within a factor <strong>of</strong> 2 <strong>and</strong> 5 metrics (FA2 <strong>and</strong> FA5, Figure 6‐18b, top), <strong>the</strong> CMAQ <strong>and</strong> TKE<br />

vertical mixing approaches are clearly performing better than <strong>the</strong> OB70 <strong>and</strong> ACM2 methods<br />

<strong>and</strong> <strong>the</strong> PPM horizontal advection solver is clearly performing better than Bott. For <strong>the</strong> FA2,<br />

<strong>the</strong> TKE/PPM is <strong>the</strong> best performing configuration (11.4%) followed by CMAQ/PPM (10.9%),<br />

Whereas for <strong>the</strong> FA5 <strong>the</strong> reverse is true with CMAQ/PPM being <strong>the</strong> best performing<br />

configuration (23.4%) followed by TKE/PPM (22.2%).<br />

There is essentially no difference in <strong>the</strong> PCC statistic using <strong>the</strong> two horizontal advection solvers<br />

(Figure 6‐18b, bottom right). According to <strong>the</strong> PCC metric, CMAQ is <strong>the</strong> best performing<br />

vertical diffusion approach (0.52) followed by TKE (0.37 <strong>and</strong> 0.38), OB70 (0.35) <strong>and</strong> ACM2 (0.26<br />

<strong>and</strong> 0.27).<br />

The final panel in Figure 6‐18b (bottom right) displays <strong>the</strong> overall RANK statistic. The RANK<br />

statistics orders <strong>the</strong> model performance <strong>of</strong> <strong>the</strong> CAMx configurations without PiG as follows:<br />

1. CMAQ/PPM (1.94)<br />

2. CMAQ/Bott (1.90)<br />

3. TKE/PPM (1.70)<br />

4. OB70/PPM (1.66)<br />

5. TKE/Bott (1.65)<br />

6. ACM2/PPM (1.60) (tied)<br />

7. OB70/Bott (1.60) (tied)<br />

8. ACM2/Bott (1.54)<br />

Based on this analysis <strong>the</strong> CMAQ Kz coefficients is <strong>the</strong> best performing vertical diffusion<br />

approach followed by TKE <strong>and</strong> <strong>the</strong> PPM horizontal advection algorithm is performing slightly<br />

better than Bott. The vertical diffusion algorithm has a greater effect on CAMx model<br />

performance compared to <strong>the</strong> choice <strong>of</strong> horizontal advection solvers.<br />

126

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

Saved successfully!

Ooh no, something went wrong!