The next call will be Thursday
(not Wednesday!) the 22nd of February at 2pm EDT.
Meeting Jan 31 - Feb 1 2007,
Chicago
Attendees:
Illinois: Buzz
Asselmeier
Indiana:
Michigan: Dennis McGeen
an Allan Ostrander (phone)
Minnesota: Chun Yi
New York: Carlos
Mancilla
Ohio: Tom Velalis
Ontario: Peter Wong
Pennsylvania: Karen Gee
Wisconsin: Orlando Cabrera-Rivera
GLC: Jon Dettling and Pete Giencke
EPA: Suzanne King, Steve Rosenthal
and Erin Newman
Project's software consultant: Mark
Young
Server update
The GLC has resolved the issues
with the server that crashed recently and the database instances have been
restored and should be available to connect to through ODBC using the same
protocols that would have worked prior to the crash. If anyone has any difficulties,
they should let the GLC know. The only component not yet working is the ArcSDE
instance that ran the mapping applications on CAROL. For the time being, the
mapping capabilities will not be working. The GLC is currently assessing
whether it is more efficient to work on restoring that system or focusing on
development of similar capability on their new server.
The new server should be set up
for use in developing the new databases and software within the next month.
Once it is up, Mark can work on establishing a draft database there and the GLC
can assist people in establishing connections.
RAPIDS Concepts
Mark Young led the discussion
on the definitions of the new data model. In the new
model, sources are now related to geographies and inventories. Multiple
inventories could be maintained within the same database
instance , but each source record would only be associated
with one inventory.
Mark provided a definition for Inventory:
"An inventory is a collection
of sources and all data associated with those sources. An inventory includes
sources, emission units, release points, controls, processes, activity data and
emissions data. An inventory may include multiple scenarios for activity and
emissions data."
Orlando asked about
the need to have each source associated with a single inventory, rather
than with multiple inventories. Mark responded that this is
advisable because source characteristics can change from one inventory to the
next and although these characteristics could be given time-stamps or something
similar, it is much simpler to just replicate the source records rather than
these other schemes. Mark also noted that some options for making new
inventories might be developed, such as duplicating an inventory and allowing a
user to alter sources.
Chun Yi asked about where the
scenarios and how this relates to inventories. Mark responded that the scenario
is a child record of the
inventory.
Mark provided a definition for Scenarios:
"A scenario is a collection
of related activity and emissions data within an inventory. Multiple scenarios
may exist within an inventory. For example, an Annual Average inventory and an
Ozone Season inventory can coexist as different scenarios within a master
inventory. Scenarios share a common set of sources, but may be used to produce
multiple subsets of emissions data."
Orlando noted that
he's currently working on an area source inventory for LADCO in which he
wants average daily emissions based on the annual data. He's
been able to do this with the current RAPIDS database, but it required some
customized queries and calculations. This would be able to be done more simply
in the newer model with scenarios.
Pete pointed out that he has some
suggestions about how to handle the geography. This may replace some of what
Mark has under the state, nation and county tables.
Mark explained that most of the
tables have codes that relate the table to each other. These will all be 16 characters.
The RID field will each have 20 characters. The comment fields can be made much
longer in terms of the limit and made variable length. Postgres claims that
there is no performance impairment for variable length fields as there were
in Oracle. Chun Yi asked about the
potential for converting these to Oracle and whether any modifications would be
needed. Mark responded that the reproduction in Oracle will take some effort
regardless of what the field characteristics are.
Chun Yi expressed that she would
like for Oracle to be supported. This isn't an option for the project due to
the budget needed to support Oracle and the additional programming time to
support two databases. Jon will correspond with Dwain and Mark Olsen in Minnesota to try to resolve
these concerns.
Mark provided a definition for Periods:
"A period is a pre-defined
date/time range. Periods are associated with activity and emissions data. The
use of periods is intended to simplify record comparison operations and to help
ensure that all related records carry precise dates and times."
Mark is still uncertain that
periods as a concept will work as a replacement to the start date and end data
method. Mark noted that he sees a possibility of having seasonal activity data
in the current system. Mark needs to work with the group to figure out what
information is needed to calculate emissions for activity data that changes.
