April 1990
EVALUATION OF THE REGIONAL OXIDANT MODEL (VERSION 2.1)
USING AMBIENT AND DIAGNOSTIC SIMULATIONS
by
Thomas E. Pierce
Kenneth L. Schere
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dennis C. Doll
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Warren E. Heilman
Computer Sciences Corporation
Research Triangle Park, NC 27709
Project Officer
Kenneth L, Schere
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ATMOSPHERIC RESEARCH AND Dff»QSURE ASSESSMENT LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
April 1990
EVALUATION OF THE REGIONAL OXIDANT MODEL (VERSION 2.1)
USING AMBIENT AND DIAGNOSTIC SIMULATIONS
by
Thomas E. Pierce
Kenneth L. Schere
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dennis C. Doll
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Warren E. Heilman
Computer Sciences Corporation
Research Triangle Park, NC 27709
Project Officer
Kenneth L. Schere
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
NOTICE
The information in this document has been subject to the United States
Environmental Protection Agency's peer and administrative review and has been
approved for publication as an EPA document. Mention of trade names or
commercial products does not constitute endorsement or recommendation for
use.
AFFILIATIONS
Messrs. Pierce, Schere, and Doll are on assignment to the U.S.
Environmental Protection Agency from the National Oceanic and Atmospheric
Administration, U.S. Department of Commerce. Dr. Heilman is currently with
the U.S. Forest Service in East Lansing, Michigan.
11
-------
ABSTRACT
This report discusses am evaluation of the latest version of EPA's
Regional Oxidant Model, ROM2.1. In the ambient evaluation, model estimates
were compared with ambient measurements of hourly surface ozone collected on
26 days during the summer of 1985 in the Northeastern United States.
Observed and modeled maximum daytime concentrations agreed, on average, to
within 2 ppb or 1.4% (79 ppb versus 77 ppb). The model tended to
underestimate at the higher extremes of the frequency distribution. The
95th-percentile value was underestimated by 8 ppb or 6.6% (127 ppb versus 119
ppb), and the overall maximum value was underestimated by 50 ppb or 22.7%
(219 ppb versus 169 ppb). Underestimates at the upper percentiles tended to
be more prevalent in the southern and western portions of the model domain.
Concentrations at the lower end of the frequency distribution were slightly
overestimated. Estimated and observed spatial patterns of three day maximum
ozone generally showed good agreement. ROM2.1 improved noticeably over
ROM2.0 with regard to the orientation of the high-ozone plumes in the
Northeast Corridor and the depiction of high concentrations along the coast
of Maine. Similar to ROM2.0, a tendency to underestimate peak concentrations
near Washington, DC was again evident with ROM2.1. A unique aspect of the
ambient evaluation was an assessment of the model's ability to estimate
boundary conditions for the Urban Airshed Model. Near the New York City
metropolitan area, estimated and observed boundary conditions agreed to
within 4 ppb or 7.6% (57 ppb versus 61 ppb). Model performance was degraded,
however, during some situations with dynamic mesoscale wind flow conditions.
ROM2.1 also underwent a series of diagnostic tests to investigate the
accuracy of its numerical solution algorithms. When the model was subjected
to extremely steep concentration gradients (steeper than those observed in
the ambient atmosphere), the model did not conserve mass during a 48 h
simulation, deviating by as much as 18% from the initialized value. With
more realistic concentration gradient tests, however, the model conserved
mass.
iii
-------
ROM2.1 will be undergoing a series of improvements based largely on the
findings of this evaluation. The next version of the model will include
refinement of vertical layer specifications and turbulence parameters, and a
correction to the mass imbalance problem. Later versions of the model will
likely include a dynamic meteorological processor that will simulate
nonsteady-state flows.
IV
-------
CONTENTS
Abstract iii
Figures vi
Tables viii
Acknowledgements x
1 Introduction 1
2 Episode Selection for the Ambient Evaluation 4
3 Data Base Development for the Ambient Evaluation 8
4 Ambient Evaluation 14
5 Diagnostic Evaluation 55
6 Summary and Recommendations 90
References 96
Appendix: Statistical Summaries for the August 1985 Episode 98
-------
FIGURES
Number Pa£e
1 Daily exceedances of ozone in the Northeastern
United States • 7
2 The ROMNET modeling domain 11
3 ROM grid points overlaying the UAM
OMNYMAP domain 11
4 Ozone monitoring sites divided into five
geographical groups 12
5 OMNYMAP region monitoring sites used for
developing boundary conditions 12
6 Quantile-quantlie plots of daytime hourly
ozone for the July 1985 episode 26
7 Comparison of observed and modeled daily
maxima for the July 1985 episode 32
8 Surface weather maps for July 9-11, 1985 35
9 Spatial patterns of maximum ozone
for July 9-11, 1985 36
10 Surface weather maps for July 13-15, 1985 37
11 Spatial patterns of maximum ozone for
July 13-15, 1985 . 38
12 Surface weather maps for July 18-20, 1985 39
13 Spatial patterns of maximum ozone for
July 18-20, 1985 40
14 Surface weather maps for August 13-15, 1985 41
15 Spatial patterns of maximum ozone for
August 13-15, 1985 42
16 Mean residuals versus wind persistence for the
UAM boundary 45
17 Mean residuals versus daily average wind
direction for the UAM boundary 45
18 Division of the OMNYMAP boundary into eight groups .. 46
19 Mean daytime ozone concentrations by UAM
group for the UAM boundary .,
vi
-------
FIGURES (continued)
Number Page
20 Mean residuals for each UAM group
experiencing incoming flow 47
21 Ozone concentrations for July 10, 1985 (0800 EST) 48
22 Ozone concentrations for July 10, 1985 (1000 EST) 49
23 Ozone concentrations for July 10, 1985 (1200 EST) 50
24 Ozone concentrations for July 10, 1985 (1400 EST) 51
25 Ozone concentrations for July 10, 1985 (1600 EST) 52
26 Hourly cloud cover and solar fluxes for
grid cell (45,21) 53
27 Hourly layer thicknesses for grid cell (45,21) 53
28 Hourly wind speeds for grid cell (45,21) 54
29 Anticyclonic rotational flow used in
test cases 1A, IB, and 1C 68
30 Initial species concentration field in ROM layer 1
for test case 1A 69
31 Time series of normalized mass ration within the
entire ROM domain for test case 1A 70
32 Three-dimensional schematic of height of top of
ROM layer 1 for test case IB 71
33 Three-dimensional schematic of height of top of
ROM layer 2 for test case IB 72
34 Time series of normalized mass ratio within the
entire ROM domain for test case IB 73
35 Time series of normalized mass ratio within the
entire ROM domain for test case 1C 74
36 Purely divergent wind field used in test case 2 75
37 Three-dimensional schematic of height of top of
ROM layer 1 for test case 2 76
38 Three-dimensional schematic of height of top of
ROM layer 2 for test case 2 77
VII
-------
FIGURES (concluded)
Number Page
39 Three-dimensional schematic of height of top of
ROM layer 3 for test case 2 78
40 Initial species concentration field in ROM layer 1
for test case 2 • • 79
41 Time series of normalized mass ratio within the
entire ROM domain for test case 2 80
42 Quasi-constant wind field used in test cases 0,
OA, and OB 81
43 Initial species concentration field in ROM layer 1
for test case 0 82
44 Time series of normalized mass ratio within the
entire ROM domain for test case 0 83
45 Initial species concentration field in ROM layer 1
for test case OA 84
46 Time series of normalized mass ratio within the
entire ROM domain for test case OA 85
47 Time series of normalized mass ratio within the
entire ROM domain for test case OB 86
48 Frequency distribution of paired grid 0 concentration
differences 87
49 Frequency distribution of paired grid NO concentration
differences 88
50 Frequency distribution of paired grid ROG concentration
differences 89
A-l Quantile-quantile plots of hourly ozone from
the August 1985 episode 100
A-2 Comparison of observed and modeled ozone
maxima from the August 1985 episode 103
viii
-------
TABLES
Number Page
1 Ozone monitoring sites used for developing
UAM boundary conditions 13
2 Summary statistics for all hourly and maximum
ozone concentrations 26
3 Comparison of the July and August 1985 episodes 27
4 Statistical summary by geographical group for the
July 1985 episode 28
5 Summary statistics for UAM boundary conditions 43
6 Daily statistics of UAM boundary condtions 44
7 Maximum percentage changes in total mass during
a simulation 95
A-l Summary statistics of hourly daytime ozone
for the August 1985 episode 99
A-2 Summary statistics of daily ozone maxima for
the August 1985 episode 99
ix
-------
ACKNOWLEDGMENTS
The authors gratefully acknowledge the contributions made by Trudy
Boehm, Carlie Coats, Jeanne Eichinger, Susan Hallyburton, Sundar Jambunathan,
Dianne Jordan, Don Olerud, Chet Wayland, and Jeff Young of Computer Sciences
Corporation under contract number 68-01-7365.
-------
SECTION 1
INTRODUCTION
After realizing that summertime episodes of photochemical smog were a
regional and not a localized urban phenomenon, EPA embarked on a research
program in the middle 1970's to develop a regional-scale computer model for
simulating the transport and fate of ozone (0») and its precursors. Research
O
efforts included extensive field studies (Vaughan, 1985) and model
development (Lamb, 1983). This work eventually resulted in the creation of
the first operational version of the EPA Regional Oxidant Model (ROM1)
(described by Lamb, 1983 and 1984; and Lamb and Laniak, 1985). Testing of
ROM1 and the analysis of field data prompted additional improvements,
including the ability to model biogenic emissions, horizontally-varying layer
thicknesses, and improved deposition relationships. After a period of
development, the first application version of the ROM (Version 2.0) was made
in 1986. It has since been used for several studies in the Northeast and the
Gulf Coast region of the U.S., and it has undergone an intensive evaluation
by Schere and Wayland (1989) using field data collected in 1980.
Additional needs by the northeastern states and the emergence of new
model improvements prompted the development of ROM2.1 as reported by Young et
al. (1989). Improvements to ROM2.1 over ROM2.0 included an expanded modeling
domain, an updated methodology for computing biogenic emissions of
hydrocarbons, revisions in the objective wind field interpolator to remove
known biases in the low-level wind flow, and a more sophisticated
anthropogenic emissions processor.
Creation of ROM2.1 was primarily motivated by the needs of the Regional
Ozone Modeling for Northeast Transport (ROMNET) project (EPA, 1988a). In
general, the objective of ROMNET is to investigate interurban transport of
ozone and its precursor emissions in the Northeast to support state air
pollution control agencies in the development of State Implementation Plans
-------
to achieve the National Ambient Air Quality Standard CNAAQS) for 0 . In this
regard, ROM is being used to estimate regional 0 concentrations and to
examine the effectiveness of a variety of hydrocarbon (VOC) and nitrogen
oxides (NO ) emissions control strategies for reducing 0 . Another major
X J
objective of the ROMNET project is to use ROM calculations for deriving
boundary conditions for urban-scale applications with the Urban Airshed Model
(UAM) (Godowitch and Schere, 1989). Because of ROM's importance for
comparing the estimated effectiveness of emission control strategies and for
estimating boundary conditions, representatives of the ROMNET project asked
EPA's Office of Research and Development to evaluate the performance of
ROM2.1. This report summarizes the evaluation.
The approach we took complements the work of Schere and Wayland (1989),
who evaluated the previous version of ROM, Version 2.0. The primary
difference between the two evaluations is that Schere and Wayland used a
special field-study data set from 1980, while our evaluation used
routinely-collected data from 1985. These data include the anthropogenic
emissions inventory, which was adapted from the 1985 National Acid
Precipitation Assessment Program (NAPAP) inventory (EPA, 1988b). Our
evaluation, unfortunately, could not be as comprehensive as that of Schere
and Wayland (1989). They had access to extensive field measurement data for
ozone, hydrocarbons, and nitrogen oxides as well as aircraft transects; we
were able to use only routine data stored in EPA's Aerometric and Information
Retrieval System (AIRS). After surveying AIRS, it was concluded that
nitrogen oxide and hydrocarbon measurements were too coarse for a rigorous
model evaluation. Therefore, we limited the evaluation of ROM2.1 to hourly
observations of surface ozone from state and local agency monitoring sites.
In this report, we compare observed and modeled ozone concentrations for
selected periods of high ozone observed during the summer of 1985. Periods
from 1985 were chosen because they correspond to the base year emissions
inventory. The objectives of the ambient evaluation were (1) to examine
overall evaluation statistics to determine whether a general bias exists in
the model calculations, (2) to look at spatial patterns of maximum
concentration to determine whether a spatial bias exists, and (3) to examine
-------
the model's applicability for determining UAM boundary conditions.
In addition to the ambient evaluation, we performed a series of
"stressful" diagnostic tests on the model. In the original development of
the first generation ROM (ROM1), a number of diagnostic tests were performed
to probe the accuracy of the numerical algorithms used to solve the equations
that simulated physical and chemical processes (Lamb and Laniak, 1985).
These tests were designed in a hierarchial fashion beginning with a
chemistry-only simulation. Transport was then added to the chemical
simulation, then vertical mixing, and finally source emissions. These tests
represented the next step after independent (external to the ROM framework)
tests of the model's numerical algorithms (Lamb, 1983). Results from this
original set of diagnostic tests demonstrated that the model faithfully
represented solutions to the relevant equations.
Since the time of ROM1, numerous changes have been made to the entire
ROM modeling system culminating in the most recent version of the
second-generation ROM model (ROM2.1). .Although there were significant
changes to the chemical and physical processes simulated within the ROM, few
changes were made to the basic numerical algorithms employed to solve these
processes. Nevertheless, a new round of diagnostic testing was proposed for
ROM2.1 that would complement the earlier work done with ROM1. We performed
these diagnostic tests on the production version of ROM2.1 used for the
ambient evaluation exercises described earlier. In the diagnostic evaluation
section, the accuracy of the numerical algorithms will be assessed by
evaluating the model's ability to conserve mass for these diagnostic tests.
-------
SECTION 2
EPISODE SELECTION FOR THE AMBIENT EVALUATION
Because we wanted the model simulations to correspond to the base year
of the 1985 NAPAP emissions inventory, we examined ozone monitoring data from
1985 for candidate episodes. Daily exceedances (hourly concentrations
greater than 120 ppb) of observed ozone in the northeastern U.S. during 1985
are shown in Figure 1. Two clusters of episodes were apparent: middle July
and early August. We examined these episodes further for possible starting
and ending dates for model simulations, recognizing that the ROM is designed
for three-day segments starting at noon and that the first segment should be
initialized with "clean" (low ozone/precursor concentrations) conditions.
The episodes we selected for modeling for this evaluation were July 7-22 and
August 7-16. Brief descriptions of meteorological conditions and observed
ozone concentrations are given below.
THE 7-22 JULY 1985 EPISODE
Two rather distinct synoptic patterns characterized this period,
occurring on July 7-15 and July 16-22. From the 7th to the 15th, surface
troughs and frontal boundaries predominated in the modeling domain. From the
8th to the 10th, a surface trough persisted along the coast from southern New
England to the Delmarva area and then westward to a front along the Ohio
Valley. From the llth to the 13th, an east-to-west frontal boundary was
located along the Virginia/North Carolina border westward to the Ohio Valley.
