Local and urban scale modeling - Air Pollution Modeling and Its Application - page 78

3. Results and Discussion

3.1. Overall results

Results from the global statistical analysis are presented in the Table 1. These
results show that overall HYSPLIT was the best performing model, exhibiting the
highest model composite RANK of 2.00 and the highest individual RANK in 5 of 7
tracer simulations. By contrast, the lowest performing model was CALPUFF, with
a composite RANK score of 1.47. Overall, SCIPUFF and FLEXPART exhibited
similar behavior with composite RANKs of 1.78 and 1.72 respectively.

Table 1

Tracer Study SCIPUFF CALPUFF FLEXPART HYSPLIT

CAPTEX-1 2.23 1.81 2.54 2.35

CAPTEX-2 1.92 1.59 1.65 2.10

CAPTEX-3 1.25 1.10 1.49 1.83

CAPTEX-4 1.82 1.69 1.53 1.88

CAPTEX-5 1.30 1.20 1.81 1.93

CAPTEX-7 2.40 2.06 1.76 1.90

ETEX-1 1.51 0.87 1.24 1.98

Overall RANK 1.78 1.47 1.72 2.00

3.2. European tracer experiment at a glance

Results from the ETEX-1 experiment significantly influence the overall RANK
score for the CALPUFF modeling system (Table 1) and warrants further exami-
nation. Figure 2 displays the time series of the FMS scores over each of the 30
sampling intervals. CALPUFF exhibits marginal agreement during the first 36 h of
the simulation, with the FMS scores varying between 8% and 22%. CALPUFF

performance deteriorates rapidly after 36 h, with the FMS dropping to the 2-4%

Fig. 2. Time series of figure of merit in space (FMS) for ETEX-1