Jon wondered what issues Mark is
now seeing that may be problems with the period code concept. Mark said that he
thinks this may cause some issues on data import, where it may be problematic
for the system to what periods exist and what the codes are for these. Jon
suggested that the codes could perhaps just be a concatenation on start and end
date. Mark suggested we could keep both for the time being and see if one is
more useful than the other.
Chun Yi noted that she would like
to see the capability for hourly emissions. These can be helpful in doing
dispersion modeling and risk assessment. Orlando
suggested the possibility of having the
capability of estimating emission using emission rates in mass per unit time
(LB/HR). This will facilitate the use of source specific emission's test data
in RAPIDS.
Mark provided a definition for Attributes:
"Attributes provide a way of
extending core entities in the data model. Currently, RAPIDS 2.x is a flexible
attribute data model, where the use of metric codes and the activities table
provide ways of storing data elements not contained in the core tables. The
RAPIDS 3.x design supports flexible attributes, in a similar manner, via the
use of an attributes table. However, the attributes table is intended for data
elements not needed in emissions calculations."
Mark also provided a definition
for Activity Data:
"Activity data include any
data elements that can potentially be used in emissions calculations (aside
from emission factors and SCC-related constants, which are stored in the
emission factors table). Activity records can be assigned to various data
levels, from the geographic levels down to the process level, as necessary.
Activity records are also designed to be referenced and used by calculation
protocols.
An inventory is a collection of
sources and all data associated with those sources. An inventory includes sources,
emission units, release points, controls, processes, activity data and
emissions data. An inventory may include multiple scenarios for activity and
emissions data."
Mark is planning to move as much
information as possible out of the current activity table and into other
tables. The concept of activity will be retained so that future concepts can be
included that aren't currently envisioned. The attributes table will include
information that isn't used for calculations. This could be characteristics of
various entities within the model. In most cases the new Activities table will
contain things that are actually activities. Activity data will be things that
are or could be used to conduct calculations. Orlando noted that the
current system only allows Stream Activity records to be used in calculations.
The use of other activity parameters such as process and
geographic activities will be an improvement.
Multi-level data
In the current system there can be
multiple levels of emissions data. For example, the current activities
table has records that can correspond to multiple levels.
There will be no streams in the
current model, so the process is the lowest level of definition.
Orlando pointed out
that its currently not possible to have multiple constants for things like heat
value of coal or something similar. Mark suggested these would best be placed
in the emission factors table. Orlando
gave an example of using a more specific type of wood for a throughput that has
a more specific heat content.
Mark suggested the possibility of
having a region or group level where sources or process could belong to
entities other than those factors have been pre-defined (e.g., geogrphies,
NAICS, SCC).
Orlando noted that
there are some calculations that use different factors for urban and rural
areas. The current RAPIDS doesn't account for this in the emission
calculations. The new system may be able to. The potential of having
multi-level data being able to refer to a group might be a good way of
addressing this.
Mark explained that the system can
search at multiple levels for the most specific data. State factors would
override generic and so forth on down to process.
Tom and Buzz noted that the NEI
model doesn't allow for a percentage fugitive to be modeled. While in theory
two stacks could be defined to represent this, the NEI won't allow that. This
may be possible in the new RAPIDS.
The core tables in the model are
currently the same as what was presented by phone in December. The reference
tables will draw heavily on RAPIDS 2.X, but will likely eliminate unnecessary
information and may make some other changes.
Mark explained some of the steps
in getting the database set up and configured. He'll need to work with the GLC
to ensure the proper paths are used.
In some cases, database views may
be created that allow queries of data to be done more quickly without having to
join tables. RAPIDS 2.X has used these to some extent, but most of the end
users don't seen to have been using these.
Regarding controls, Chun Yi asked
about the use of controlled emission factors. This should be handled the same
way in the new system.
Data import and export are another
module that can be included. If it is done correctly, it should only be
required to define the mappings and the system can automate the conversion with
no additional programming required. It will be an ongoing process to define
these imports and exports.
