Volume 1 - The Atmospheric Studies Group at TRC
Volume 1 - The Atmospheric Studies Group at TRC Volume 1 - The Atmospheric Studies Group at TRC
Figure 4-11. Graphical depiction of MG,VG model performance results for CALPUFF configurations with Modeled Iy for minimum σ v = 0.37 m/s (:37 labels) and 0.5 m/s (:50 labels). A -- Modeled Iy, CALPUFF Turb(z), Draxler Fy E -- Modeled Iy, AERMOD Turb(z), Draxler Fy C -- Modeled Iy, CALPUFF Turb(z), Variable TLy G -- Modeled Iy, AERMOD Turb(z), Variable TLy Final Report Vol.1 74
Performance of CALPUFF Configurations Using Standard COARE in CALMET The performance of all eight CALPUFF configurations tested are intercompared using runs made with the standard COARE option, denoted here as c10d, with a minimum σ v = 0.37 m/s. Results from Appendix A are collected and processed to form the geometric performance measures. OCD5 results are also included as a reference point. Results of applying the model evaluation software, including 95% confidence limits, are listed in Table 4-16 and Figure 4-12. Note that OCD5 with observed Iy produces a peak χ/Q less than 0.001 µs/m 3 in one of the Carpinteria fumigation study periods. This is seen as a “zero” by the performance evaluation software which precludes forming measures based on ln(Co/Cp). The OCD5 peak χ/Q for this period is increased to 0.001 µs/m 3 . With predicted Iy, OCD5 produced a peak χ/Q equal to 0.010 µs/m 3 for this period, whereas the observed is 5.20 µs/m 3 . MG,VG results in Figure 4-12 fall into three groups. The first includes CALPUFF configurations with computed TLy based on the turbulence at puff height, the height of the surface layer, and the Monin-Obukhov length. This group shows the strongest bias toward overprediction, with an MG of about 0.5 (factor of two overprediction). This overprediction may result from details chosen for the implementation of this new algorithm (i.e., more development work might improve on this performance), or it may result from problems inherent in using properties of the boundary layer that are poorly known at times. Whatever the reason, the implementation tested here performs poorly compared to the use of the standard Fy function for lateral cloud growth in CALPUFF. The second group includes the two OCD5 results, for modeled and observed Iy. While the tendency toward overprediction is about 10-20%, the scatter is relatively large at VG = 3 to 4. Some of the scatter is due to the fumigation event that OCD5 “misses”. Even with this singular event that tends to increase the MG (reduces the degree of overprediction), the overall MG using the predicted Iy is significantly different from 1.0 because the 95% confidence interval does not overlap 1.0. The third group includes CALPUFF simulations with the Draxler Fy function for lateral cloud growth in the near field and turbulence profiles from either CALPUFF or AERMOD. All 4 configurations have an MG that is near 1.0, and are not significantly different from 1.0. VG for these lies between 1.7 and 1.9, which is slightly greater than the 1.6 that indicates about a factor of two scatter. Within this group, use of the AERMOD turbulence profiles leads to slightly larger concentrations (smaller MG). Using observed Iy also leads to slightly larger concentrations, and larger scatter. The small difference in MG between using the CALPUFF and AERMOD turbulence profiles is statistically significant, but the difference between using observed and predicted Iy is not. Final Report Vol.1 75
- Page 25 and 26: June 4 (1130-1230 CET) L. Turbulenc
- Page 27 and 28: 4. MODEL PERFORMANCE EVALUATION 4.1
- Page 29 and 30: The CALMET preprocessors TERREL, CT
- Page 31 and 32: All sampler locations are used as r
- Page 33 and 34: Table 4-2b Over-water Meteorologica
- Page 35 and 36: Meteorological data used in the OCD
- Page 37 and 38: . . CARPINTERIA, CA 3814 . UTM Nort
- Page 39 and 40: Year Month Day Table 4-4 Over-water
- Page 41 and 42: Datum: NAS-C (North American 1927)
- Page 43 and 44: Table 4-5 Source Characterization f
- Page 45 and 46: Table 4-6a Over-water Meteorologica
- Page 47 and 48: Geophysical Processing Gridded land
- Page 49 and 50: Table 4-7 Source Characterization f
- Page 51 and 52: tape format called GF-3. These data
- Page 53 and 54: . . Strait of Oresund 6230 6220 UTM
- Page 55 and 56: . . Strait of Oresund 6230 6220 UTM
- Page 57 and 58: Table 4-10 SEA.DAT Meteorological D
- Page 59 and 60: the numbers refer to ICOARE values,
- Page 61 and 62: that are simulated during the one-h
- Page 63 and 64: 4.3 Evaluation Results Cameron, Car
- Page 65 and 66: improve model performance relative
- Page 67 and 68: Figure 4-7. Graphical depiction of
- Page 69 and 70: Figure 4-8. Graphical depiction of
- Page 71 and 72: Figure 4-9. Graphical depiction of
- Page 73 and 74: Figure 4-10. Graphical depiction of
- Page 75: Table 4-15 Performance Statistics f
- Page 79 and 80: Table 4-16 Performance Statistics f
- Page 81 and 82: Table 4-16 (continued) Performance
- Page 83 and 84: Figure 4-12. Graphical depiction of
- Page 85 and 86: into CALPUFF, and the simulations a
- Page 87 and 88: Table 4-18 Oresund Results with Min
- Page 89 and 90: Table 4-20 Oresund Results with Min
- Page 91 and 92: Table 4-21 Performance Statistics f
- Page 93 and 94: 4.4 Recommendations A minimum σ v
- Page 95 and 96: SEA.DAT and OZONE.DAT files are pro
- Page 97 and 98: 36 34 32 30 Latitude 28 26 24 22 -1
- Page 99 and 100: WMO Number WBAN Number Station Iden
- Page 101 and 102: WMO Number WBAN Number Station Iden
- Page 103 and 104: WMO Number WBAN Number Station Iden
- Page 105 and 106: 36 72363 72340 34 72249 72248 72235
- Page 107 and 108: 36 34 32 41008 30 42035 42007 42040
- Page 109 and 110: 36 34 32 30 Latitude 28 26 24 22 -1
- Page 111 and 112: COOP State Station Name Table 5-4.