Weak high pressure was located over New York with a weak ridge extending from
New England to Pennsylvania. On the 14th, a frontal boundary approaching the
east coast from the west set up weak southwesterly flow along the Northeast
Corridor. Cloudiness was widespread throughout the domain from the 9th to
the 14th, and showers were occasionally observed in the western portion of
the domain.
-------
Southwesterly flow and hot temperatures predominated during the second
portion of this episode. On the 16th, a cold front was located along the
east coast but moved offshore by the 17th. High pressure over the Great
Lakes produced a light northwesterly flow along the Northeast Corridor. The
high progressed southeastward on the 18th and gradually weakened as it moved
into Virginia and North Carolina on the 19th and 20th. Weak southwesterly
flow was observed along the Corridor on the 19th and 20th with relatively hot
daily maximum temperatures (upper 80*s and low 90's [°F]). By the 21st, a
surface trough was located along the east coast while a cold front approached
from the west. Maximum temperatures near 90 F were again observed along the
East Coast on the 21st. The cold front swept through the Northeast Corridor
on the 22nd.
Exceedances of ozone were observed along the Corridor on the 9th, 13th,
19th, and the 20th. Peak hourly concentrations were 165 ppb in central New
Jersey on the 9th; 218 ppb in New York City on the 13th; 163 ppb in northern
New Jersey on the 19th; and 152 ppb in southern Connecticut on the 20th.
Isolated exceedances were observed in the New York City metropolitan area on
the 10th and 21st and near Philadelphia on the 16th.
THE 7-16 AUGUST 1985 EPISODE
The 7th to the 10th was characterized by a high pressure ridge that
extended from New England through central Pennsylvania into western Virginia
and West Virginia. Weak northerly and northeasterly flow occurred along the
Corridor. Showers were observed along the Corridor on the 8th. A cold front
moved through the Corridor on the llth and was located offshore on the 12th.
High pressure moved southeastward from north of the Great Lakes on the 12th,
setting up a ridge along the east coast on the 13th. Relatively high
temperatures were observed in the Baltimore/Washington area on the 13th (low
90's). On the 14th and 15th, high pressure extended from off the East Coast
into the southeastern U.S. Weak but relatively well-defined southwesterly
flow developed along the east coast and high temperatures were observed from
New England to Washington, DC (upper 80's to low 90's). A cold front
approached the East Coast on the 15th and moved offshore on the 16th.
-------
Widespread exceedances of ozone were first observed on the 9th,
extending from Wilmington, DE, to New York City. Exceedances were observed
from New York City to Washington, DC, on the 13th and throughout the Corridor
on the 14th and 15th. On other days, only scattered exceedances were
observed along the Corridor.
On the 9th, peak concentrations ranged from 160 ppb in Wilmington to 164
ppb in New York City. On the 13th, a peak concentration of 201 ppb was
observed near New Brunswick, NJ, with values greater than 150 ppb in
Baltimore, Trenton, and New York City. Peak concentrations on the 14th
included 187 ppb in Baltimore, 184 ppb in New York City, and 169 ppb near
Portland, ME. On the 15th, peak values ranged from 188 ppb to 219 ppb in
southern and central Connecticut, and concentrations along the Maine coast
were above 150 ppb.
-------
1MMY 2MMV 1MJN ZftJIM 1CKAX 34-JUl 7-AUO 21-AUQ 4-8CP
Figure 1. Daily exceedances of ozone in the northeastern U.S. during
1985. An exceedance consists of an hourly concentration greater than
120 ppb.
-------
SECTION 3
DATA BASE DEVELOPMENT FOR THE AMBIENT EVALUATION
Basically, two types of databases were needed for the ambient
evaluation: hourly concentrations of ozone from (1) model estimates and (2)
observations. Development of these two data bases is described below.
MODEL ESTIMATES
To produce the modeled database, we executed the ROM for the two
episodes discussed In Section 2. Model inputs included National Weather
Service surface and upper-air meteorological data, observed ozone
concentrations for estimating boundary conditions, and hydrocarbon and
nitrogen oxide emissions (both anthropogenic and biogenic). More details on
model input requirements are given by Young et al. (1989).
The ROM is a three-layer Eulerian grid-scale model that estimates hourly
photochemical species concentrations for a 64 by 52 grid as shown in Figure
2. Each grid cell is 1/6° latitude by 1/4° longitude, or approximately 19 km
by 19 km. For this study, we evaluated only hourly ozone concentrations from
layer 1; concentrations from this layer most closely represent surface ozone
observations. In layer 1, an individual grid cell is typically -100 m in
vertical extent at night, and 200-500 m deep during the day.
We developed three different model databases for the evaluation. (1)
Point estimates from gridded data: For the portion of the evaluation
concerned with general statistics, we interpolated gridded estimated ozone
values to actual monitoring locations using a biquintic interpolation scheme
(Lamb, 1985) that is consistent with the methodology used in the model. (2)
Contoured values of gridded data: For analyzing spatial patterns, we used an
objective contouring algorithm to produce computer graphics depicting
concentration fields based on the gridded ROM data. (3) Interpolated values
-------
derived using the ROM/UAH interface methodology: For the portion of the
evaluation concerned with boundary conditions, we employed a fairly elaborate
interpolation scheme, described below, that is consistent with the ROM/UAM
interface developed for the ROMNET program (Godowitch and Schere, 1989).
1. A band of three ROM grid cells surrounding the UAM New York
Metropolitan area domain (Rao, 1987) was defined (see Figure 3).
2. Three-hour running averages, centered on the hour, were
calculated for each hour for each ROM grid cell noted in the
three-cell band.
3. Using the hourly concentration averages from step 2, an hourly
spatial average was taken of each three-cell set normal to the
UAM outer boundary and the result noted on the UAM boundary.
4. These concentrations were then spatially interpolated (using
linear averages) to the UAM grid cell centers along the
boundary.
This procedure allowed us to transform boundary conditions from a 19-km grid
size to the grid size of the UAM domain, typically 4 to 10 km.
OBSERVATIONS
We developed the following three observation databases, which are
analogous to the three model databases: (1) a set of observations used for
developing overall statistics; (2) a set of observations used for creating
contour plots; and (3) a set of observations used for developing UAM boundary
conditions. We obtained ozone concentrations from monitoring data archived
in EPA's Aerometric Information and Retrieval System (AIRS) (EPA, 1989),
which contains a national database of hourly ozone (0_) concentrations and
J
information on monitoring sites. Hourly 03 concentrations were selected for
sites located in the U.S. portion of the ROMNET domain for the two episodes
that we modeled with ROM. (Canadian 0 monitoring data are not included on
AIRS and were not readily available for this analysis.) An extensive review
and screening of the data was performed. We included only daytime values
(0800 h LST to 1900 h LST) in the evaluation because nighttime observations
-------
are influenced by localized processes that often include scavenging of 0 by
NO emissions and therefore do not reflect vertically-integrated 0_
X »*
concentrations in layer 1 of the ROM (Schere and Wayland, 1989).
Furthermore, we excluded a site's data on days missing more than 25% of their
observations. We also examined the data for extremely high or low values.
Several sites, such as Poughkeepsie (NY), were eliminated because mean
daytime 0 concentrations were consistently below 50 ppb and may have
reflected local NO scavenging. Of the more than 200 sites in the original
database, 187 of these were used in computing general statistics. For
portions of the analysis, the data were divided into five geographical groups
(Figure 4).
For the spatial plots, little effort was made to eliminate sites because
only maximum concentrations were considered. As a result, the plots showed a
few locations with extremely low values (due either to poor data recovery or
local NO scavenging). We ignored these sites when manually contouring the
data.
The monitoring data used for developing the UAM boundary condition
database were given special consideration. The approach we followed is
consistent with the guidance given by Rao (1987). Because so few monitoring
sites were available for developing boundary concentrations, we distributed
the monitored data among six locations along the UAM boundary: south,
southwest corner, west, northwest corner, north, and east (Table 1). The
assignment of particular sites to a boundary location depended on the
prevailing wind direction for that day. If more than one site was available
for a location, the hourly concentrations were averaged. Monitors used in
our analysis of boundary conditions are shown in Figure 5. After averaging
at the six locations, concentrations were spatially interpolated (using
linear avearaging) along the UAM boundary. To be consistent with the ROM/UAM
interface methodology, we then used the hourly concentrations to create
three-hour running averages as described earlier.
10
-------
n
Figure 2. The ROMNET modeling domain. Dots represent model grid cell
centers
<«3<-
4X4 -
43<4 -
. -
4464 -
. 12
-'
i
-
~"
-
4 .
-
•v
-
_j
-
N
-
"N
EJ
-
T
r\
.
.
.
;\
IS
•
-
\
-
<
-
-
-.
2
-
-
Y
-
1
•v
-
^
-
. 564 .
—
Oi
-
?
-
— I
K
.
_j
K-
-J
/
i
\
•
-
i
-
—
/
J
>,
-
-
r—
1.
\
?
*S
-
r
-
~
.1
X
cj
-
L
-
-
^
<$
**•
-c
£>
-••"
-
^-~
^
•-•
-
f
c
•
.
^
.
1
-'
0
f
-
II
^*
X1
. • • •
t
.
-•
•
.
"
^T
rf-
••
^
.
C
«
^-
J'
•Hi.
^
— ,
X
.^
604 . 444 • 01)4
.
•
—
/
r^
^
»jr
_
•
7'
x**
nfl
*^
4
•
(
m
7
r^
•*».
^>
V
•
.
•
•
~^j
•
-
•
^
f
P^
71
*
. *f
• X
m «
V
• *
• IT
• M
• *
X
m X
K
, , K
4
. V
Figure 3. ROM grid points overlaying the UAM domain for the New York
City metropolitan area (also referred to as the OMNYMAP domain).
11
-------
I
Figure 4. AIRS ozone monitoring sites divided into five geographical
groups.
Figure 5. The monitoring sites used for developing boundary conditions
for the OMNYMAP domain.
12
-------
Table 1. Ozone monitoring sites used for developing UAM boundary
conditions along the OMNYMAP domain. The arrows indicate which sites are
used on each day of the episode; the direction of the arrow indicates the
location(s) along the boundary that the data are used (T = south, „ =
southwest corner, -» = west, * = northwest corner, ^ = north, and «- =
east).
July
Monitoring site ID 07080910111213141516171119202122
Ocean Co, NJ CLM t
McGuire AFB, NJ MCG ^ t „ t ^ t f
Trenton, NJ TR2 ^^^^^^^ ^-^^^ ^ 71^ ^
Trenton, NJ TR1 *T T T.t at » t j.t T ^7rT T.^
Flemington, NJ FLM T -» „-» *-» t -» -» -> -» „-» -» -> -> „-» -» „-»
Bristol, PA BST „ * „ , , , , 7,7,
Morris Co, NJ NJ1 -» -^ -» ^
Allentown, PA ALT -»->-»-»-» ^->-> ->->->->
Easton, PA ETN ^-»-> ->-» ->-»^ ^-» ->
Scranton, PA SCR -»-»->->-> » » -» ^-»^->
Carbondale, PA CRB -» -> ^ * * a
Rensselaer, NY REN 4-4- 4, 4-
Afi5LW3.nl MA AGA ^^^^^^4'^4'^'^4'4'4'4^^
Pittsfield, MA PTS ^ * 4- 4- 4- 4,
Chicopee, MA CCP ^ ^
Ware, MA WAR ^ 4'
Groton, CT CRT <-<-*-<-<-<-<-<-<-<-<-<-<-<-<-<-
Kent Co. RI PRK <- <- <-
Providence, RI PRV <- «-
August
Monitoring site ID 07080910111213141516
Ocean Co, NJ CLM T t
McGuire AFB, NJ MCG t t t t T
Trenton, NJ TR2 ^/j>-^7i^,^,^^/^/|s/f.
Trenton, NJ TR1 TI/^TI/H TI /^^/^^/^
Flemington, NJ FLM -» -» -» ^ -7 x-* 7.^ ^ ->?.->
Bristol, PA BST 71 TI T> TI TI TI 71
Morris Co, NJ NJ1 -» -»
Allentown, PA ALT -»-» -»-»-»-»-»-»
Easton, PA ETN -»-> ->->-»-»->->
Scranton, PA SCR ^ * -» ^ ^ ->
Carbondale, PA CRB " ^ » »
Rensselaer, NY REN *
Agawam, MA AGA 4,4,4,4-4,4,4,4,4,4,
Pittsfield, MA PTS ^ ~J'
Chicopee, MA CCP 4, 4- 4, 4-
Ware, MA WAR 4< 4, 4- 4,
Groton.CT CRT <-«-«-<-<-«-<-<-«-<-
Kent Co, RI PRK «-
Providence, RI PRV
13
-------
SECTION 4
AMBIENT EVALUATION
To emphasize different aspects of model performance, we broke the
ambient evaluation of ROM2.1 into three parts: overall statistics, spatial
patterns, and UAM boundary conditions.
OVERALL STATISTICS
Using the modeled and observed databases of hourly ozone concentrations,
descriptive statistics were computed using SAS (1985) procedures. From these
general statistics, we attempted to assess the overall temporal and spatial
performance. The intent here was not to evaluate ROM in a deterministic
manner—that is, by comparing closely paired observations and estimates. We
believe that using ROM in a strictly deterministic manner is inappropriate
because (1) ROM's grid cells are of significant size (19 km by 19 km), (2)
stochastic variations in atmospheric flow cannot be resolved using the
current network of meteorological observing sites, and (3) the degree of
spatial variability of actual 0_ concentrations cannot be detected by the
current monitoring network. In fact, Schere and Wayland (1989) and Lamb and
Hati (1987) also recommend against the deterministic use of models such as
the ROM.
Summary statistics for all modeled and observed daytime hourly
concentrations for the combined July and August episodes are shown in Table
2, together with statistics for daily maxima at all sites. The observed and
modeled data sets were treated separately in calculating the statistics. The
mean observed and modeled values agree fairly closely; the hourly means agree
to within 10 ppb and the daily maximum means agree to within 1 ppb (which is
not significant at the 99'/. confidence level). Although the model
overestimated the mean hourly value, it tended to underestimate the hourly
and daily maximum values in the upper extremes of the frequency
distributions. The 95th-percentile hourly value was underestimated by 1 ppb,
and the maximum was underestimated by 50 ppb; the daily maximum values show a
14
-------
similar tendency.
Statistics for the individual July and August episodes are shown in
Tables 3a and 3b; they are similar to those for the combined episodes.
Because results for the two episodes are similar to each other, we limit our
discussions in the rest of this report primarily to the July episode. A
complete list of statistics for the August episode is available in the
appendix.
To examine spatial trends in the data, we separated the monitoring sites
into five geographical groups (Figure 4) as follows: (1) Northern Corridor,
(2) Southern Corridor, (3) Ohio Valley-Middle Atlantic, (4) Interior
Northeast, and (5) Great Lakes. Statistics for each group are summarized in
Tables 4a and 4b. Depending on the statistic, model performance was better
in some groups than in others. The statistic we preferred to examine was the
95th-percentile value from the daily maximum concentrations, because it is
robust and can be related to NAAQS assessment. As shown in Table 4b, all
concentrations at the 95th percentile were underestimated except for group 1,
which was overestimated by 5%. It is encouraging to note that, except for
group 3 which was underestimated by 19%, all estimated and observed
concentrations at the 95th percentile agreed to within 10%.