Mark discussed the possibility of
denormalizing some of the data so that some information components are
replicated at multiple places. For example, SCC codes could be in the process
table and in the emission table. Another way to cover some of the same needs
might be to use views that provide this function. Mark will need to do some
testing to see what works best.
A system catalogue and a data
dictionary are additional components that the system should contain. The
catalogue would contain key portions of code that define commonly-used system
components, such as formatting of how tables appear on the screen.
Mark walked through several
examples of how actual data would be represented, including examples for point,
area and mobile data types.
Within the discussion on point
source data representation, there was a lengthy discussion on the
representation of capture, control and fugitive emissions.
During the discussion on area
source, Orlando
raised a concern over the current RAPIDS, which won't allow the emissions
calculator to operate on a single process; a whole source must be estimated.
Mark expressed that he would like the new RAPIDS to be able to calculate at the
process level and thinks this can be done.
Confidentiality
Mark asked about what is needed in
the data model to support confidentiality requirements. In RAPIDS 2.X, any
activity record can be flagged as confidential. In the new model, the activity
records may need to be confidential but not the emissions. There may also need
to be a confidential flag at the source level. Sources that are confidential
generally only allow reporting of the facility ID information and the
emissions. When records are imported, this may require the system to create
some dummy sources.
We will need to have confidentiality
flags at the source and activity level. It may make sense for the data model to
have a confidentiality field at each stage in the model, even if these are not
used in practice. The confidentiality will have to operate at the export level,
so that things are censored when they leave the system. Emission factors may
also be confidential.
Note: The current model has a
metric of "Confidential" at the source level. All
activity metrics, regardless of level type, can be flagged as
"confidential" records.
Calculation
approaches
Within the activities table, a
material conversion can be done. This converts, for example, coal to CO (same
as emission calculation). In addition, there are unit conversions, material
values and scalar values.
Chun Yi asked about cases where
two conversions may be needed. Mark explained that in the current system, there
are standard units. For example, pounds is the standard unit for mass. No mass
unit conversion will require more than two conversions: one to pounds and one from
pounds to the desired unit. The SQL statement can be set up so that this is
done in one step.
Within any field that might be
used for calculations, there will be a set of standard numeric fields to
handle how the units work. These include a value amount, an exponent, a
calculation value (double precision), and various unit numerators and
denominator codes. Mark would like at some point to have an equation editor
that allows some significant amount of flexibility in equations that can be
used. Some examples were given of complicated equations, such as landfill
emissions which are based on a logarithmic decay over time, and residential
wood combustion, which can have complicated methods for apportioning by
population.
The group talked about the timeframe
for implementation. A first step once the database model is created is to
create scripts to build out the database. The software mark has will generate
SQL to build the tables themselves. The code for establishing the database and
setting users and privileges will need to be established. Following that, the
imports can be programmed.
2005 Inventory
Timeframe
The group discussed the timeframe
within each state for completing the 2005 inventory.
IL: Criteria is finished. Adding
the toxics should not be difficult.
MI: Michigan should have their HAPs for point
sources by the EPA's June 1 deadline. Dennis has done criteria pollutants for
area sources. Adding the toxics should not be difficult. He may also be able to
add some categories. Michigan
has formed a Manganese workgroup. Detroit's
monitoring programs have detected elevated levels of manganese in air and soil.
Dennis has gotten some of the regional data to use in providing some
information to this efforts.
MN: Minnesota has done some restructuring of
their EI strategy. Their goal is to have everything completed by June 1. They
have completed the inventory for many of the top facilities and have collected
information from several hundred other facilities. They have also collected
information on VOC-using facilities. Chun Yi is currently working on converting
the criteria pollutant information to RAPIDS to do HAPS. The airport
inventory has been completed. They are still waiting on locomotives. For
commercial marine vessels, Minnesota
is attempting to join the LADCO contract for this category. They are
considering using the 2002 onroad and nonroad data. Minnesota has compiled the latest FIRE
database of emission factors and added them to their RAPIDS. She will share
this with the group. Chun Yi also has compiled a table of state-specific
factors that she can also share. These were derived by filling gaps in FIRE.
