The New Emissions Modeling Framework System Emissions Modeling

The New Emissions Modeling Framework System Emissions Modeling Team Marc Houyoux, Madeleine Strum, Rich Mason, Norm Possiel David Misenheimer, Darryl Weatherhead, Larry Sorrels Bill Benjey, George Pouliot EMF Contractor: UNC Institute for the Environment Alison Eyth, Qun He, Alexis Zubrow, Darin Del Vecchio, Dongmei Yang 1

Overview n Background n Development status and benefits n How EPA is using the EMF n EMF Future n EMF Demo 2

EMF Motivation n Improve timeliness, quality, and transparency of emissions data used in air quality models Air Quality Modeling Emissions Inventory Emissions Modeling Data handling SMOKE processing QA checks Standard procedures Documentation Reproducibility Summaries and Analysis 3

EMF Goals n Make emissions modeling faster and more accessible n Create and implement emissions modeling protocol n Provide data/processing transparency & tracking n n n Data with versions, metadata, QA status Emissions modeling applications and associated data Data queries and summaries n Meet needs for multi-pollutant modeling n Create tools that can be used by EPA and others (e. g. , RPOs, states) 4

EMF capabilities n Data Manager n n n Case Manager n n n Organizes/stores data Provides tracking of data changes using data versions Supports metadata “QA Manager” facilitates QA capabilities and tracking of QA steps, summaries, and queries Organizes model steps, inputs, & parameters (e. g. , SMOKE) Runs Emissions Modeling steps at NCC, records history and registers outputs Control Strategy Tool (Co. ST) n n n Stores database of control “measures” (practices and technologies) Supports user to create control strategies (multi-pollutant functionality) Applies controls to emission inventories for use in SMOKE n All done in a multi-user environment n SMOKE and other tools are run from EMF, but are not within EMF 5

EMF Development Status and benefits 6

Timeline Oct ‘ 04 Project start Apr ’ 05 Co. ST team decides to use EMF 2005 Jan ‘ 06 Surrogate Tool & EMF Betarelease installed at EPA Mar ’ 06 Initial Speciation Tool 2006 Sep ’ 05 Final Design Document Sep ’ 07 Updated Mar ’ 07 Speciation Initial Tool EMF server move to NCC 2007 May ‘ 06 First production data loaded Oct ‘ 06 First Case “Inputs” Setup (2002 v 2) 2008 May ’ 07 EMF Training Design Oct ‘ 07 2020 cc EMF Demo nears completion Production Development Data Manager Emis. View Oct ’ 07 Case Mgt Complete Runs EMF Speciation Database Tool at NCC (2 TB disk) Maintenance Case Manager QA Manager 7

Major milestones met so far n EMF-funded components all exist in working form n Data Manager, including QA Manager n Case Manager n Control Strategy Tool (Co. ST) n Spatial Surrogate Tool (Create challenging SMOKE inputs) n Speciation Tool n Server obtained, installed, and running at NCC n OAQPS using EMF for supporting ongoing projects n EMF training module created and training held at 2007 EPA Emission Inventory conference 8

How Data Manager helps EPA n Stores data in a Postgre. SQL database accessible by all. All team n n n members can find the latest data. Others’ data changes shared. Access by staff other than emissions modelers. Supports multiple versions of datasets. We can access both old and current versions of data, for reproducibility and transparency. Users can edit data in a user interface to the database. Convenient. Automatically records metadata, such as who/when makes changes. Allows users to add what/why of changes and manual metadata entry. Eases our burden for keeping track of details other ways. Adds transparency after work is done. Exports SMOKE-ready files, including documentation in file headers. Saves us time & prevents possible format errors. We can better focus on analysis and modeling needs. 9

How QA Manager helps EPA n Stores QA steps, who/when steps performed, & QA status along with the data. We can easily find QA status of data. We can see the overall QA status, what QA steps complete, who performed steps, and the results. Users can add steps as needed. n Tracks QA steps by dataset version. We know what steps were performed on each version. Helps prevent skipping QA. n Supports running SQL queries on the data to perform QA steps and summaries. Give us the flexibility and convenience of database queries. Runs within EMF and results stored for present and future use. n Templates of QA steps (queries) can be created and reused. Helps ensure QA of data gets done, decreases learning curve for new emissions modelers. n Tracks both automated (run by EMF) and external and/or manual QA steps. Gives us needed flexibility to get work done. 10

How Case Manager helps EPA n Data in a “Case” (e. g. , SMOKE run) linked to datasets in Data Manager. We have a single system to use for data handling and running SMOKE. n Exports all SMOKE-ready files needed for a SMOKE run with a few mouse clicks. Saves time and prevents possible errors associated with manually writing out many data files & formats from many other data sources. n User interface for SMOKE. Reduces learning curve for new emissions modelers. We have a permanent record of SMOKE inputs, settings, and location of outputs. n Submits SMOKE jobs, preprocessing, and postprocessing to a compute server (“amber”) or EMF server. Supports desktop access to data and SMOKE cases, while running at NCC compute cluster to meet computational needs. 11

How EPA is using the EMF 12

Uses to date n Data Manager and Case “Inputs” functions used for: n n n 2002 -based CAP platform v 3 (2002, 2009, 2014, 2020, 2030) 2002 -based CAP/HAP platform v 3 (2002) Ozone NAAQS Final Rule OTAQ’s Locomotive Marine Rule ORD 2005 -based platform Detroit Multi-Pollutant Study n Actually run using EMF Case Manager: n n n SMOKE 2020 base case to demonstrate we can replicate the “old” way of running SMOKE Speciation Tool runs for ORD to create SOA-enabled speciation profiles for SAPRC 99 and CB 05 A few sectors on the 2002”ae” (v 4 platform), in development 13

EMF in FY 2008 n In-house efforts to setup EMF for operational use n EMF work assignment n Support and maintenance n Add “quick and easy” emissions sensitivity tool for AQMG n Co. ST n Completion of O 3 NAAQS RIA Mobile Control Strategies n NOx/SOx NAAQS RIA Control Strategies n Manage SMOKE inputs: n OTAQ’s Bond Rule and SECA / IMO modeling n Detroit Multi-Pollutant Study n Upcoming runs of SMOKE with EMF: n 2002 ae (v 4) platform n New base runs for 2003, 2004, 2005, & 2006 for CDC/Phase, CMAQ evaluation, and Accountability 14

EMF future n Ongoing work to refine Data Manager, QA Manager with a more robust protocol, Case Manager, Co. ST n As part of OAQPS’s daily operations, EMF is assisting us manage emissions modeling for 8+ projects simultaneously with few staff and high quality n Possibility for adapting to meet data management and modeling needs of others in AQAD and EPA n n EMF could be “Environmental Modeling Framework” Readily adaptable to support other applications such as CMAQ 15

EMF demo 16