A valuable tool for assessing model performance is a quantile- quantile
plot, which can be used to compare the frequency distributions of sorted
observed and sorted estimated concentrations. The concentrations are sorted
from highest to lowest and then plotted on an x-y plot. The x-axis depicts
observed data, and the y-axis depicts estimated data.
Quantile-quantile (hereafter refered to as QQ) plots of hourly daytime
concentrations for each of the five geographical groups during the July
episode are shown in Figures 6a-6e. In each of these plots, the solid line
denotes a perfect fit, and the dashed lines show a ten percent deviation from
this line. The findings from the QQ plots are consistent with overall
statistics shown previously. Model estimates from groups 2, 3, and 4 tended
to be lower than observations in the higher quantiles; underestimates were
15
-------
evident in the top 30 percent of group 2, top 50 percent of group 3, top 20
percent of group 4. Group 1 showed general overestimates in the upper
quantiles (except for the top 1%), and group 5 ranged from good agreement to
overestimates at the upper values. Overestimates were evident at the lower
ends of the frequency distribution for all five groups. Median values in
general agreed quite well. In groups 1 and 5, median values were
overestimated by about 20% (around 10 ppb); in groups 2, 3, and 4, the
median was overestimated by 10% or less. In the Northeast Corridor,
differences between the northern portion of the Corridor (group 1) and the
southern portion (group 2) were evident. Group 2's estimates agreed much
better with observations than did group 1's. Similar to what Schere and
Wayland (1989) noted in their evaluation of ROM2.0, estimates in the southern
part of the Corridor tended to be lower than the observations at the upper
end of the frequency distribution, although estimates were within 10% of
observations except for the maximum value. The tendency to underestimate
peak concentrations in group 2 will be discussed below in the spatial
analysis section. The region showing the greatest underestimate in the upper
values was group 3, an area removed from the extensive metropolitan area
along the Northeast Corridor. In groups 4 and 5, agreement was quite good.
In an effort to understand how well the ROM tracked daily maximum
concentrations, we computed frequency distributions of maximum (observed and
estimated) ozone concentrations from the data in each group for each day.
Figures 7a-7e compare box plots of modeled and observed daily maximum
concentrations for all but the first two days of the July episode. The first
two days are not shown because the model is strongly influenced by initial
conditions and tends to underestimate ozone concentrations during this
"start-up" period. The box plots denote the maximum, 75th percentile, 50th
percentile (or median), 25th percentile, and minimum value. These plots show
similar results to those noted in the QQ plots for hourly concentrations;
however, they also provide temporal information.
On a daily basis, the model tended to track the observations fairly
well. For all of the regions combined, there were about equal numbers of
days with underestimates and overestimates. As discussed for the QQ plots,
16
-------
group 1 tended to experience the most overestimates (12 out of 14 daily
medians were overestimated) while group 3 tended to experience the most
underestimates (medians for 12 out of 14 days were underestimated). The
daily maximum plots also show that as one moves away from the Corridor
(groups 1 and 2), median daily maximum concentrations tended to approach
background values.
One of the ROM's most important uses is to simulate days having ozone
exceedances (hourly concentrations greater than 120 ppb). In group 1, eight
exceedance days were observed while ten exceedance days were estimated, thus
indicating that the ROM tends to slightly overestimate in the northern
portion of the Corridor. Further to the south in group 2, however, ten
exceedance days were observed but only six were estimated. In group 3, six
exceedance days were observed but none were estimated. The monitors
reporting these six exceedances were in smaller urban areas such as Richmond,
Norfolk, and Charleston (WV) than those found in the Northeast Corridor.
This suggests that the coarse grid resolution (19 km) of the ROM may be
insufficient to adequately resolve smaller urban plumes. Also,
naturally-occuring hydrocarbon and NO emissions are probably more important
in these areas because of the relatively lower anthropogenic emission
densities; so, uncertainties in estimating naturally-occurring emissions may
have contributed to poor performance in group 3. Despite the relatively poor
model performance in the southern portion of the domain, performance in the
Great Lakes, Interior Northeast, and the Northern Corridor groups seems quite
good.
SPATIAL PATTERNS
For this portion of the evaluation, we examined spatial patterns of
maximum hourly ozone for four distinct three-day episodes that represent a
range of meteorological conditions and model performance.
The 9-11 July Episode
This was a period of unsettled weather with weak pressure gradients.
17
-------
For much of the episode, a weak surface trough was draped from Long Island to
Virginia to northeastern Ohio to Illinois (Figure 8). By the 10th,
southwesterly flow became better defined in advance of a cold front moving
eastward through the Great Lakes. Of special interest on this day was a
mesoscale high pressure area associated with a line of thunderstorms in
central Pennsylvania, which eventually moved through the southern part of
the corridor and significantly reduced ozone concentrations. This feature
will be discussed below in the section on background concentrations. By the
morning of the llth, less humid conditions and northwesterly flow prevailed
as the cold front had pushed off the East Coast.
During this period, monitors picked up ozone plumes above 120 ppb from
Wilmington, DE, to Boston (Figure 9). Observed and estimated maximum hourly
concentrations greater than 160 ppb were observed near central New Jersey and
coastal Connecticut. Agreement over southern Connecticut was good, although
the modeled plume appears to be shifted slightly east and south of observed
values. The model also shifted the high concentrations just off the coast of
New Jersey near Atlantic City. Simulating small-scale features during this
episode was particularly difficult because of the weak wind flow, the
occurrence of the mesoscale high, and the persistence of a coastal trough.
Overall, however, estimates of maximum ozone for this episode agreed fairly
well with observed data.
The 13-15 July Episode
Figure 10 shows a meteorological scenario typically associated with
elevated ozone concentrations in the Northeast. A weak cool high pressure
system centered over New York on the 13th gave way to a warm front on the
14th. By the 15th, southwesterly flow was fully established. Unsettled
conditions prevailed in the western portion of the model domain. Showers
were reported on the 14th and the 15th in western New York and Pennsylvania
in advance of a slow moving cold front.
During this episode, two separate areas of high ozone values were
observed (Figure 11). A modest area of exceedances (maximum of 132 ppb) was
18
-------
noted from Boston to the southern coast of Maine. High concentrations were
also noted in distinct "blobs" near Philadelphia and New York City.
Concentrations greater than 200 ppb were observed near Wilmington, DE, (211
ppb) and Bayonne, NJ (218 ppb). Although an area of observed exceedances
stretched from Washington, DC, to central Long Island, no well-defined plume
downwind of New York City was apparent.
Except north of New York City, estimated ozone concentrations were lower
than those observed during this episode. However, Figure 11 shows that the
intermittent nature of the high ozone concentration areas was well
represented. Near Boston, the modeled and observed concentrations agreed
closely as noted by the placement of the 140 ppb contour on both plots.
Unfortunately downwind of Wilmington, DE, the model underestimated the peak
observed concentration by about 80 ppb. Part of the poor performance may be
due to a coastal trough that existed during this episode as shown in the
synoptic analysis for July 15th (Figure 10). Pagnotti (1987) points out that
small-scale features such as this tend to produce localized pockets of high
ozone that would not be well-represented by a regional model such as the ROM.
The 18-20 July Episode
This period began with weather forecasters tracking Tropical Storm Ana,
which moved harmlessly off the New England coast. Figure 12 shows that high
pressure held firmly over the Northeast, gradually shifting southward and
weakening. By the 19th, southerly flow had become fully established over the
Corridor. On the 20th, a coastal trough was evident along the Atlantic
Coast.
Under light wind conditions, high concentrations were once again
positioned in small, distinct areas (Figure 13). Values above the standard
stretched from the Delmarva peninsula to central Massachusetts. The highest
concentration (164 ppb) was observed near Trenton, NJ. A small but
impressive area of high concentrations (maximum of 152 ppb) was observed
along the coast of Maine. With minor exceptions, the model showed reasonable
agreement to the peak observed values during this period (Figure 13). A
19
-------
plume over 120 ppb extended from near Baltimore northward to off the coast of
Maine. Pockets over 160 ppb were estimated near New York, which agreed, in
general, with the observed values. Except for missing the high
concentrations over the Delmarva peninsula, ROM appears to do a good job
predicting the magnitude and placement of the maximum ozone plume during this
period.
The 13-15 August 1985 Episode
This episode was similar to the July 13-15 episode. Weak cool high
pressure was positioned over the Northeast on the 13th (Figure 14). The
system had moved eastward by the 14th allowing weak southwesterly flow to
ensue. By the 15th, a cold front began moving through the western portion of
the domain. Southwesterly flow strengthened, but a hint of a coastal trough
was evident in the surface weather map.
Monitoring data revealed an extensive area of ozone exceedances that
stretched from Wilmington, DE, to coastal Maine (Figure 15). Highest
concentrations were observed near Hartford, CT (219 ppb). An isolated area
of high concentration (maximum of 187 ppb) was observed just north of
Baltimore.
The general orientation of the modeled ozone plume is excellent when
compared to observations (Figure 15). The highest estimates, however,
approached only 200 ppb, more than 20 ppb less than the highest observation.
The model did simulate an area of high concentrations (greater than 180 ppb)
off the coast of Maine, close to where relatively high concentrations were
observed. While the model did not replicate the extreme peaks, it
satisfactorily estimated the pattern and shape of the ozone plume near the
Northeast Corridor.
BOUNDARY CONDITIONS
One of the intended uses of ROM is to estimate boundary conditions for
the Urban Airshed Model (UAM). UAM modeling will be needed for resolving
20
-------
small-scale gradients of ozone near urban areas and for determining the
degree of nonattainment in the urban airshed relative to the NAAQS for 0 .
The grid resolution of the UAM varies from 2 to 10 km, whereas the ROM has a
grid resolution of about 19 km. For this evaluation, we compared ROM
estimates of ozone boundary concentrations from layer 1 with boundary
concentrations derived from monitoring data taken near the OMNYMAP domain;
the development of the two databases was described in Section 3. This
evaluation was limited to near-surface concentrations because measurements
were available only from surface-based monitors. The 73 UAM grid points used
in our analysis stretch from the coast of New Jersey clockwise to the coast
of Connecticut. We excluded grid boundaries over the Atlantic Ocean because
of an obvious lack of ozone monitoring data there.
We began the analysis by looking at the overall performance of the ROM
in estimating daytime boundary conditions. Table 5 shows summary statistics
for the 20, 155 data points available; these data were taken from valid
combinations of estimated and observed daytime (0800 LST to 1900 LST) hourly
concentrations for 73 grid locations for 24 days of data. The model slightly
overestimated the mean concentrations (by 5 ppb or 7.6%). At higher
quantiles, however, the model tended to underpredict in a manner consistent
with the statistics we presented earlier. In addition, the larger standard
deviation of the observed concentrations shows that observations exhibited
more variability than the model estimates.
Because there was no overwhelming bias apparent in the model, we
examined the boundary condition calculations in further detail. Table 6
compares estimated and observed mean concentrations on a daily basis. While
the mean values summed over all the days agreed to within 10%, there was some
day to day variability. Percent differences for individual days ranged from
-24% to +49%. However, estimated and observed mean values were within ±10
ppb on 14 of the 22 days. Other pertinent information on mean wind
persistence, wind speed, and wind direction is also noted in Table 6.
Figure 16 displays model performance as a function of wind persistence.
Surprisingly, the more persistent flow conditions were associated with
21
-------
greater overestimates, as noted by the mean fractional errors which dropped
from 237. percent during very persisent conditions to -18% during low
persistence.
Next we looked for a relationship between model performance and daily
average wind direction. Figure 17 shows, as expected, that southerly and
southwesterly winds were most prevalent during the model simulations. In
terms of model performance, model estimates and observations agreed more
closely when southerly winds pervailed. With southerly winds, there were 10
days of overestimates and 8 days of underestimates, for an average error of
5%. With northerly winds, the model overestimated on all five days, and the
average error was +26%.
We then geographically stratified the data set to look for spatial
biases in the estimates. Figure 18 shows the eight groups of grid cells
formed, with group 1 located in the south and group 8 located in the east.
Mean observed and estimated concentrations for the eight groups are shown in
Figure 19. The highest observations and estimates occurred in group 1, which
is located in the center of the Northeast Corridor and would experience high
ozone concentrations typical of southwesterly flow conditions. It is
noteworthy that the observed and modeled patterns in Figure 19 resemble each
other; both show a minimum along the northern boundary and a maximum for
group 1. Best agreement occurred for groups 3 and 5.
The above statistics have included all grid points regardless of the
wind direction. Next using the same eight grid cell groups, we examined
model performance for individual groups experiencing incoming flow on a given
day. For example, if southwesterly winds predominated, we then analyzed the
daytime concentrations for group 2. Figure 20 shows the mean residuals for
each group for days having upwind flow. As we saw in Table 6, residuals
varied appreciably from day-to-day, ranging from -15% to +25%. However, 8 of
13 days experienced mean residuals less than 10%.
There were a few days when the model performance was admittedly
disappointing. In particular, day 191 showed a 25% overestimate in group 2.
22
-------
Although we lacked the resources to analyze every day in detail, we examined
this day in order to gain some insight on model performance.
The 1200 GMT surface weather map for 10 July 1985 (Julian day 191) gave
some indication as to why observed concentrations were lower than expected in
eastern Pennsylvania and western New Jersey. This map (Figure 8) shows a
mesoscale high over south central Pennsylvania which was apparently induced
by a cluster of thunderstorms. A stationary front extending from a low over
central Indiana stretched eastward to northern Virginia and merged into a
surface trough that stretched northeastward along the Atlantic seaboard. The
main weather feature was a combined cold and occluded front positioned along
the St. Lawrence Valley southward through the Ohio Valley.
To examine why the daytime ozone observations were underestimated in
group 2, we generated hourly plots from both the observations and the
estimates. Figure 21 shows the observed and estimated concentration fields
for 0800 EST (one hour after the 1200 GMT surface weather map). Two features
are given on the observed plot: weather data reported at each National
Weather Service surface observing site, and measured hourly ozone
concentrations. The circled ozone measurements denote the sites used in
generating the observed UAM boundary conditions for group 2. The 18 gridded
model values used in defining group 2's boundary conditions are enclosed by a
rectangle. At 0800 EST, estimates averaged more than 20 ppb higher than the
observed values. The weather data indicate that the surface trough is still
located along the Atlantic Seaboard. Ahead of this trough, winds were
generally from the south; behind the trough, winds were generally from the
west.
Moving to 1000 EST, estimates were still about 20 ppb higher than the
observations (Figure 22). Observed concentrations in northeastern
Pennsylvania were particularly low. This area was experiencing overcast
skies, temperatures around 22 C, and high relative humidity. This suggests
the likelihood of shower activity which may have caused lower ozone
concentrations by allowing increased vertical transport and reduced
photochemical production.