NY: The point source data is about
75% done; they are currently in the stage of "technically completing"
these data points. Their area source categories match up with the GLC list very
well, so they should have a very complete area source inventory.
OH: OEPA is working on completing
the SIP inventory in conjunction with LADCO. After that, Tom will start working
on HAPs and hopes to have both these done by June 1.
ON: Point sources are
directly-reported and these are already received. Area sources will be similar
to 2002. They will need to do the mobile estimates. As usual, they will need a
10-day buffer to report the data.
PA: Karen is currently working on
the SIP inventory. She's hoping to have the CAPs to EPA by June and she will
then work on the toxics. Their 2002 inventory submission was much more complete
in terms of the number of pollutants. This was due to just giving the data in
its current format rather than converting to RAPIDS. They have developed a new
emission data entry software under a contract with CIBER.
WI: Most area sources are
completed. Point sources are done. They will begin working on the Non-road.
This time, Wisconsin
has done outdoor wood boilers and portable gas containers. He has added SCCs to
RAPIDS for these. These SCCs can be added to the repository. The OWB
methodology was done based on some use statistics from New York and some emission factors. Wisconsin's inventory
for these is currently just for CAPs.
By June, most states should have
the initial stages of the inventory done. That will be a good time to do some
comparisons of emission factors and other information for consistency. In
addition to the usual QA/QC checks, some additional checks can be done to
identify the coverage of pollutants for various source types and the SCCs used.
A portion of the next meeting can be devoted to this. Jon will begin working on
a framework for how to do this inter-comparison.
Area Sources
The group discussed the Area
Source Methodologies that have been compiled in the past and whether it was
worth doing an update of these at the current time. These were last reviewed
and updated in 2004 in preparation for the 2002
inventory release. Jon distributed a list of the Area Source categories for
which methodologies have been written. There area a few additional categories
people in the group have reported on in the past and Jon added a few of these
to the list.
The EPA has recently announced
that they will be organizing a series of discussion groups to update the NEI
guidance for several area source categories. They haven't provided details yet
regarding when these groups will begin and what their timeframe is for
completion. It would be good to have involvement from our group to each of
these to provide input and receive information on potential
changes.
Chun Yi will participate in the
solvent use group. Chun Yi noted that Minnesota
has also been doing swimming pools, prescribed burning and forest fires. These
categories could have methodologies added. Chun Yi also noted that they may be
doing some work on fuel combustion mercury emission factors. She can share this
with the group when it is completed.
Orlando is currently
working on an update of their pesticide methodology. The EPA's assumption of
90% evaporation is likely to be erroneous. Wisconsin is working with some university
folks to put together information based on actual volatility.
Tom suggested concentrating on
methodologies where the group currently does not have one. EPA has a commercial
cooking methodology, but the regional project has never officially adopted
this.
Orlando pointed out
that in addition to updating methodologies, there is also a need to standardize
SCCs among states. Although SCCs to use for reporting have been listed in the
methodologies for the group, there is still wide variation in how these are
reported among states. Orlando
will also look at residential wood combustion.
Several states have found traffic
marking emissions to be very low and some are considering dropping this
category. In Wisconsin,
the only compound emitted was methanol.
Several of New York's area sources will be less
accurate this time. Surveys were done previously for some categories and won't
be done this time. Generic information will be used instead.
Dennis will look at Lamp Breakage.
They have had a staff member in the past working in this area. She may have
developed some new information that Dennis could pass on to others.
The methodologies that are
developed should be able to be handled by RAPIDS. The equations that are used
should be summarized and provided to Mark so that he can evaluate whether
RAPIDS may be used for some of the estimations. Tom stated that he has found
that nearly every methodology can be summarized into AxBxC=D. In some cases,
the activity would need to be calculated prior to applying this calculation and
that might need to be done outside of RAPIDS to provide the activity.
In the next few months, the group
should do an assessment for each methodology to evaluate which pollutants and
emission factors each state or province is reporting or using. This can be done
with the interim results of the 2005 inventory.
Tom commented that the protocols
are rather lengthy and can be shortened. He volunteered to come up with a list
of the basic elements that should be in the protocol.
