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 ...

20.04.2013 Views

the duration the tracer resides on the 600 km arc, with values of 14 hours (1998EPA_PG) and 13 hours (1998EPA_CAL). Since the 1998 EPA CALPUFF runs estimated that the tracer arrives after the sampling started (hour 3), then this is a true overstatement of the tracer residence time and not an artifact of the tracer sampling starting after, or at the same time, the observed tracer arrived at the arc. There are a couple exceptions to the initial CALPUFF simulations performed in this study that understated the observed tracer duration on the arc by approximately a factor of 2, which are discussed below. The BASEA_PG scenario estimates that the tracer is on the arc for 12 hours, the same as the observed. However, it estimates the tracer leaves three hours earlier (hour 14) than observed (hour 11). Why the BASEA_PG tracer plume time statistics are so different from the two companion turbulence dispersion CALPUFF sensitivity tests (BASEA_CAL and BASEA_AER) is unclear. The same meteorological fields were used in the three BASEA CALPUFF sensitivity tests and the only difference was in the dispersion options. This large difference in the CALPUFF predicted tracer residence time due to use of the PG versus CAL or AER dispersion options (12 hours versus 6‐7 hours) was not seen in any of the other CALPUFF sensitivity experiment configurations. Although use of the PG dispersion sometimes increases the estimated tracer residence time on the arc by one hour in some of the CALPUFF sensitivity tests (Table 3‐14). The EXP2C series of experiments have estimated tracer plume duration times (11‐13 hours) that is comparable to what was observed. EXP2C uses 36 km MM5 data and CALMET was run using a 4 km grid resolution with no meteorological observations (NOOBS = 2). When meteorological observations are added, either surface data alone (EXP2B) or surface and upper‐air measurements (EXP2A), the tracer duration statistics degrades to only 5 to 8 hours on the arc. It is interesting to note that all of the “C” series of experiments (i.e., use of no meteorological observations in CALMET) exhibit better plume residence time statistics than the experiments that used meteorological observations (with the exception of BASEA_PG discussed previously). But only experiment EXP2C (and BASEA_PG) using 36 km MM5 data and CALMET run with 4 km grid resolution was able to replicate the observed tracer residence time. Most of the initial CALPUFF sensitivity tests were unable to reproduce the observed tracer residence time on the 600 km arc, as was done in the EPA 1998 study using earlier versions of CALPUFF. Even the BASEA_CAL sensitivity test, which was designed to be mostly consistent with the 1998EPA_CAL simulation, estimated tracer plume residence time that was half of what was observed and estimated by the 1998EPA_CAL simulation. In addition to using difference versions of the CALPUFF model (Version 4.0 versus 5.8), the BASEA_CAL simulation also did not invoke the slug option as was used in 1998EPA_CAL (MSLUG = 1). The use of the slug option is designed for near‐source applications and is not typically used in LRT dispersion modeling, so in this study the initial CALPUFF sensitivity tests did not use the slug option for modeling of the 600 km arc. The effect of the slug option is investigated in additional CALPUFF sensitivity tests discussed later in this Chapter. 41

20% 10% 0% ‐10% ‐20% ‐30% ‐40% ‐50% ‐60% ‐70% 20% 10% 0% ‐10% ‐20% ‐30% ‐40% ‐50% ‐60% ‐70% 1998EPA_PG 1998EPA_CAL BASEA_AER BASEA_CAL BASEA_PG EXP1A_AER EXP1A_CAL EXP1A_PG Figure 3‐7. Percent difference in the predicted and observed duration of time tracer is residing on the GP80 600 km arc for the CALPUFF sensitivity tests using puff model formulation and no puff splitting. 42 EXP1B_AER EXP1B_CAL EXP1B_PG EXP1C_AER EXP1C_CAL EXP1C_PG EXP2A_AER EXP2A_CAL EXP2A_PG EXP2B_AER EXP2B_CAL EXP2B_PG EXP2C_AER EXP2C_CAL EXP2C_PG EXP3A_AER EXP3A_CAL EXP3A_PG EXP3B_AER EXP3B_CAL EXP3B_PG EXP3C_AER EXP3C_CAL EXP3C_PG MMIF12_CAL MMIF12_PG

<strong>the</strong> duration <strong>the</strong> tracer resides on <strong>the</strong> 600 km arc, with values <strong>of</strong> 14 hours (1998EPA_PG) <strong>and</strong><br />