23
-------
By 1200 EST, thunderstorm activity had begun (Figure 23) as indicated by
the Wilmington (DE) observation. Overestimates by the model continued
throughout group 2. Although overcast and relatively cool conditions
predominated in eastern Pennsylvania, Atlantic City experienced hot, sunny
conditions as indicated by their temperature of 34°C. The estimated
concentration field shows a strong gradient across group 2, ranging from 79
ppb in the northwest to 102 ppb in the southeast. The observed
concentrations, in contrast, range from 40 ppb to 69 ppb. Observed
concentrations as high as 111 ppb were reported in central New Jersey but
these were not prescribed for use in the UAM/ROM interface methodology.
After investigating additional sources of meteorological data, we found that
Philadelphia reported a wind gust of 45 miles per hour. This information,
which was not available in the routine modeling data base, strongly suggests
the nearby presence of organized convective activity such as a squall line.
Figure 24 shows that by 1400 EST, shower activity had moved into
Atlantic City. The temperature in Atlantic City dropped over 5 C in two
hours and observed ozone concentrations in the area dropped over 50 ppb.
Easterly winds reported at both Philadelphia and Wilmington are probably
indicative of outflow from thunderstorm activity in central New Jersey.
Meanwhile over northeastern PA, skies are beginning to clear although ozone
concentrations remain low. Predictions in group 2 seem less affected by the
storm activity and are as high as 112 ppb.
By 1600 EST, (Figure 25) observed concentrations near Scranton had
rebounded to near modeled values (65 ppb versus 75 ppb). In the area near
Atlantic City, which was still experiencing shower activity, estimated
concentrations were as high as 125 ppb but observed concentrations were only
around 30 ppb.
The meteorological data we used in the ROM offer some clues as to why
the ROM overestimated concentrations in group 2. Figures 26 through 28 show
time series of selected meteorological parameters that were calculated for
grid cell (45,21), the cell for which modeled concentrations were underlined
24
-------
in each of the plots shown earlier. The simulated cloud cover and solar
fluxes in Figure 26 did show some effect of the observed cloudy conditions
but probably not enough. Cloud cover increased to more than 9054 by 1300 EST
but only slightly reduced the computed solar flux. The model layer heights
(Figure 27) show some increase, with layer 3 growing to 1700 m by 1100 EST.
However, the localized convective activity probably caused more vigorous
vertical and horizontal mixing than was simulated in the model. Horizontal
dilution in the model also was probably smaller than actual dilution as
indicated by the relatively low modeled wind speeds in Figure 28. Apparently
the small-scale (relative to the model) cluster of thunderstorms was not
captured in the overall interpolation of meteorological data.
This case study was noteworthy because it highlighted some of the ROM* s
limitations. ROM is a regional-scale model that was designed for application
in relatively benign, steady-state summertime conditions. The occurrence of
a mesoscale surface trough and localized thunderstorm activity resulting in
dynamic subgrid-scale atmospheric processes seems to have affected the ROM's
ability to estimate boundary conditions along portions of the UAM domain on
day 191. The results indicate that development of UAM boundary conditions
from either observed or modeled data should include careful examination of
the effects of mesoscaie meteorological conditions, which as we have shown,
can cause localized perturbations in ozone concentrations.
25
-------
TABLE 2. SUMMARY STATISTICS FOR ALL DAYTIME HOURLY AND
DAILY MAXIMUM OZONE CONCENTRATIONS (IN PPB)
FOR THE ROMNET REGION (JULY AND AUGUST 1985
DATA MERGED)
Cone .
type
95th
Mean Std. dev. Percentile Maximum
n obs. model obs . model obs . model obs . model
Daytime
hourly 40,534 55.2 64.5 26.8 19.2 102.0 101.1 219.0 169.2
Daily
Maximum 3,707 78.6 77.5 26.7 22.2 127.0 118.7 219 0 169.2
26
-------
TABLE 3A. COMPARISON OF THE JULY AND AUGUST 1985 EPISODES
(DAYTIME HOURLY STATISTICS, IN PPB)
95th
Mean Std. dev. Percentile Maximum
Episode n obs. model obs. model obs. model obs. model
July
1985 25,129 56.8 66.6 26.0 19.3 102.0 103.2 218.0 169.2
August
1985 15.405 52.7 61.1 28.0 18.7 102.0 97.0 219.0 161.3
TABLE 3B. COMPARISON OF THE JULY AND AUGUST 1985 EPISODES
(DAILY MAXIMUM STATISTICS, IN PPB)
95th
Mean Std. dev. Percentile Maximum
Episode n obs. model obs. model obs. model obs. model
July
1985 2,271 79.9 79.6 25.1 22.5 125.0 120.9 218.0 169.2
August
1985 1,436 76.6 74.1 29.0 21.3 131.0 114.1 219.0 161.3
27
-------
TABLE 4A. STATISTICAL SUMMARY BY GEOGRAPHICAL GROUP
FOR THE JULY 1985 EPISODE (DAYTIME HOURLY
CONCENTRATIONS, PPB).
Group
1
2
3
4
5
Mean
n obs . model
6206 60.2 78.4
6776 65.2 72.6
5902 54.3 57.3
3604 47.8 56.4
2641 45.2 57.9
Std. dev.
obs. model
27.8 21.9
27.5 19.0
23.7 11.8
20.1 12.8
19.5 11.3
95th
Percentile
obs. model
111.0 118.7
110.0 107.6
95.0 79.4
83.0 81.3
80.0 78.5
Maximum
obs . model
218.0 169.2
211.0 159.8
155.0 103.5
126.0 121.4
108.0 108.0
TABLE 4B. STATISTICAL SUMMARY BY GEOGRAPHICAL GROUP
FOR THE JULY 1985 EPISODE DAYTIME (DAILY
MAXIMUM CONCENTRATIONS, PPB).
Group
1
2
3
4
5
Mean
n obs. model
558 88.0 97.8
609 90.7 87.9
537 75.1 65.9
328 66.1 64.7
239 63.4 67.2
Std. dev.
obs. model
28.9 23.1
23.8 19.5
19.1 12.1
17.9 14.5
17.8 11.7
95th
Percentile
obs . model
138.0 144.9
130.0 120.8
109.0 88.0
96.0 91.7
94.0 89.3
Maximum
obs. model
218.0 169.2
211.0 159.8
155.0 103.5
126.0 121.4
108.0 108.0
28
-------
200
CL
O.
150
Q
LJ
I—
O
Q 100
LJ
LJ
-z.
o
M
O
50
I I I I I I I I I I O I /I I xl I I I I
5968 OBS. DATA PTS_
6321 PRED. DATA PTS-
I
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure 6a. Quantile-quantile plot of daytime hourly ozone for the July 1985
episode (Group 1, Northern Corridor).
200
_Q
Q.
Q.
150
Q
LJ
I—
U
Q 100
LJ
o:
Q_
o
M
o
50
6510 OBS DATA PTS
7203 PRED. DATA PTS
1 I 1 I I I I I I I \ I I I I I I I
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure 6b. Quantile-quantile plot of daytime hourly ozone for the July 1985
episode (Group 2, Southern Corridor).
29
-------
5735 OBS DATA PTS
632) PRED. DATA PTS
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure 6c. Quantile-quantile plot of daytime hourly ozone for the July 1985
episode (Group 3, Ohio Valley-Middle Atlantic).
200
_Q
Q-
CL
150
LJ
O
Q 100
LJ
o:
Q_
LJ
M
O
0
C
** / '
/ / /
/ / /
/A/
7 / /
j&/
~ +**^y/// 3474 OBS DATA PT^~
//' 3969 PRED. DATA PTS-
/ I 1 1 ! 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 ] | ! J_ j
) 50 100 150 200 250 300
OZONE (OBSERVED), ppb
Figure 6d. Quantile-quantile plot of daytime hourly ozone for the July 1985
episode (Group 4, Interior Northeast).
30
-------
200
_Q
CL
Q.
150
CJ
O 100
LJ
cr
CL
LJ
O
r\i
O
50
I [ I I
2543 OBS. DATA PTS^
3087 PRED. DATA PTS-
50
100
150
200
250
300
OZONE (OBSERVED), ppb
Figure 6e. Quantile-quantile plot of daytime hourly ozone for the July 1985
episode (Group 5, Great Lakes).
31
-------
300-
280-
260-
240-
220-
200-
^3 160-
z ,40-
S 120-
100-
80-
60-
40-
20-
o-
i
T
i
0
1
3
I
0
1
I
0 ,
1
1
I '
) '
0
1
>
'
> 1
3
1
C
> '
(
)
c
T
)
1
r
3 *
1
1 I ! 1 1 1 1 1 1 1 i 1 1 1 1 >
89 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204
Julian
Figure 7a. Comparison of
July 1985 episode
(Group
doy
observed (o) and
1,
(1985)
modeled
(x)
maximum ozone for the
Northern Corridor).
300-
280-
260-
240-
220-
200-
£ 18°-
3 160-
Z ,40-
S 120-
100-
80-
60-
40-
20-
o-
i
c
I :
i
0
89 190 191 192 19
3
C
3 194
T
! if
i iJ
1 1
C
t
\
J
3 >
1 .
X
1
c
1 i i i i i r 1 1 r1
95 196 197 198 199 200 201 202 203 204
Julian doy (1985)
Figure 7b. Comparison of
July 1985 episode (Group
observed (o) and modeled
2, Southern Corridor).
(x)
maximum
ozone for the
32
-------
300-
230-
260-
240-
220 -
200-
£ 180-
•S 160-
UJ
§ ,40-
S 120-
100-
80-
60-
40-
20-
o-
i
i!
!
1 °
3 I *
1
0
1
1
1
I
1
- ': ,
) X 0 C
1 i1
C
1
) *
1
i
1
J
1 1
'1
) 1
1
1
•I
I
1 1 1 1 1 1 1 1 1 1 1 1 ( 1 1 |
89 190 191 192 193 194 195 196 197 1 98 199 200 201 202 203 204
Julian day (1985)
Figure 7c. Comparison of observed (o) and modeled (x) maximum ozone for the
July 1985 episode (Group 3, Ohio Valley-Middle Atlantic).
300
280
260
240
220-
200-
3 160-
z 140-
o
o 120-
100-
80-
60
40-
20-
o-
i
I
0
i
>• I! !
i ;i |
i
0
)
1
<
) '
i:
| ,! I
i
0
1
89 190 191 192 193 194 195 196 197 198 199 2
Julian day
[1985)
Figure 7d. Comparison of observed (o) and modeled
July 1985 episode (Group 4, Interior Northeast).
1
T
I
i
1 1 1 1 !
00 201 202 203 204
(x)
maximum ozone for the
33
-------
300-
280-
260-
240-
220-
200-
^180-
Q.
~ 160-
bJ
Z 1 40 '
O
o 120-
100-
80 -
60-
40-
20-
o-
c
i
I
) 1
1
T
1
i
I 1 o- >
1 ' "
o1
1
1 1
1
189 190 191 192 19
1 ,[ I
I 0 ' 1 > 1
. 0 ' i >. t
1
i r •
i i i
)
I
,
[loir
0 x
ll
1
' o'
1
3 194 195 196 197 198 199 200 2
T
1 o , T I
1
1 o'
1
i i I
01 202 203 204
Juhan day (1985)
Figure 7e. Comparison of observed (o) and modeled (x)
July 1985 episode (Group 5,
maximum ozone for the
Great Lakes).
34
-------
m° <*2
-------
Observed July 9-11, 1985
Maximum Hourly Ozone ( p p b )
ROM 2 I J u 1 y 9- 1 1 , i 985
Maximum Hourly Ozone (ppb)
Figure 9. Spatial patterns of maximum ozone for July 9-11, 1985.
36
-------
July 13
July 14
July 15
Figure 10. Surface weather maps for July 13-15, 1985.
37
-------
Observed July 13-15 1985
Maximum Hourly Ozone ( p p b )
ROM 2.1 July 13-15, 1985
Maiimuin Hourly "Ozone (ppb)
Figure 11. Spatial patterns of maximum ozone for July 13-15, 1985.
38
-------
fj^^^w^v ^&
&?" '~j!^vj$ti$i\ • ^u ^V
~\«i.i'^i"«»rtJ5Hli5^./Bi'3 \zort Vi,o\ rJ*»loz A
>*•'$,.j}ir) \zozi Vi«\
^ fSP «|\
r <«or J^y'Z \
_!iUii',,.^-'f /l-^ia;
^
July 18
July 19
July 20
Figure 12. Surface weather maps for July 18-20, 1985.
39
-------
Observed July 18-20 1985
Maximum Hourly Ozone ( ppb )
ROM 2 I July 18-20, 1985
Maximum Hourly Ozone ( p p b )
Figure 13. Spatial patterns of maximum ozone for July 18-20, 1985.
40
-------
August 13
August 14
\
Figure 14. Surface weather maps for August 13-15, 1985.
41
-------
Air Qualily data Maximum Ozone ( P pb
observed A u g 13 1985 thru A u g 15 1985
ROM 2 I August 13-15, 1985
Maximum Hourly Ozone (p p b '
Figure 15. Spatial patterns of maximum ozone for August 13-15, 1985.
42
-------
TABLE 5. SUMMARY STATISTICS FOR UAM BOUNDARY CONDITIONS
(HOURLY DAYTIME OZONE, PPB, FOR THE JULY AND
AUGUST 1985 EPISODES)
Statistic
Modeled
Observed
Percent
Difference
n 20,155
Mean 60.9
Std. deviation 17.7
95th Percentile 92.9
Maximum 146.7
20,155
56.6
25.2
102.0
179.3
+7.6%
-29.8%
-8.9%
-18.2%
43
-------
TABLE 6. DAILY SUMMARIES OF UAM BOUNDARY CONDITIONS1
Date
(1985)
Mean ozone
(ppb)
model obs „
Mean wind
Percent
difference
Direction
(o)
speed
(ms'1)
Persistence2
July 8
July 9
July 10
July 11
July 12
July 13
July 14
July 15
July 16
July 17
July 18
July 19
July 20
July 21
46
59,
66
63.
61.
72.
64.
67
61.
57
56
74
70
53
4
.3
.8
.5
.0
.6
.5
.0
.7
.8
.2
.2
.8
.8
35.
70.
57.
45.
56.
70.
58.
44.
51.
46.
48.
79.
74.
60.
1
1
1
5
5
0
9
9
1
1
9
6
1
2
+32.2%
-15.4%
+17.0%
+40.0%
+8.0%
+3.7%
+9.5%
+49.2%
+20.7%
+25.4%
+14.9%
-6.8%
-4.5%
10.6%
250
171
233
320
165
187
215
200
169
35
70
222
265
177
5.2
3.8
4.6
4.7
4.6
3.1
5.9
5.2
3.0
4.4
3.7
5.9
4.4
4.1
0.89
0 72
0.80
0.91
0.80
0.52
0.89
0.94
0.73
0.83
0.22
0.94
0.85
0.55
August
August
August
August
August
August
August
August
8
9
10
11
12
13
14
15
46.2
54.9
55.9
52.3
56.6
59.7
69.2
80.1
42.
72.
52.
54.
40.
62.
69.
72.
3
0
1
5
3
5
7
7
+9.
-23,
+7,
-4.
+40,
-4,
-0,
+ 10.