Benzo(a)pyrene
report
Steve Rosenthal and Erin Newman of
EPA joined the group for the discussion on the benzo(a)pyrene report. Jon sent
the latest draft to the group prior to the meeting. The only section of the
report Jon sees doing significant revisions to is the portion of the conclusion
dealing with long-term trends. He would like to get the group's input regarding
any information that would help make a general assessment of changes in the
emissions from the main source categories over the past 15 years. Steve agreed
that this type of approach would be very useful.
Orlando noted that the
emission factor for certified woodstoves is higher than for conventional. This
is true in both the MARAMA report and in the NEI guidance. Steve was surprised
to hear this and has information from other studies that shows the toxic
emissions from these units are far lower. These two sets of information should
be compared.
Steve noted that the report could
mention outdoor wood boilers as a potentially growing source and that this
might lead to an increase in the emissions from wood burning.
EPA has an annual study of the
amount of waste burnt in barrels. Erin Newman can provide some information on
this. This might be useful for checking with the estimation methods for open
burning of residential waste.
Steve mentioned Creosote treated
wood and asphalt sealcoat as possible contributors. These were raised as part
of the NY-NJ Harbor study that was recently
completed. In addition, he suggested the potential of including emissions for
structural fires, tire fires, agricultural burning.
Erin
mentioned that OAQPS has a group working on uncontrolled combustion. Although
burn barrels have been well characterized, there are other uncontrolled
burning. An example is the burning of agricultural
plastics. It has not been determine if the burning of
agricultural plastics is a significant source of HAP emissions. The uses
of plastics on farms for various purposes have been increasing. She can provide
the URL and password to anyone who is interested. Erin's
burn barrel group has a website that has information from some of these
sources.
EPA's Dioxin
Study
The EPA released in November the
report for their recently completed dioxin emission study, which presents
emissions for 1987, 1995 and 2000. The 600+ page report has a wealth of
information in it that the group may find useful, including many emission
factors from a wide variety of emission sources. The EPA has said that an
actual database with the inventory data will be released in the coming months.
When it is, it may be useful to do a comparison of their data with ours to see
how drastic the differences are.
There are numerous things in the
report that might be used to improve the quality of the dioxin and furan
information in our own inventory. At the least. A compilation of emission
factors from the report could be done and those that are new could be added to
our database for future use. In many cases, the report includes emission
information at the congener level for the set of congers believed to contribute
substantially to toxicity. Our inventory is currently only set up to handle the
2,3,7,8-dioxin and furan congeners individually and has a catch-all material
code for PCDD and PCDF. We've often gotten input from toxicologists in the past
that congener-level data is most useful to them. There appears to be sufficient
information available to provide this information if material codes are added
and the group protocol is modified to allow this. PCBs are a similar set of
group and some of these (the co-planar or "dioxin like" PCBs) are
discussed in the EPA's dioxin report. The group recently added material codes
for PCBs at the chlorination level. At that time, they received some input from
toxicological staff that also suggested that congener level, rather than chlorination
level, is the most preferred way of classifying these.
However, estimating emissions for these materials also depend on the
availability of emission factors.
Erin was not
aware of any firm plans by the EPA to update the inventory they have compiled
for years beyond 2000, so the availability of more recent information will
become desirable. If the regional inventory had data of the quality contained
in the EPA report, she could see many uses for it. She agreed that information
at the congener level is highly preferred.
Erin noted
that if the group was looking to take on some additional PBT-related projects,
there is a need for information of any type regarding emissions of
"chemicals of emerging concern." Information is slowly becoming available
that might allow inventorying at least some sources of these compounds,
although no-one has yet taken on the task of incorporating them into an
inventory. Given the group's goal of providing information to support programs
dealing with deposition to the Great Lakes,
this might be a suitable thing to work on. Jon agreed and suggested a first
step might be to do a review of the information that is out there on sources of
these contaminants and identify which, if any, are able to be inventoried at
some level. This was also a topic that Ed Piche raised during the group's
recent meeting in Toronto.
The group set the next conference
call time for Thursday the 22nd of February at 2pm EDT. We will discuss the
next in-person meeting at that time.