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 ...
statistic for indicating overall model performance of a LRT dispersion model for the ETEX evaluation. 2.4 2.0 1.6 1.2 0.8 0.4 0.0 CALPUFF SCIPUFF Rank (RANK) (Perfect = 4) HYSPLIT Figure ES‐6. RANK statistical performance metric for the five LRT models and the ETEX tracer field experiment. Table ES‐7. Summary of ETEX model ranking using the eleven ATMES‐II statistical performance metrics and their average rankings that are compared against the rankings by the RANK composite model performance metric. Statistic 1 st 2 nd 3 rd 4 th 5 th FMS CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF FAR HYSPLIT FLEXPART CAMx SCIPUFF CALPUFF POD CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF TS CAMx HYSPLIT SCIPUFF FLEXPART CALPUFF FOEX CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF FA2 CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF FA5 CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF NMSE HYSPLIT CAMx CALPUFF FLEXPART SCIPUFF PCC or R SCIPUFF HYSPLIT CAMx FLEXPART CALPUFF FB HYSPLIT CAMx CALPUFF FLEXPART SCIPUFF KS CAMx SCIPUFF HYSPLIT FLEXPART CALPUFF Avg. Ranking CAMx HYSPLIT SCIPUFF FLEXPART CALPUFF Avg. Score 1.55 2.27 2.73 3.82 4.64 RANK Ranking CAMx HYSPLIT SCIPUFF FLEXPART CALPUFF RANK Score 1.9 1.8 1.8 1.0 0.7 Spatial Displays of Model Performance Figures ES‐7 and ES‐8 display the spatial distributions of the predicted and observed tracer concentrations 36 and 60 hours after the beginning of the ETEX tracer release. CALPUFF advects the tracer too far north keeping a circular Gaussian plume distribution and fails to 24 FLEXPART CAMx (1‐KS/100) FMS/100 (1‐FB/2) R^2
eproduce the northwest to southeast diagonal orientation of the observed tracer cloud. The other four LRT dispersion models do a much better job in reproducing the observed tracer cloud spatial distribution. SCIPUFF tends to overestimate the tracer cloud extent and surface concentrations. FLEXPART, on the other hand, underestimates the observed tracer cloud spatial extent and CAMx and HYSPLIT do the best job overall in reproducing the spatial extent of the observed tracer cloud. Figure ES‐7. Comparison of spatial distribution of the ETEX tracer concentrations 36 hours after release for the observed (top left), CALPUFF (top right), SCIPUFF (middle left), FLEXPART (middle right), HYSPLIT (bottom left) and CAMx (bottom right). 25
- Page 1 and 2: Documentation of the Evaluation of
- Page 3 and 4: FOREWARD This report documents the
- Page 5 and 6: 2.4.3 ATMES‐II Model Evaluation A
- Page 7 and 8: EXECUTIVE SUMMARY ABSTRACT The CALP
- Page 9 and 10: OVERVIEW OF APPROACH Up to six LRT
- Page 11 and 12: CALPUFF performance is evaluated by
- Page 13 and 14: Table ES‐2. ATMES‐II spatial an
- Page 15 and 16: • CALPUFF tended to overstate the
- Page 17 and 18: • The best performing CALPUFF con
- Page 19 and 20: 2009a). The key findings from the C
- Page 21 and 22: 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2
- Page 23 and 24: 2.4 2 1.6 1.2 0.8 0.4 0 EXP4A EXP4B
- Page 25 and 26: Figure ESS‐4. RANK statistical pe
- Page 27 and 28: Evaluatioon of Six LRT T Dispersion
- Page 29: Table ES‐6. Summary of model rank
- Page 33 and 34: ETEX LRT Dispersion Model Sensitivi
- Page 35 and 36: CONCLUSIONS OF LRT DISPERSION MODEL
- Page 37 and 38: The CAMx and CALGRID Eulerian photo
- Page 39 and 40: July 1980. Both experiments examine