.2%
.8%
.3%
.0%
.4%
.5%
.7%
.2%
221
110
168
200
2
159
222
331
4
8
3
4
4
4
3
5
.7
.7
.6
.0
.4
.1
.8
.4
0
0
0
0
0
0
0
0
.87
.48
.80
.80
88
.68
.70
.94
1. All land-based UAM boundry grid cells were used, for daytime hours only.
2. Persistence - (vector mean wind speed)/(scalar mean wind speed) Based upon
all daytime hourly wind observations from National Weather Service stations
near the OMNYMAP domain.
44
-------
MEAN DAIL^
30
20
10
0
-10
-20
C
RESIDUALS
- V
1-
1 ' '"
: • • •
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
PERSISTENCE
Figure 16. Mean residuals versus wind persistence for the UAM boundary.
MEAN DAILY RESIDUALS
30 r
20
10
-10
-20
180 210 240 270 300 330 360 30 60 90 120 150 180
WIND DIRECTION
Figure 17. Mean residuals versus daily average wind direction for the UAM
boundary.
45
-------
Figure 18. Division of the OMNYMAP boundary into eight groups.
CONCENTRE
100
90
80
70
60
50
40
30
20
10
0
VTIONS (PPB)
-
; *::^^^ ./^
\Jr^*^-^^"^
-
-
-
i i i i i i i i
12345678
MEAN
OBSERVED
MEAN
PREDICTED
UAM GROUP
Figure 19. Mean daytime ozone concentrations by UAM group along the UAM
boundary.
46
-------
£ io
e
< -io E-
Q
-20
i i i i
i i
2345
UAM GROUP
MEAN OVER ALL DAYS
Figure 20. Mean residuals for each UAM group experiencing incoming flow.
47
-------
/ J5 25 03
/ " >! D' " '
^ "!'r 4(
-------
41 41 41 SJ SI S3 II 71
31 41 4S 41 /SI SI II 71
Jl 4! 47 SI UU U 70 7j
Figure 22. Ozone concentrations for July 10, 1985 (1000 EST). Observed data
are in the top plot, and gridded model estimates are in the bottom plot.
49
-------
/ * ' 77 85 •
I 6, n i
} u ' ' i — i
23 ^ G4 \ Jr"
\3^&& 1 ,-*. 4V\r?*o-^ — A°^'
ft? 7 /?" 1^,7, iufy
SI 51 It II II 11 II 10 SI SI 5/1 SI 11 71 11 || II IS 11 11
SJ SI 51 II IS 10 17 Jl II 17 /SI 11 71 11 II 11 11 11 14 II
II SO II II Jl 17 n !l ii II Ul M 71 11 1! II 1! 100 II M
Jl II 41\_lO Jl 17 Jl II II St SI 71 IS 'YJJ ,j, 17 II IS M 15
II l« II 111 IJ 11 10 41 SO SI 71 IJ IJ |(7 ,00 II 17 11 fJK^/l
II 41 57 ll'. Si II II II 70 71 II |,t ,,, ,,, II 17 II 11 y/j'/\T&
17 57 51 II I?"1 57 11 II 71 IS ,,, IM IM ,(J 11 II If 7J_ji/74 51
O ^ - *j *s*-*~ — •» — ~~
IJ 15 17 II li 71 11 70 10 H,||i |4»..''lYl~~1T''ir 11^71 !^ 47 50
II 11 II 11/11 71 10 "|7 ._ 11 'D^rVTlll ** •'.' 'jfav^fl^-pl !l 50 17
II 17 II 7l> 7! 10 11 14 ^1 .'joV-YyT'^n ,,, l«~r!jt-''7j 55 II 44 11
II II _7I ^7» IJ II 11 (1JJ*»S_ 11 I'^-IT'J'^ffl l] " 41 Jl "
-------
SI 51 51 41 II I] It II 41 SI II SI 54 II
SI SI 50 II II 10 II 17 4D I] /II SO 51 77
41 41 41 II II 17 17 11 41 14
\_(l 41 II II II 44 17
71 71
II II
_U ll_
I! 14
IS 14
19 II
Figure 24. Ozone concentrations for July 10, 1985 (1400 EST). Observed data
are in the top plot, and gridded model estimates are in the bottom plot.
51
-------
/ ' ' 50 66
/ 70 5?
^ 1 ^Q/Xi, L
~\ ^ 55 Z6 JT i
„ o,B
41 II 41 It 17 II 41 11 II 1! il SO 11 17 71 71 71 11 11 71
17 II 41 |7 II 11 11 17 11 41/44 4! 41 11 7! 14 II 14 11 77
I
44 " " " 1) 11 > " M " "'" * ' 11 11 ' < II II H 11 71
4! 11 11 _ll 41 41 41 40 41 47 SI 1! 11 ' -t'l 17 Ii 10 14 M 71
41 41 4! Ii1' 17 II 1! 41 11 SI 11 71 II 17 17 II II 1] Ctpf ft
1! 41 10 S7\ 14 11 10 41 S5 II 10 11 ,04 |0, ,j, II II 11 \T L i>
14 _47 14 11 lT~^ll IS 11 II II 10 ,0, ,,j ,,, ,,, II MJ 7t_Vl! 4)
ii ~u ii ii i/ ii 41 ii 7i ,00 -in jn-i'sllTiTTor _,«'''"'! «>j, " 37
ii u ii ii ^ii ii 17 ~~7j , 10 'iij^-r'n in in ut' 'i^fj^ ^ 4I ]l Jl
14 II 11 11 71 II 14 14 ii .'itT-'YjI'VsTTTl I/T~r4o«-'" !l 1! 34 14
|— *• '•'/'' -
'i i-rTio n ss 17 ii u ji
10 71 ~n !*l 11 71 !J j^4 1 ~ftf~fri "Til 111 111 H II 1! Jl 33 11 11
14 10 II II -4^ 1C UJ Tltilll 117 II! 107 " '« '« SI 11 14 14 I]
H II II II II ii] H4 | n / I J | 1 1 0 ' 3 '1 ' « '3 1 ! 47 11 11 17 17
13 II ig, ll^.-n, HI III 1H? Ill in H >S 71 II 57 44 10 41 11 11
100 IJI__ UL] »11 111 II! Ill I4)| I17 l< *» '3 7J 14 it 4? 11 41 41 11
-nV IH/r'l«l 101 117 104 101 |/l " " " '! " !< '5 41 II ]« 11 ]|
7_y I4;/ II II M IJ. Ml^ij II 74 7! 17 11 41 41 41 31 If 11 ]7
1
Figure 25. Ozone concentrations for July 10, 1985 (1600 EST). Observed data
are in the top plot, and gridded model estimates are in the bottom plot.
52
-------
2500 r-
2000 -
1500 -
-, 1
X
a
o 1000-
solar flux
cloud cover
o
2
•o
§
HOUR
Figure 26. Hourly cloud cover and solar fluxes for grid cell (45,21).
HEIGHT (M)
Figure 27.
1800
1600
1400
1200
1000
800
600
400
200
°c
A, /\ top of Layer 3
_ -^ ^x top ot Layer 2.
\J \ ~~ -
- >/ \--*
top of Layer 1
....^ .-"-^
/
) 5 10 15 20
HOUR
Hourly layer heights for grid cell (45,21).
53
-------
Q
u:
UJ
a.
HOUR
Layer 1 wind speed
Layer 2 wind speed
Figure 28. Hourly wind speeds for grid cell (45,21),
54
-------
SECTION 5
DIAGNOSTIC TESTS
The earlier ROM diagnostic tests performed by Lamb and Laniak (1985)
began with a chemistry-only simulation and added processes from that point.
One of the problems with that hierarchy of tests is that the chemistry aspects
of the ROM are, perhaps, the most complex and non-linear of all processes
simulated. Testing those aspects first may well mask problems in numerical
algorithms designed to test simpler processes, such as horizontal transport
and vertical flux. The current series of diagnostic tests were designed with
a reverse hierachical order from the original series of tests. That is,
horizontal transport and vertical fluxes were tested first, and the chemistry
was activated only in the last test. This hierarchy proceeds from the simpler
to the more complex simulation processes.
The current series of tests were designated as Tests 1, 2, and 3, and
were designed to test numerical procedures for transport and vertical flux,
horizontal transport only, and transport and chemistry only, respectively.
Test 1 was further subdivided into a number of subtests, all of which are
described below. Later, an even simpler series of tests for horizontal
transport (Tests 0) were added.
DESCRIPTION AND RESULTS OF TESTS
In order to perform diagnostic tests on ROM2.1 and its most critical
processors under known atmospheric conditions, a specialized processor system,
STG4, was developed to allow the specification of certain wind field and model
layer patterns in an analytically defined atmosphere. By predetermining the
wind field and model layer depth patterns in time and three-dimensional space,
the capacity of the ROM and specific processors to obey various physical laws
may be tested.
55
-------
In the production version of the ROM processor network, sequential stages
of the processors are executed toward the preparation of the complete ROM
input file. Among the data types processed in the first four stages are
included all of the meteorological parameters such as wind, temperature,
density, cloud cover, etc. Also processed in these first four stages are the
definitions for all three of the ROM layers. All of these physical variables
and various others, as required by the ROM processors subsequent to stage
four, are defined by STG4.
Particular wind flow and layer definition sets for each test case are
defined in STG4. Routines there define the wind field and the vertical layer
boundaries, each of which may be dependent on the various physical quantities
that are analytically defined in STG4. Eight test cases have been defined and
incorporated into STG4.
The analytically defined atmospheres in STG4 were developed from the
ideal gas law and the atmospheric physical relationships defined by the
hydrostatic and hypsometric equations. Horizontal diffusion due to turbulent
eddies is set to zero for these tests. The temperature lapse rate is defined
to be adiabatic, which is a good approximation for the actual atmospheric
mixed layer where the ROM is designed to work. The only atmospheric
definitions which do not approximate the real atmosphere are the neglect of
the effect of water vapor on the specific gas constant for air and the
assumption of no horizontal diffusion due to turbulent motion. The neglect of
water vapor effects was necessary to achieve a tractable solution of the
atmospheric definition equations.
Test Case 1A
This case is designed to test the horizontal transport and interlayer
mass flux mechanisms in the model. For this test case, a circular
anticyclonic wind field, shown in Figure 29, is specified identically for all
three model layers. This wind field exhibits perfect solid body rotation on a
56
-------
spherical earth. Therefore, the divergence in the horizontal plane is zero
for all locations in the definition region. This aspect of the wind field
guarantees that any vertical mass exchange will be the result of changes in
the depth of the vertical layers, and not as the result of convergence or
divergence of the horizontal winds.
A temporally oscillating pattern of ROM layer depths was used to provide
the mechanism for vertical mass flux between ROM layers. The three layer tops
are kept horizontally flat at all times, with the heights of the top of layer
one and two oscillating up and down in time. The layer tops are defined by:
Zjd.J) - 500 + (300) [ sin [ (2* / 60,000) t ] ] m
Z2(I,J) - 1000 + (300)[ sin [ (2* / 60,000) t ] ] m
Z3(I,J) - 1500 m
where t is the elapsed time, in seconds, from the beginning of the simulation.
This layer top definition represents an up and down oscillation of layer two
itself, with its thickness remaining constant in time (500 m). The depth of
the full model domain is fixed over all space and time (1500 m).
For test case 1A, all chemical reactions are turned off in ROM2.1. Only
one chemical species, used as a mass tracer, is given a spatially-dependent
initial concentration that differs from the prescribed background
concentration. This species is given the following initial concentration
distribution:
C(I,J) - [ C(I0,J0) - B ][1 - (R / 2.5)] + B R < 2.5
C(I,J) - B R > 2.5
where I and J are the grid column and row indices, respectively, C(I0,J0) -
1 ppm is the initial species concentration at grid cell (I0,J0), B - 0.001 ppm
is the background concentration, and R is the distance, in grid cell lengths,
from the center of cell (I0,J0) to the center of cell (I,J). The background
concentration, B, is also used as the concentration value at inflow boundary
57
-------
cells.
The initial concentration distribution is applied to all three model
layers, with I0 and J0, for each layer, given by:
Layer 1: I0 - 18, J0 - 14
Layer 2: I0 - 30, J0 - 35
Layer 3: I0 - 42, J0 - 14 .
The initial concentration distribution for Layer 1 is depicted in Figure 30
for test case 1A. It should be noted that the initial concentration
magnitudes are completely arbitrary, because the chemical reactions are turned
off for this test case. The gradients in the concentration field and the
total mass are the significant factors in these tests. The concentrations can
change only in response to the atmospheric dynamics or to any numerical errors
associated with the ROM transport scheme. Concentration "clouds" identical to
the one depicted in Figure 30 are also initialized in ROM layers 2 and 3,
except that they are centered at different locations in the horizontal grid.
As an initial test of the adequacy of the numerical algorithms we will
examine the mass conservation within the model domain through the simulation
period for each of the tests. Ideally, total mass within the domain will
remain the same throughout the time period, despite the advective and vertical
flux processes occurring in Test 1A. The rate of angular rotation chosen for
the wind field in this test causes the mass field to make one complete
revolution in 100 time steps. Figure 31 presents a time history of the total
mass in the model domain through slightly more than one complete revolution.
The mass is represented as normalized mass; that is, the ratio of the mass in
the domain at some time ti compared to the initial mass field at t0,
M(t1)/M(t0).
There is an oscillation evident in the mass field over time. The
periodicity corresponds to that of the oscillation in the ROM layer depths for
this test. Mass increases as much as 6% in the model domain during the first
58
-------
third of the simulation. Later, a mass decrease of around 2% is evident. The
overall trend in the mass field is toward a decrease over time. The degree of
mass change is not particularly strong here, although the test does suggest
that mass conservation errors may occur when the model layer interfaces change
significantly over time.
Test Case IB
Test case IB is very similar to 1A, except that the layer heights vary
over space instead of over time. This case is also designed to test the
transport and numerical algorithms of the model. The wind field and the
initial concentration fields for the three ROM layers are the same for test IB
as they were for 1A.
The heights of the layer one and layer two tops are specified as
functions of the radial direction from the grid cell containing the wind field
rotation axis (30,21) to the grid cell centers. Starting with the radial line
to the north of the center of rotation, and moving clockwise, the top of layer
one decreases in height by an amount proportional to the sine of the compass
direction of the radial line. Thus, along a radial line pointing east from
the wind field rotation axis, the height of the top of layer one reaches a
minimum value. As the radial line sweeps around to the south, the height
returns to the original value. For grid cell locations with column indices
less than 30, the heights of the top of layer one are symmetrical with the
layer one heights at grid cell locations with column indices greater than 30.
A three-dimensional view of the cell heights in layer 1 is given in Figure 32.
The height of the top of layer two is defined in the opposite sense of
layer one, and is symmetrical with layer 1 at all locations. Layer 2 heights
are shown in Figure 33. The top of layer 3 is defined to be horizontally
flat. All layer heights are held constant in time for test IB. Analytic
expressions for the layer 2 top heights are given by:
Z^I.J) - 500 - (300) [ sin [ n/2 - tan-1(x/y ) ] m
59
-------
Z2(I,J) - 1000 + (300) [ sin [ w/2 - tan^C x/y ) ] m
Z3(I,J) - 1500 m
where x is the distance from the grid cell longitude to the axis of rotation
longitude, and y is the distance from the grid cell latitude to the axis of
rotation latitude.