- Page 113 and 114: COOP State Station Name Table 5-4.
- Page 115 and 116: Table 5-4. (Concluded) Hourly NWS P
- Page 117 and 118: Table 5-5. Ozone Stations Station I
- Page 119 and 120: Table 5-5. (Continued) Ozone Statio
- Page 121 and 122: Table 5-5. (Continued) Ozone Statio
- Page 123 and 124: 36 317 339 361 383 405 427 449 471
- Page 125 and 126: - Figure 5-8. “Inputs/Outputs & R
Performance of CALPUFF Configur<strong>at</strong>ions Using Standard COARE in CALMET<br />
<strong>The</strong> performance of all eight CALPUFF configur<strong>at</strong>ions tested are intercompared<br />
using runs made with the standard COARE option, denoted here as c10d, with a<br />
minimum σ v = 0.37 m/s. Results from Appendix A are collected and processed to<br />
form the geometric performance measures. OCD5 results are also included as a<br />
reference point. Results of applying the model evalu<strong>at</strong>ion software, including 95%<br />
confidence limits, are listed in Table 4-16 and Figure 4-12. Note th<strong>at</strong> OCD5 with<br />
observed Iy produces a peak χ/Q less than 0.001 µs/m 3 in one of the Carpinteria<br />
fumig<strong>at</strong>ion study periods. This is seen as a “zero” by the performance evalu<strong>at</strong>ion<br />
software which precludes forming measures based on ln(Co/Cp). <strong>The</strong> OCD5 peak<br />
χ/Q for this period is increased to 0.001 µs/m 3 . With predicted Iy, OCD5 produced a<br />
peak χ/Q equal to 0.010 µs/m 3 for this period, whereas the observed is 5.20 µs/m 3 .<br />
MG,VG results in Figure 4-12 fall into three groups. <strong>The</strong> first includes CALPUFF<br />
configur<strong>at</strong>ions with computed TLy based on the turbulence <strong>at</strong> puff height, the height<br />
of the surface layer, and the Monin-Obukhov length. This group shows the strongest<br />
bias toward overprediction, with an MG of about 0.5 (factor of two overprediction).<br />
This overprediction may result from details chosen for the implement<strong>at</strong>ion of this<br />
new algorithm (i.e., more development work might improve on this performance), or<br />
it may result from problems inherent in using properties of the boundary layer th<strong>at</strong><br />
are poorly known <strong>at</strong> times. Wh<strong>at</strong>ever the reason, the implement<strong>at</strong>ion tested here<br />
performs poorly compared to the use of the standard Fy function for l<strong>at</strong>eral cloud<br />
growth in CALPUFF.<br />
<strong>The</strong> second group includes the two OCD5 results, for modeled and observed Iy.<br />
While the tendency toward overprediction is about 10-20%, the sc<strong>at</strong>ter is rel<strong>at</strong>ively<br />
large <strong>at</strong> VG = 3 to 4. Some of the sc<strong>at</strong>ter is due to the fumig<strong>at</strong>ion event th<strong>at</strong> OCD5<br />
“misses”. Even with this singular event th<strong>at</strong> tends to increase the MG (reduces the<br />
degree of overprediction), the overall MG using the predicted Iy is significantly<br />
different from 1.0 because the 95% confidence interval does not overlap 1.0.<br />
<strong>The</strong> third group includes CALPUFF simul<strong>at</strong>ions with the Draxler Fy function for<br />
l<strong>at</strong>eral cloud growth in the near field and turbulence profiles from either CALPUFF<br />
or AERMOD. All 4 configur<strong>at</strong>ions have an MG th<strong>at</strong> is near 1.0, and are not<br />
significantly different from 1.0. VG for these lies between 1.7 and 1.9, which is<br />
slightly gre<strong>at</strong>er than the 1.6 th<strong>at</strong> indic<strong>at</strong>es about a factor of two sc<strong>at</strong>ter. Within this<br />
group, use of the AERMOD turbulence profiles leads to slightly larger concentr<strong>at</strong>ions<br />
(smaller MG). Using observed Iy also leads to slightly larger concentr<strong>at</strong>ions, and<br />
larger sc<strong>at</strong>ter. <strong>The</strong> small difference in MG between using the CALPUFF and<br />
AERMOD turbulence profiles is st<strong>at</strong>istically significant, but the difference between<br />
using observed and predicted Iy is not.<br />
Final Report Vol.1 75