- Page 41 and 42: 1.3 ORGANIZATION OF REPORT Chapter
- Page 43 and 44: puffs expand until they exceed the
- Page 45 and 46: that performance evaluation be base
- Page 47 and 48: The ETEX real‐time LRT modeling p
- Page 49 and 50: The ETEX study has formulated the f
- Page 51 and 52: In this study we expand the LRT mod
- Page 53 and 54: AM ∩ AP FMS = × 100% (2‐2) A
- Page 55 and 56: Factor of α (FAα): FAα represent
- Page 57 and 58: 3.0 1980 GREAT PLAINS FIELD STUDY 3
- Page 59 and 60: compact discs, which were used to o
- Page 61 and 62: ILEVZI = 1 Layer of winds to use in
- Page 63 and 64: MCHEM = 0 No chemical transformatio
- Page 65 and 66: Table 3‐6. CALPUFF/CALMET experim
- Page 67 and 68: Table 3‐11. CALPUFF/MMIF sensitiv
- Page 69 and 70: evaluation studies and evaluate whe
- Page 71 and 72: Tables 3‐13 and Figures 3‐2 thr
- Page 73 and 74: 140% 120% 100% 80% 60% 40% 20% 0%
- Page 75 and 76: 30% 20% 10% 0% ‐10% ‐20% ‐30%
- Page 77 and 78: 120% 100% 80% 60% 40% 20% 0% ‐20%
- Page 79 and 80: 20% 10% 0% ‐10% ‐20% ‐30% ‐
statistic for indicating overall model performance <strong>of</strong> a LRT dispersion model for <strong>the</strong> ETEX<br />
evaluation.<br />
2.4<br />
2.0<br />
1.6<br />
1.2<br />
0.8<br />
0.4<br />
0.0<br />
<strong>CALPUFF</strong><br />
SCIPUFF<br />
Rank (RANK) (Perfect = 4)<br />
HYSPLIT<br />
Figure ES‐6. RANK statistical performance metric for <strong>the</strong> five LRT models <strong>and</strong> <strong>the</strong> ETEX tracer<br />
field experiment.<br />
Table ES‐7. Summary <strong>of</strong> ETEX model ranking using <strong>the</strong> eleven ATMES‐II statistical performance<br />
metrics <strong>and</strong> <strong>the</strong>ir average rankings that are compared against <strong>the</strong> rankings by <strong>the</strong> RANK<br />
composite model performance metric.<br />
Statistic 1 st 2 nd<br />
3 rd<br />
4 th 5 th<br />
FMS CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
FAR HYSPLIT FLEXPART CAMx SCIPUFF <strong>CALPUFF</strong><br />
POD CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
TS CAMx HYSPLIT SCIPUFF FLEXPART <strong>CALPUFF</strong><br />
FOEX CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
FA2 CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
FA5 CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
NMSE HYSPLIT CAMx <strong>CALPUFF</strong> FLEXPART SCIPUFF<br />
PCC or R SCIPUFF HYSPLIT CAMx FLEXPART <strong>CALPUFF</strong><br />
FB HYSPLIT CAMx <strong>CALPUFF</strong> FLEXPART SCIPUFF<br />
KS CAMx SCIPUFF HYSPLIT FLEXPART <strong>CALPUFF</strong><br />
Avg. Ranking CAMx HYSPLIT SCIPUFF FLEXPART <strong>CALPUFF</strong><br />
Avg. Score 1.55 2.27 2.73 3.82 4.64<br />
RANK Ranking CAMx HYSPLIT SCIPUFF FLEXPART <strong>CALPUFF</strong><br />
RANK Score 1.9 1.8 1.8 1.0 0.7<br />
Spatial Displays <strong>of</strong> Model Performance<br />
Figures ES‐7 <strong>and</strong> ES‐8 display <strong>the</strong> spatial distributions <strong>of</strong> <strong>the</strong> predicted <strong>and</strong> observed tracer<br />
concentrations 36 <strong>and</strong> 60 hours after <strong>the</strong> beginning <strong>of</strong> <strong>the</strong> ETEX tracer release. <strong>CALPUFF</strong><br />
advects <strong>the</strong> tracer too far north keeping a circular Gaussian plume distribution <strong>and</strong> fails to<br />
24<br />
FLEXPART<br />
CAMx<br />
(1‐KS/100)<br />
FMS/100<br />
(1‐FB/2)<br />
R^2