The normalized mass ratio for the entire model domain during the
simulation of test IB is shown in Figure 34. The mass total remains within
close proximity of the original mass amount during the simulation. Maximum
changes of about 4% from the original mass can be seen.
Test Case 1C
This test case is designed to examine the accuracy of the numerical
transport scheme alone, without vertical fluxes occurring. The wind field and
initial mass fields for each ROM layer used here are identical to those used
for test case 1A. The layer top heights are horizontally flat and are
constant in time for test 1C. This allows for an examination of the ability
of the ROM to conserve mass during periods of rotational (non-divergent) flow
with no vertical transport across the layer interfaces. The layer top heights
are defined as follows:
Zi - 500 m
Z2 - 1000 m
Z3 - 1500 m.
The layer top heights act as material surfaces, resulting in no transport of
air across the layer boundaries.
The normalized mass ratio for test 1C is shown in Figure 35. We see
here that the total mass within the domain grows by nearly 7% in the first
third of the simulation, and then stabilizes later at a mass increase of
approximately 8% over the initial mass. Considering that there are no
60
-------
physical or chemical sources or sinks of mass in the model simulation, this is
a significant mass increase.
Test Case 2
This case is a somewhat more complex test of the numerical transport
algorithms in the ROM. For this test case, a purely divergent (convergent)
flow field is used. The layer average wind fields for each of the model
layers are identical, resulting in no vertical wind shear. A west-to-east
zonal wind field is defined for each layer such that speed maxima exist along
the western and eastern boundaries of the modeled region, and speed minima
exist along the central longitude of the modeled region. The speed of this
flow is defined for any grid cell location by the following equation:
U - Uw[0.75 + 0.25( cos(2wx/D) ) ]
where Uw is the latitude-dependent wind speed at the western edge of a
particular grid cell row, x is the distance of a given grid cell center to the
grid cell center along the western edge of the modeled region along a
connecting latitude line, and D is the distance from the grid cell center on
the western edge to the grid cell center on the eastern edge of the modeled
region along a connecting latitude line. This definition results in speeds at
the central longitude of exactly one-half the speeds at the western and
eastern edges of every grid cell row. There exists, therefore, convergent
flow in the western half of the model domain, and divergent flow in the
eastern half. This windfield is depicted in Figure 36.
It was desired for test case 2 that the layer boundaries represent
material surfaces, or surfaces across which there is no flux of material.
Thus, the layer heights must be determined such that, given the layer average
wind fields and the physical variables analytically defined in STG4, there is
no transport of air across the layer boundaries. This stipulation requires
that the layer heights be determined based on the defined wind field. The
resulting field of spatially varying layer heights for model layers 1, 2, and
61
-------
3 are shown in Figures 37, 38, and 39, respectively. Here, although there are
spatial variations in the layer heights, there is no flux of material between
layers.
The initial concentration field for test case 2, ROM layer 1, is shown
in Figure 40. It is similar to that shown in Figure 30 for test case 1, with
two exceptions. First, the cloud is centered at the east side of the grid to
allow for westerly zonal wind flow. Second, the peak concentration along each
row has been stretched from a width of 1 grid cell to 3 grid cells, resulting
in quasi-elliptical concentration cross sections instead of circular ones.
Identical mass clouds exist in layers 2 and 3. with their centers displaced
southward from the layer 1 cloud by 10 grid cells for layer 2 and 20 cells for
layer 3. Background concentrations outside the cloud mass are identical to
those for test case 1A.
The mass conservation results for test case 2 are shown in Figure 41.
As before, this plot presents a time history of the normalized mass
[M(t)/M(t0)] in the entire model domain during the simulation. While the
concentration cloud remains within the domain the normalized mass should
always be equal to 1.0. The cloud mass traverses the model domain from west
to east in 90 time steps. We see that during the first half of this
simulation the total mass increases by nearly 18% within the grid. This
corresponds to the region of convergent winds and increasingly deep layer
heights. As the cloud mass enters the region of divergent winds and
increasingly shallow layer heights the mass increases tend to level off and
eventually the mass begins to decrease. This decrease, however, occurs at a
much slower rate than the earlier mass increase, thus resulting in a net mass
increase of approximately 13% at the end of the simulation.
Results of this test show some disturbing features. Mass conservation
should be preserved in any numerical transport scheme. Results of test case 2
clearly show that mass is not being conserved here. Since the test
formulation was such that the model layer interfaces act as material surfaces,
there should be no vertical flux of mass. To attempt to isolate the mass
62
-------
increase phenomenon further, a simpler test (Test 0) was proposed which would
examine the numerical transport in a less divergent wind field.
Test Case 0
Test cases 1C and 2, described above, showed that the numerical
transport scheme produced significant mass increases for simulations of 90-100
time steps for purely rotational and divergent flow fields, respectively.
Since this was an unexpected and somewhat disturbing finding, we decided to
proceed to a test 0 series, instead of the planned test 3 (transport, vertical
flux, and chemistry). In the test 0 series we attempt to simplify the
transport environment to isolate the cause of the mass increase.
The ROM layer heights for test case 0 were spatially and temporally
constant over the model domain and were set at the values used in test case
1C. The wind field prescribed was "essentially" constant in space and time.
The winds were zonal from west to east with a velocity corresponding to:
U - (5/6) D/tc
where D is the distance from the western edge of the modeled region to the
eastern edge of the modeled region along the particular latitude, and tc is
the time required for a parcel of air to travel the distance D (180,000 s) .
The north-south extent of the model domain varies from 38°N to 45°N, and the
rms value of the wind speed over the domain is 5.6 m/sec, with very small
variations in wind speed from row to row. The wind field used in test 0 is
shown in Figure 42.
The initial concentration field for test 0 is the same as that specified
for test case 1A, except that the locations of the cloud centers are as
follows:
Layer 1: I0 - 5. J0 - 30
Layer 2: I0 - 5, J0 - 20
63
-------
Layer 3: I0 - 5, J0 - 10.
The initial concentration field for ROM layer 1 is shown in Figure 43.
Background and inflow boundary concentrations were again set to .001 ppb. The
concentration cloud traverses the model domain in just over 100 time steps.
The mass conservation time history for the model domain in test case 0
is shown in Figure 44. This plot shows an initial mass increase of about
0.33% per time step, while at the end of the simulation, the increase was
about 0.065% per time step. A total mass increase of about 10% was noted
toward the end of this simulation. This result is significant because the
test has isolated the numerical errors to the horizontal transport scheme,
without the complicating effects of vertical redistribution of mass.
Test Case OA
To determine the degree to which the noted mass increases from the
numerical advection algorithms were a function of the magnitude of initial
concentration gradients, a second test was performed in which the
concentration gradient was diminished. The concentration distribution in each
layer is given by:
C(I,J) - [ C(I0,J0) - B ][ 1 - (R / 9.5) ] + B R < 9.5
C(I,J) - B R > 9.5
Concentrations above the background value cover a much larger area within each
layer than the concentrations in test case 0. In order to assure that no mass
from the initial distribution within each layer is lost through the boundaries
of the modeled region during the simulation period, an adjustment is made in
positioning the initial cloud concentrations. The peak concentration
locations are placed farther away from the region boundaries than in test case
0. The locations are given by:
Layer 1: I0 - 12, J0 - 22
64
-------
Layer 2: I0 - 12, J0 - 20
Layer 3: I0 - 12, J0 - 18.
Figure 45 shows the initial concentration field in ROM layer 1 for test case
OA. The wind fields and layer heights for this test case are identical to
those used in test case 0.
The time history of the normalized mass within the model domain for this
simulation is shown in Figure 46. There are not as many time steps considered
during the simulation as there were in test 0 because the larger size of the
concentration cloud causes the mass to begin leaving through the outflow edge
sooner in this test. Nevertheless, it is clearly evident from the figure that
the mass is essentially conserved during the simulation, with a mass increase
of only 0.6% at the end of the simulation. The contrast between the results
of this test and test 0 demonstrate that the numerical artifact of the mass
increase is a function of the magnitude in concentration gradient. This
result is not surprising given the nature of finite difference approximations.
Test Case OB
In the final test case presented here we investigate the effect of
"clipping" negative concentration predictions from the model run. Many
numerical transport schemes, including the one used in the ROM, produce small
negative ripples of concentration near the edges of large concentration
gradients when these gradients are advected (McRae et al., 1982). The
negative concentrations are a result of the limitations of finite difference
approximations to large sub-grid gradients. The negative values, in
themselves, do not pose a problem for the advection routines. However, when
the advection solution is passed to the chemistry portion of an air quality
model there must be no negative concentrations since these are not defined in
a chemical simulation. Typically, as here, the negative values produced by
the numerical transport solution are "clipped", or set to either zero or a
very small positive number. In our case they were set to 10"16. Since we are
not solving the chemical equations thus far in these analytical tests it is
65
-------
possible to retain any negative concentrations resulting from the transport
simulation. In test case OB we have suppressed the negative clipping, and
allow the propagation of any negative concentrations.
The layer depths are the same as those used in test case 0, constant in
space and time. The wind field is also the same as in test case 0, simple
zonal flow. The initial concentration field is the same as that used in test
case 2, with sharp concentration gradients surrounding a flatter core. The
mass conservation results for this test are shown in Figure 47. The
normalized mass ratio indicates an initial increase of mass of about 0.5%,
dropping back to about 0.4% for the remainder of the simulation. This value
should be compared to the nearly 10% (and rising) normalized mass ratio at the
end of the test 0 simulation. It is apparent that the clipping of negative
concentrations has introduced a significant mass increase during the advection
of sharp concentration gradients.
Now that the cause of the mass increase has been isolated, the next area
of concern is the degree to which the mass imbalance affects actual ambient
simulations with the ROM model. From the testing performed above it has been
found that the degree of mass increase is a function of the concentration
gradients within an advected field. The tests have generally employed very
sharp gradients to stress the numerical algorithms. These gradients are
larger than those found in the real atmosphere for regional trace gases The
question we must answer is whether the mass increases effected by the
concentration gradients and the attendant chemical responses in ambient
simulations are serious enough to warrent consideration of a mass correction
algorithm to the numerical transport scheme.
We have made a first attempt to answer this question. A two-day ambient
ROM simulation has been performed for July 6-7, 1988, a particularly severe 0
episode in the Northeast U.S. The model domain used is the ROMNET domain
described in earlier sections. In addition to the base run, containing any
mass imbalances caused by the "clipping" of negatives, a simulation was
performed in which a first-order correction was made to the advected
66
-------
concentration field to assure mass conservation. The algorithm used for the
correction was based on a global assessment of "negative mass'1 clipped out of
the domain during an advection step for a single chemical species and ROM
level. This negative mass was then added back to the clipped field to
selected areas in proportion to the absolute value of local concentration
gradients. This latter simulation is referred to as the "corrected base run"
Grid differences, A-B, were constructed for each time step over each of
the two 24-hr periods of simulation of the base case simulation result (A)
minus the corrected result (B) for key chemical species. Normalized percent
differences, (A-B) x 100/A, were also calculated. Results are presented here
for ROM layer 1, where the largest grid differences were seen. For each day
there were 64 x 52 (grid cells) x 48 (time steps), or 159,744 grid
differences. Figure 48 presents frequency distributions of these differences
for 03 for each of the two days. Figure 48(a) shows the distribution of
residuals (ppb) and Figure 48(b) shows the distribution of normalized
residuals (% change from base). The figure shows that between the 5 and 95
percentile levels the absolute grid differences were less than 1.3 ppb and
less than 1.7%. Figures 49 and 50 present similar plots for NOX (N0+N02) and
ROG (carbon-weighted sum of reactive organics), respectively. These figures
show that the differences for ROG are also quite small, generally less than 1%
through most of the distribution. The results for NOX show the largest grid
differences, with the corrected version producing 8-9% less NOX than the base
run at the 95 percentile level, with 100% differences at the 100 percentile
level. (Note that the large negative percent differences at the 0 percentile
level are associated with very small concentrations.) The largest NOX
discrepancies are associated with large NOX point source emissions.
67
-------
7/7 7
7 7 7/
I , h h i,
uu
V \ \ \
i\ix
X^^
'/'/ / f
7 / / / /^ x •"
/////'"
111 r r ' ' ' '
7 /////'> >
'[ 1 1 1 T r r r >
1 J J J t
fit'
Mil
M ^
\ \ \ \ ^ r r
\ \ \ \ \ N ^N
\\\\NN^V
V\\\\\v^
V\\\\\vv
x,v\>vvv^
s*
f t
1 t
\ 1
\ ^
*^
^» -+
* -*
* •
t ®
« •
«. •-
*x •-
•^ •-
. .
• 4
tr . *-
ri
mi,,
4m
my,
.D8019^]PF015Z.DAT;i
Windfield for LAYER i — DATE & HOUR 80194:12
Grid stats ~ MAX: 26.30 @ row 1, col 60: RMS: 14.83 m/sec
Figure 29 Anticyclonic rotational wind field used in test cases 1A, IB, nncl
1C. The length of the largest vector corresponds to 26.3 m/s.
-------
VO
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
1)
12
11
10
9
8
7
6
3
4
3
2
1
1 2 3
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
111
1 1 1
1 1 1
111
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
|
1
|
1
1
1
1
1
1
1
1
1
1
1
,
1
'
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
'
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10
11
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
11
12
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
12
13
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
13
14
I
1
1
I
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
14
13
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
15
16
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
*
1
\f/i
w
'°P>
1
1
1
1
1
1
1
I
1
1
1
16
17
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
*
!£
?
<4(
^
5fc
i
i
i
i
i
i
i
i
i
t
i
17
18
1
1
1
1
1
1
1
1
1
I
I
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
it
OF
&
iOf
•or.
i
l
i
i
i
i
i
i
i
i
i
18
19
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
^
^
a)
U
&
t
i
i
i
i
i
i
i
i
i
i
19
20 21
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
. 1 1
vk i
A>| i
ifol 1
i i
i i
i i
i i
i i
i i
i i
i i
i i
i i
i i
i i
20 21
22
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
22
23
1
1
1
1
1
1
1
1
1_
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
23
24
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
*
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
24
23
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
|
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
25
24
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
26
27
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
27
28
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
28
29
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
29
30
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
30
31
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
31
32 33 34
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 t 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
111
1 1 1
1 1 1
111
1 1 1
1 1 1
1 1 1
1 I 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
1 1 1
32 33 34
35 36
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
* |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
35 36
37
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
*
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
37
38
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
I
1
1
|
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
38
39 40
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
• |
1 1
1 1
} |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
39 40
41 42
1 t
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
• I
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
41 42
43 44
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
t 1
1 1
1 1
1 1
1 1
t 1
1 1
1 1
1 1
1 1
1 1
1 1
| 1
1 1
1 1
| |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
43 44
45
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
|
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
43
46 47
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 t
1 1
1 1
1 I
1 1
1 1
1 I
1 1
1 1
1 1
1 1
| |
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
46 47
48 49
1 1
1 1
1 1
1 I
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
• 1
1 1
1 1
1 |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
48 49
50
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
•
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
50
51 52
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 t
1 1
1 1
1 1
1 t
1 1
1 1
I 1
1 1
1 1
| |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
51 52
53 54
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
j 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
53 54
55
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
!•
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
55
56
1
1
1
1
1
I
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
J
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
56
37
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
|
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
37
38
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
59
39
t
I
1
1
|
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
i
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
59
60
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
|
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
60
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
|J
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Figure 30. Initial species concentration field in ROM layer 1 for test case
1A. Units are ppm x 1000; contours are drawn at values of 100,
200, 400, 600, 800, 1000.