13 hours (1998EPA_CAL). Since <strong>the</strong> 1998 EPA <strong>CALPUFF</strong> runs estimated that <strong>the</strong> tracer arrives<br />

after <strong>the</strong> sampling started (hour 3), <strong>the</strong>n this is a true overstatement <strong>of</strong> <strong>the</strong> tracer residence<br />

time <strong>and</strong> not an artifact <strong>of</strong> <strong>the</strong> tracer sampling starting after, or at <strong>the</strong> same time, <strong>the</strong> observed<br />

tracer arrived at <strong>the</strong> arc. There are a couple exceptions to <strong>the</strong> initial <strong>CALPUFF</strong> simulations<br />

performed in this study that understated <strong>the</strong> observed tracer duration on <strong>the</strong> arc by<br />

approximately a factor <strong>of</strong> 2, which are discussed below.<br />

The BASEA_PG scenario estimates that <strong>the</strong> tracer is on <strong>the</strong> arc for 12 hours, <strong>the</strong> same as<br />

<strong>the</strong> observed. However, it estimates <strong>the</strong> tracer leaves three hours earlier (hour 14) than<br />

observed (hour 11). Why <strong>the</strong> BASEA_PG tracer plume time statistics are so different from<br />

<strong>the</strong> two companion turbulence dispersion <strong>CALPUFF</strong> sensitivity tests (BASEA_CAL <strong>and</strong><br />

BASEA_AER) is unclear. The same meteorological fields were used in <strong>the</strong> three BASEA<br />

<strong>CALPUFF</strong> sensitivity tests <strong>and</strong> <strong>the</strong> only difference was in <strong>the</strong> dispersion options. This large<br />

difference in <strong>the</strong> <strong>CALPUFF</strong> predicted tracer residence time due to use <strong>of</strong> <strong>the</strong> PG versus CAL<br />

or AER dispersion options (12 hours versus 6‐7 hours) was not seen in any <strong>of</strong> <strong>the</strong> o<strong>the</strong>r<br />

<strong>CALPUFF</strong> sensitivity experiment configurations. Although use <strong>of</strong> <strong>the</strong> PG dispersion<br />

sometimes increases <strong>the</strong> estimated tracer residence time on <strong>the</strong> arc by one hour in some<br />

<strong>of</strong> <strong>the</strong> <strong>CALPUFF</strong> sensitivity tests (Table 3‐14).<br />

The EXP2C series <strong>of</strong> experiments have estimated tracer plume duration times (11‐13<br />

hours) that is comparable to what was observed. EXP2C uses 36 km MM5 data <strong>and</strong><br />

CALMET was run using a 4 km grid resolution with no meteorological observations (NOOBS<br />

= 2). When meteorological observations are added, ei<strong>the</strong>r surface data alone (EXP2B) or<br />

surface <strong>and</strong> upper‐air measurements (EXP2A), <strong>the</strong> tracer duration statistics degrades to<br />

only 5 to 8 hours on <strong>the</strong> arc. It is interesting to note that all <strong>of</strong> <strong>the</strong> “C” series <strong>of</strong><br />

experiments (i.e., use <strong>of</strong> no meteorological observations in CALMET) exhibit better plume<br />

residence time statistics than <strong>the</strong> experiments that used meteorological observations<br />

(with <strong>the</strong> exception <strong>of</strong> BASEA_PG discussed previously). But only experiment EXP2C (<strong>and</strong><br />

BASEA_PG) using 36 km MM5 data <strong>and</strong> CALMET run with 4 km grid resolution was able to<br />

replicate <strong>the</strong> observed tracer residence time.<br />

Most <strong>of</strong> <strong>the</strong> initial <strong>CALPUFF</strong> sensitivity tests were unable to reproduce <strong>the</strong> observed tracer<br />

residence time on <strong>the</strong> 600 km arc, as was done in <strong>the</strong> EPA 1998 study using earlier versions <strong>of</strong><br />

<strong>CALPUFF</strong>. Even <strong>the</strong> BASEA_CAL sensitivity test, which was designed to be mostly consistent<br />

with <strong>the</strong> 1998EPA_CAL simulation, estimated tracer plume residence time that was half <strong>of</strong> what<br />

was observed <strong>and</strong> estimated by <strong>the</strong> 1998EPA_CAL simulation. In addition to using difference<br />

versions <strong>of</strong> <strong>the</strong> <strong>CALPUFF</strong> model (Version 4.0 versus 5.8), <strong>the</strong> BASEA_CAL simulation also did not<br />

invoke <strong>the</strong> slug option as was used in 1998EPA_CAL (MSLUG = 1). The use <strong>of</strong> <strong>the</strong> slug option is<br />

designed for near‐source applications <strong>and</strong> is not typically used in LRT dispersion modeling, so in<br />

this study <strong>the</strong> initial <strong>CALPUFF</strong> sensitivity tests did not use <strong>the</strong> slug option for modeling <strong>of</strong> <strong>the</strong><br />

600 km arc. The effect <strong>of</strong> <strong>the</strong> slug option is investigated in additional <strong>CALPUFF</strong> sensitivity tests<br />

discussed later in this Chapter.<br />

41

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

Saved successfully!

Ooh no, something went wrong!