-------
TOTAL MASS RATIO: TEST CASE 1A
1
1
1
o 1
^ 1
a;
™ 1
LTI
1.
r^i
-• 1 .
<:
2
o 1-
z
1.
0.
0.
.20
.18
.16
. 14
. 12 1
. 10
.08
.06
.04
.02
.00
98
96
0
10
20
I
30
40
50 60
TIMESTEP
70
80
90
T
100 110
Figure 31. Time series of normalized mass ratio (M(t)/M(t0) ] within the
entire ROM domain for test case 1A.
-------
0.200E-t03
*2$DUA31rCRCROZONEl.080I91DMFOeSY.DAT:1
Data statistics: MAX = 500.0 h£AN = 275.5
data for 8019i:l2 plotted IVSEP-89.12:23
Perm Zl i> above msl Interface surface elevation
SIGMA = 79.57
Figure 32. Three-dimensional (x. y, z) schematic of height (m) of top of ROM
layer 1 for test case IB.
-------
$2$DUA31 rCRLROZOEi .D80191DMF066Y.DAT: 1
Data statistics: MAX = 1300. MEAN = 1225. SIGMA = 79.58
data for 80J9t:i2 plotted 1^SEP-89.12:30
Porm 22 n above msl Interface surface elevation
Figure 33. Three-dimensional (x, y, z) schematic of height (m) of top of ROM
layer 2 for test case IB.
-------
TOTAL MASS RATIO: TEST CASE 1 B
—I
OJ
o
\-
•X.
\rt
in
o
^
o
1.
1.
1.
1.
1.
1.
0.
0.
0.
0.
0.
0.
0.
25-
20-
15-
10-
05-
oo-
95-
90-
85-
80'
75-
70-
65 H
^-^ ^~^ ^— -~— _
, i 1 i I .,.,.,.,. i .,.,.,.,
0 10 20 30 40 50 60 70 80 90 100 11
TIMESTEP
Figure 34. Time series of normalized mass ratio [M(t)/M(t0) ] within the
entire ROM domain for test case IB.
-------
TOTAL MASS RATIO: TEST CASE 1C
1 . 22
1 . 20
1.18
o 1 . 16
< 1.14
^ 1 . 12
on
<
2 1 08
•\J
- 1 06
2
1 .04
o
z
1.02-
i.oo-
0.98
0
10
20
30
40
I ' l
50 60
TIMESTEP
I
70
1
80
r
90
100 110
Figure 35 Time series af normalized mass ratio [M(t)/M(C0) ] within the
entire ROM domain for test case 1C.
-------
NO INVERSION: $2$DUA3i:CRUROZONEI.DSOI94 DPFOISW.DAT:i
Windfield for LAYER i — DATE & HOUR 80194:12
Grid stats — MAX: 9.97 @ row i, col l: RMS: 7.39 m/sec.
Figure 36. Purely divergent wind field used in test case 2. The length of
the longest vector corresponds to 9.97 m/s.
-------
0.105E4O1
0.950E403
0.850E-K)3
0.750E403
0.650E403
0.550E403
0.450E403
O.lOSE+0-i
0.950E403
0.850E+03
0.750E+0?
0.650E405
0.450E403
«-2$DUA31: [RIROZCNE1 .D80191DMF065W. DAT: I
Data statistics: MAX = 1022. hEAN = 707.5
data for 8019TJ2 plotted P-MDV-89a5:-11
Parm Zl •> above msl Interface surfacfc elevation
SIGMA = 182.5
Figure 37. Three-dimensional (x, y, z) schematic of height (m) of top of ROM
layer 1 for test case 2.
-------
0.210E401
0.190E+04
0.170EW
0.150E404
o.uoe+01
0.900E+03
0.210E+01
0.190E+(H
0.170£+
-------
0.320E-KM
0.290E+04
0.260E+01
0.230E-KM
0.200E-KH
0.1706+01
0.140E-KM
0.320E-KH)
0.290E+01
0.260E+04
0.250E+01
$2$DUA3J:CRUROZCNE1.08019^DMF067W.DAT:1
Data statistics: MAX = 3222. hEAN = 2174.
data for 80191:12 plot tod 6-NOV-89.15:i7
Perm Z3 ri above msl Interface surfact elevation
SIGMA = 600.7
Figure 39. Three-dimensional (x, y, z) schematic of height (m) of top of ROM
layer 3 for test case 2.
-------
42
41
40
39
38
37
36
33
34
33
32
31
X
29
28
27
26
25
24
23
U
21
20
19
IB
17
16
15
14
13
12
II
10
9
8
7
6
5
4
3
2
1
11111
11111
1 1 1 1 1
11111
11111
11111
1 1 1 1 I
11111
11111
1 1 1 1 1
i yyyf£y»nn
i M wWScJSni
i nif^gJQ»-6m-6
i i i i i
11111
i i i i i
11111
i i t i i
i i i i t
i i i i i
11111
11111
i i i i i
i i i i i
11111
11111
i i i i i
i i i i i
i i i i i
11111
i i i i i
11111
i i i i i
i i i i i
i i i i i
11111
i i i i i
11111
i i i i i
i i
i i
i i
i i
i i
i i
i i
i i
i i
i i
uTlUL
w
H£
MJUf
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1
1
1
1
1
1
1
1
1
1
1
h
III
¥
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10
11
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
11
12
1
1
1
1
1
1
1
1
1
1
1
1
1
1
i
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
12
13
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
13
14
1
1
1
1
1
1
1
1
1
1
1
1
1
1
i
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
14
U
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
t
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
U
16
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
16
17
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
17
18
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
I
1
1
I
1
1
1
1
1
1
1
1
18
19
1
1
1
1
1
1
1
1
1
1
1
|
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
19
20
1
1
I
1
1
1
1
1
1
1
1
1
I
J
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
20
21
1
1
1
1
1
1
1
1
1
1
1
1
J
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
21
22
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
22
23
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
23
24 25
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
* |
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
24 25
26
1
1
1
1
1
1
1
1
1
1
1
1
1
i
1
1
I
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
26
27 28
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
j i
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
27 28
29 30 31 32
1111
1111
1111
1111
1111
1111
1 1 1 1
1111
1111
till
1111
1111
1111
1111
till
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
till
1111
1111
1 1 1 1
till
1111
1111
1111
1111
1111
29 30 31 32
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
1 1 1 1 t 1 1 1 1 1 1 1 1 1 1 1 1
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
1 t 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
Illllllllllllllll
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
30 51
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
30 31
32 53
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
t 1
1 1
52 33
54
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
54
55 56
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 I
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 |
1 1
55 56
57
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
37
f
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
58
59
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
59
60
1
1
1
1
1
1
1
1
.1
1
I
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
60
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
II
10
9
8
7
6
3
4
3
2
1
Figure 40. Initial species concentration field in ROM layer 1 for test case
2. Units are ppm x 1000; contours are drawn at values of 100,
200, 400, 600, 800, 1000.
-------
TOTAL MASS RATIO: TEST CASE 2
CO
O
r-
1 . 22
1 . 20
1. 18
1 . 16
* 1 . 14
£ 1-12
15
O 1.10
- 1 .08
I
-------
NO INVERSION: $2$DUA3i:CRUROZONEI.
Windfield for LAYER 1 — DATE & HOUR 8019^:12
Grid stats — MAX: 5.87 6 row 1, col i: RMS: 5.59 m/sec.
Figure 42. Quasi-constant wind field used in test cases 0, OA, and OB. The
length of the largest vector corresponds to 5 87 ni/s .
-------
42
41
40
»
w
37
U
33
34
35
32
31
30
79
28
J7
16
u
24
13
2?
II
20
19
ia
I)
16
13
14
13
12
II
10
»
I
7
6
5
4
3
1
I
1
I
I
I
1
I
1
1 1
1 1
1 1
1 1
1 I I
I I 1
I I I
1 1 I
1 I 1
1 1 I
1 I
1 1
1 1
1 1
1 1
1 1
1 I
1 t
1 1
1 1
1 1J
1 1111
111111
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
111111
1 1 1 1 1 1
1 ! 1 1 1 1
£S353')K > ' '
Figure 43. Initial species concentration field in ROM layer 1 for test case
0 Units are ppm x 1000; contours are drawn at values of 100,
200, AGO, 600, 800, 1000.
-------
TOTAL MASS RATIO: TEST CASE 0
oo
CO
1.12
1.11
1.10
o 1.09
< 1.08
a;
•f 1 .07
^ 1.06
M 1-05 1
- 1 .04
<
* 1.03
z
1.02
1.01
1.00
T~
0
—, , , , 1 , 1 1 p—
10 20 30 40 50
TIMESTEP
60
70
~~r~
80
90
Figure 't4. Time series of normalized mass ratio [M( t)/M( t0) ] within the
entire ROM domain for test case 0.
-------
ti
-------
TOTAL MASS RATIO: TEST CASE OA
1.12
1.11
1.10
o 1.09
< 1.08
a;
••f* 1.07 -
^ 1.06-
£ 1.05
-1 1 .04
<
i i-03
z
1.02
1.01
1 .00
r
0
10
20
~i ' r~
30 40
TIMESTEP
50
60
T
70
Figure 46. Time series of normalized mass ratio (M(t)/M(t0) ] within the
entire ROM domain for test case OA.
-------
TOTAL MASS RATIO: TEST CASE OB
CO
cr>
o
1-
LD
2
O
_J
ai
o
1
1
1
1
1
1
1
1
1
1
1
1
1
.12
. 11
.10
.09
.08
.07 -
.06 -
.05 -
.04 -
.03 -
.02-
.01 -
.00-
/
1 i ' i • i • i ' i ' i • i • i • i
0 10 20 30 40 50 60 70 80 9C
TIMESTEP
Figure kl Time series of normalized mass ratio [M(t)/M(t0)j within the
entire ROM domain for test case OB.
-------
.0
a
a
o
O
I
0
I)
o
m
•^
10
o
0
n
o
m
^->
13
0
u
0
u
0
u
n
o
CD
iO
O
5
4. -
3 -
2 -
1 -
0
-1 -
-2 -
-3 -
-4 ^
-5 -
-6 -
-7 -
-8
15
10 -
•5 -
0
-5 -
-10 -
-15 -
-20 -
-25 -
c
-30 -
-35
20
July 5, 1988
40
Percentile
60
+ July 7, 1988
80
100
20
D July 6, 1988
40
Percentile
60
+ July 7, 1988
80
100
Figure 48. Frequency distribution of paired grid 0, concentration differences
from ROM simulations for a base case (A) and a mass-corrected base
case (B) for July 6 and 7, 1988. Top figure (a) shows A-B
distribution and bottom figure (b) shows normalized differences,
(A-B)*100/A, as percent change from base.
87
-------
JO
a
a.
o
o
O
o
CD
0
o
I
0
CO
o
o
o
z
7 -
6 -
5 -
2 -
1 -
-1
100
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0
-10
20
D July 5. 1988
40
Percentile
60
+ July 7, 1988
80
100
20
D July 5, 1988
40
Percentile
60
+ July 7, 1988
80
100
Figure 49 Frequency distribution of paired grid N0x concentration diff-
erences from ROM simulations for a base case (A) and a mass-
corrected base case (B) for July 6 and 7, 1988. Top figure (a)
shows A-B distribution and bottom figure (b) shows normalized
differences, (A-B)*100/A, as percent change from base.
88
-------
a
a
o
0
I
«
n
o
CD
o
<
O
m
O
O
o
«
7 -
4 -
3 -
2 -
1 -
-2
10
5 -
-5 -
-10 -
-15 -
-20 -
-25
0 20
O July 6. 198S
40
Perc entile
60
+ July 7, 1988
80
100
20
D July 6, 1988
40
Percentile
60
+ July 7. 1988
80
100
Figure 50. Frequency distribution of paired grid ROG concentration diff-
erences from ROM simulations for a base case (A) and a mass-
corrected base case (B) for July 6 and 7, 1988. Top figure (a)
shows A-B distribution and bottom figure (b) shows normalized
differences, (A-B)*100/A, as percent change from base.
89
-------
SECTION 6
SUMMARY AND RECOMMENDATIONS
Using both an ambient evaluation and a series of diagnostic tests, we
evaluated Version 2.1 of the Regional Oxidant Model (ROM2.1). In the ambient
evaluation, we assessed ROMZ.l's performance for periods of high ozone in
July and August of 1985 in the northeastern U.S., using AIRS daytime hourly
surface ozone monitoring data. We compared these observations with model
estimates in three types of analyses: (1) a comparison of overall statistics
to determine whether model estimates exhibited a general bias, (2) a
comparison of spatial patterns of maximum concentrations to look for spatial
bias in the model estimates, and (3) an assessment of the model's
applicability for determining UAM boundary conditions. In the diagnostic
tests, we assessed the accuracy of numerical algorithms by evaluating the
model's ability to conserve mass; we performed five tests that involved only
the horizontal transport algorithm and two that involved both horizontal
transport and vertical flux.
In the ambient evaluation portion of the study, we found good overall
agreement. For a 26 day simulation, mean concentrations of the modeled and
observed daily maximum concentrations agreed to within I'/.. Concentrations at
the higher ends of the frequency distributions were slightly underestimated;
the 95th percentile observed daily maximum concentration was 127 ppb while
the estimated concentration was 119 ppb. The tendency to underestimate peak
concentrations is to be expected with a coarse grid model such as the ROM
because of the spatial averaging that occurs with Eulerian grid computations.
In the Northern Corridor and Southern Corridor geographical groups
(groups 1 and 2, respectively), model performance was good, particularly in
group 2. The group 1 mean observed and modeled daily maxima agreed to within
11% and the 95th-percentile observed and modeled daily maxima agreed to
within 5% (both values were overestimated). For group 2, the mean daily
90
-------
maxima were within 3% of each other and the 95th-percentile values were
within 7% (both-values were underestimated). The quantile-quantile plots of
observed and modeled daytime hourly concentrations showed the same kinds of
tendencies: overestimation in the upper quantiles of group 1 and
underestimation in the upper quantiles of group 2, as well as better overall
agreement for group 2 than for group 1—only the top 15% of group 1's
estimates were within 10% of the observations, while for group 2 the top 707.
of the estimates were within 10% of the observations. The medians in the
time series plots of daily maxima for these two groups showed analagous
underestimate-overestimate tendencies, and these plots also showed that
exceedances (values over 120 ppb) were overestimated in group 1 (ten versus
eight) and underestimated in group 2 (six versus ten).
Model performance in group 3 (the southwestern part of the domain—
southern Virginia, West Virginia, Ohio, and western Pennsylvania), an area
removed from the extensive metropolitan area of the Northeast Corridor, was
noticeably poorer than the performance for groups 1 and 2. The group 3 mean
daily maximum was underestimated by 12% and the 95th-percentile daily maximum
was underestimated by 19%. The quantile-quantile plot for group 3 showed
that the top 35% of the values were underestimated by more than 10%. The
group 3 time series plot showed that 12 out of 14 medians were
underestimated, and that exceedances were dramatically underestimated (six
versus zero). Underestimates of upper-quantlie concentrations in this group
perhaps can be attributed to the relatively small-scale urban plumes and
uncertainties in estimating naturally-occurring emissions of NO and
X
hydrocarbons. However, deficiencies in anthropogenic emissions inventories
should not be ignored, and ongoing efforts to improve them should continue.
Groups 4 and 5 (the northern part of the domain, excluding the Northeast
Corridor) had observations close to background values, and model estimates
generally showed good agreement with observations, especially for the upper
values.
Spatial patterns of the three-day maximum concentrations usually showed
reasonable model performance. The modeled magnitude and orientation of ozone
91
-------
plumes in the northern portion of the Northeast Corridor, especially around
New York City, compared well with observed plumes. The model also did an
excellent job of predicting high ozone levels around coastal sections of
Maine. We believe that ROM2.1 performed better than ROM2.0 in these areas
because ROM2.1 includes a correction for the westerly bias that occurred in
ROMZ.O's low-level wind flows. Differences between observed and modeled
plumes were most evident during episodes experiencing coastal troughs and
squall lines. Underestimates in the Washington, DC, area that were reported
in the ROM2.0 evaluation by Schere and Wayland (1989) were seen again in our
evaluation. In addition, ROM 2.1 tended to underpredict rural peak
concentrations of ozone by about 20 ppb. Few monitoring data are available
for evaluating model performance in rural areas, so it is hoped that
measurements being taken under the auspices of the National Acid
Precipitation and Assessment Program (NAPAP) will aid in future evaluation of
ozone concentrations.
The most rigorous portion of this analysis was an evaluation of the
model's ability to estimate boundary conditions for UAM application.
Although we do not recommend applying the ROM in a deterministic manner, some
type of reliable estimation scheme is needed for prescribing boundary
conditions when the UAM is to be applied for future-year emission control
strategies. Also, model estimates are need because measurements aloft are
typically not available for specifying the vertical structure of pollutant
concentrations. Furthermore, our analysis showed that using monitoring data
for estimating boundary conditions requires a great deal of subjectivity and
is fraught with uncertainty.
Overall, ROM2.1 estimates compared quite well with the monitoring data
for estimating UAM near-surface boundary conditions in the OMNYMAP domain.
The model overestimated the mean concentration for all daytime hours by just
5 ppb (8°/.) and underestimated the 95th-percentile value by 9 ppb (9%).
However, we saw significant day-to-day variability in model performance,
ranging from a maximum underestimate of 24% to a maximum overestimate of 49%.
A case study performed for July 10, 1985, demonstrated that small-scale
meteorological features can cause dramatic effects on model performance,
92
-------
apparently because such features are not captured in the ROM's overall
interpolation of meteorological data. We found that a squall line resulted
in a significant overestimate in ozone concentrations along UAM's western
boundary (the inflow boundary on that day). We therefore caution UAM users
to carefully review ROM estimates and be aware of mesoscale flow conditions.
On most days, however, ROM2.1 did a reasonable job estimating UAM boundary
conditions.
In the second part of our evaluation, we employed a series of diagnostic
tests to assess the model's ability to conserve mass. Table 7 presents a
summary of the mass conservation results for seven test cases. For all
individual layers and for all layers combined, the table shows the maximum
mass increase and maximum mass decrease (if mass decreased) that occurred
during the simulation. For test cases 1A and IB, in which mass flux was
allowed between layers, mass changes were expected to occur in individual
model layers. For the other tests, however, no change in mass was expected
for individual layers. For all tests, there should have been no change in
mass for all layers combined (total domain).
The table shows that for most test cases there were serious departures
from mass conservation. Only in test case OA (where the concentration
gradients were considerably relaxed) and test case OB (where negative
clipping was suppressed) was the total domain mass effectively conserved.
The results for the other five test cases suggest a potential problem with
the ROM's numerical transport procedures, despite earlier design tests
performed during the ROM's developmental stages (Lamb, 1983) that showed no
problems with mass conservation. We have delayed further diagnostic tests,
including examining test case 3 with chemical simulation, and further
analysis, including preservation of peak concentrations, until the mass
conservation problem has been corrected.
A first-order correction algorithm has been developed based on a global
assessment of the mass imbalance due to the clipping of negative
concentrations. This algorithm was implemented and tested on a two-day
ambient simulation with ROM during a high 0 concentration period. Results
93
-------
have shown that the differences between the corrected and unconnected
simulation results were almost always less than 1%. Differences seen in the
NO concentrations were greater than those of the other species, but
X
generally under 10%. These results suggest that the implementation of a
mass-correcting scheme in ROM's numerical advection algorithm would be a
desirable, although probably not essential feature. We plan to perform
further tests to assess the degradation in computation time with the
inclusion of the mass correction scheme. With this additional information,
we will weigh the improvements in accuracy of the transport solver with the
increases in computation time. We will also repeat the diagnostic tests
discussed in this section with the global mass-correction algorithm in place
and analyze the results in detail. Our analysis has demonstrated the value
of this type of diagnostic testing in model evaluation.
Our evaluation has suggested that further improvements to the ROM are
warranted. We are working towards improving the specification of layer
thicknesses and the computation of naturally-occurring emissions. In future
years, we hope that a dynamic meteorological processor can be incorporated
that will simulate nonsteady state flows. To continue making advances in
model development, additional monitoring data are needed for examining other
chemical species (such as NO , isoprene, formaldehyde, and HNO ) and for
X J
fully evaluating model performance in rural areas.
94
-------
TABLE 7. MAXIMUM PERCENTAGE CHANGES IN TOTAL MASS DURING A SIMULATION
ROM
LAYER
1
2
3
Total
Domain
lAa
59
-59
5
-1
72
-59
6
-2
.20
.00
.94
.23
.00
.25
.12
.55
lBa 1C3
10
-9
16
-3
1
-28
3
-3
.58 9.41
.75 -0.39
.40 13.30
.73
.91 9.07
.27
.80 8.13
.39
TEST CASE
2 0 OA OB
14.14 9.98 0.60 0.40
17.22 9.98 0.60 0.40
21.60 9.98 0.60 0.40
17.57 9.98 0.60 0.40
a Values shown include maximum positive and maximum negative changes from the
initial mass. Other tests only produced positive changes.
95
-------
REFERENCES
Godowitch, J. and K. Schere. 1989. Plans for the Development of a ROM/UAM
Interface. Internal Report, U.S. Enviromental Protection Agency, Office of
Research and Development, Research Triangle Park, NC.
Lamb, R. 1983. A Regional Scale (1000 km) Model of Photochemical Air
Pollution: Part 1. Theoretical Formulation. EPA-600/3-83-035, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Lamb, R. 1984. A Regional Scale (1000 km) Model of Photochemical Air
Pollution: Part II. Input Processor Network Design. EPA-600/3-84-085, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Lamb, R. and S. Hati. 1987. The representation of atmospheric motion in
models of regional-scale air pollution. Journal of Climate and Applied
Meteorology, 26(7): 837-846.
Lamb, R. and G. Laniak. 1985. A Regional Scale (1000 km) Model of
Photochemical Air Pollution: Part III. Tests of the Numerical Algorithms.
EPA-600/3-85-037, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
McRae, G., W. Goodin, and J. Seinfeld. 1982. Numerical solution of the
atmospheric diffusion equation for chemically reacting flows. Journal of
Computational Physics, 42(1): 1-42.
Pagnotti, V. 1987. A meso-meteorological feature associated with high ozone
concentrations in the northeastern United States. Journal of Air Pollution
Control Association, 37(6): 720-722.
SAS Institute Inc. 1985. SAS User's Guide: Basics, Version 5 Edition.
SAS Institute Inc., Cary, NC.
96
-------
Schere, K. and R. Wayland. 1989. EPA Regional Oxidant Model (ROM 2.0):
Evaluation on 1980 NEROS Data Bases. EPA-600/x-89-xxx, U.S. Environmental
Protection Agency, Research Triangle Park, NC.
U.S. Environmental Protection Agency. 1987. Application of the Urban Airshed
Model to the New York Metropolitan Area. EPA-450/4-87-011, Office of Air
Quality Planning and Standards, Research Triangle Park, NC.
U.S. Environmental Protection Agency. 1988. Anthropogenic Emissions Data for
the 1985 National Acid Precipitation Assessment Program (NAPAP) Inventory.
EPA-600/7-99-022, Air and Energy Engineering Research Laboratory, Research
Triangle Park, NC.
U.S. Environmental Protection Agency. 1988. Protocol for Regional Ozone Model
for Northeast Transport. Office of Air Quality Planning and Standards,
Research Triangle Park, NC.
U.S. Environmnetal Protection Agency. 1989. Aerometric Information and
Retrieval System (AIRS) User's Guide. Office of Air Quality Planning and
Standards, Research Triangle Park, NC.
Vaughan, W. 1985. Transport of Pollutants and PEPES. EPA-600/3-85-033, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Young, J., M. Aissa, T. Boehm, C. Coats, J. Eichinger, D. Grimes, S.
Hallyburton, W. Heilman, D. Olerud, S. Roselle, A. Van Meter, R. Wayland, and
T. Pierce. 1989. Development of the Regional Oxidant Model Version 2.1.
EPA-600/3-89-044, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
97
-------
APPENDIX
STATISTICAL SUMMARIES FOR THE AUGUST 1985 EPISODE
98
-------
TABLE A-l. Summary statistics of hourly daytime ozone concentrations (ppb)
for the August 1985 episode.
Region n
1
2
3
4
5
3606
4095
3522
2616
1566
Mean
obs. model
57.
60.
48.
45.
43.
7
1
3
6
4
71.
64.
54.
53.
55.
6
8
0
9
0
Std.
obs.
33.1
30.0
22.7
21.7
21.5
dev.
model
23.4
18.6
12.3
12.9
10.9
95th
percentile
obs. model
119.0
110.0
85.0
84.0
8l! 1
114.2
99.2
73.7
76.7
' 72. 6
Maximum
obs. model
219.0
188.0
135.0
137.0
122.0
161.3
129.1
99.6
127.0
116. 1
TABLE A-2. Summary statistics of daily maximum ozone (ppb)
for the August 1985 episode.
Region n
1 334
2 385
3 330
4 241
5 146
Mean
obs. model
88.
85.
68.
65.
64.
3
4
1
4
2
91.2
80.1
62.8
62.7
63.8
Std.
obs.
37.
27.
19.
20.
20.
7
7
6
8
1
dev.
model
23.3.
19. 1
12.2
16. 1
11.2
95th
percentile
obs. model
167.
136.
98.
100.
103.
3
0
0
9
3
141.4
112.7
82.7
96.1
81.7
Maximum
obs. model
219.0
188.0
135.0
137.0
122.0
161.3
129. 1
99.6
127.0
116. 1
99
-------
3571 DBS. DATA PTS
3680 PRED. DATA PTS
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure A-la. Quantile-quantlie plot of observed and modeled hourly ozone
for the August 1985 episode (Group 1, Northern Corridor).
200
4093 DBS DATA PTS
4416 PRED. DATA PTS-
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure A-lb. Quantile-quantile plot of observed and modeled hourly ozone
for the August 1985 episode (Group 2, Southern Corridor).
100
-------
200
_Q
CL
O
LJ
^—
cj
O 100
LJ
Q_
LJ
O
M
O
50
0"-
0
3539 OES. DATA PTS
3587 PRED. DATA PTS-
i i i i I i i i i I i i i i I i i i i I i i i i l i i i
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure A-lc. Quantile-quantile plot of observed and modeled hourly ozone
for the August 1985 episode (Group 3, Ohio Valley-Middle Atlantic).
200
.D
Q.
Q_
150
Q
LJ
I—
CJ
O 100
LJ
o:
Q_
LJ
Z
O
M
O
50
50
100
150
2591 DBS. DATA PTS
2760 PRED. DATA PTS-
200
250
300
OZONE (OBSERVED), ppb
Figure A-ld. Quantile-quantile plot of observed and modeled hourly ozone
for the August 1985 episode (Group 4, Interior Northeast).
101
-------
200
-O
Q.
Q.
150
LJ
I—
CJ
Q 100
LJ
cr
Q.
LJ
Z
o
M
O
50
1577 DBS. DATA PTS^
1656 PRED. DATA PTS-
50 100 150 200 250
OZONE (OBSERVED), ppb
300
Figure A-le. Quantile-quantile plot of observed and modeled hourly ozone
for the August 1985 episode (Group 5, Great Lakes).
102
-------
300-
280
260-
240-
220-
200-
180-
160-
s
100
80
60-
40-
20-
0
0 x
o *
219 220 221 222 223 22* 225 226 227 228
Julian day (1985) •_"
Figure A-2a. Comparison of observed (o) and modeled (x) maximum ozone for
the August 1985 episode (Group 1, Northern Corridor).
300-
280-
260-
240-
220-
200-
1'80-
^ 160-
z 140-
o '20-
100-
80-
60-
40"
20-
o-
2
c
,'
1
)
T 1 1
19 220 22
> I
X
c
1
(
)
-
c
t
t
i i i i i i I
222 223 224 225 226 227 228
Julian day (1985)
Figure A-2b. Comparison of observed (o) and modeled (x)
the August 1985 episode (Group 2, Southern Corridor).
maximum ozone for
103
-------
500-
280-
260 •
240-
220-
200-
^ 180-
3 160-
UJ
Z ,40-
3 120-
100-
80-
60 -
40-
20-
0-
2
19
> X
1
\
•>
c
1 '
) 1
1
)
<
1 [
i
i
r
1
,1
1
1 I 1 ) 1 1 1 I !
220 221 222 223 22* 225 226 227 228
Julian day ('1985)
Figure A-2c. Comparison of observed (o) and modeled (x) maximum ozone for
the August 1985 episode (Group 3, Ohio Valley-Middle Atlantic).
300-
280-
260-
240-
220 •
200-
^o ,80-
3 160-
LJ
Z ,40-
3 120-
100-
80-
60 •
40-
20-
o-
2
} 1
0 " 0
0
'
i i r i
19 220 221 222
Jul
)
-
[ T
i 1
TO o
1 1
0 1
~p 1
223 224
1
0
— 1 j
225 226
"i r
227 228
on doy (1985)
Figure A-2d. Comparison of observed (o) and modeled (x)
the August 1985 episode (Group 4, Interior Northeast).
maximum ozone for
104
-------
300-
280-
260-
240-
220-
200-
£180-
3 160-
LJ
Z 140-
S 120-
100-
80-
60-
40-
20-
o-
2
1 i
I
1 i 0 0 T
fll i: M 1:
:
1 1 1
19 220 221
' o * o I 1 r
1 ^ J ' o I
1
222 223 224 225 226 227 228
Julian day C'l985)
Figure A-2e. Comparison of observed (o) and modeled (x) maximum ozone for
the August 1985 episode (Group 5, Great Lakes).
105
------- |