EPA 910-R-15-001 d                           Alaska
             United States         Region 10      Idaho
             Environmental Protection      1200 Sixth Avenue   Oregon
             Agency           Seattle WA 98101   Washington
             Office of Environmental Assessment           September 2015
cvEPA       Combined WRF/MMIF/
                 Overwater Modeling
               Approach for Offshore
                    Emission Sources
                             Technical Summary


            Combined WRF/MMIF/
 Modeling Approach for Offshore
                 Emission Sources
                  Technical Summary

        EPA Contract No. EP-W-09-028

  Work Assignment No. M12PG00033R

                            Prepared for:

            U.S. Environmental Protection Agency
                              Region 10
                        1200 Sixth Avenue
                        Seattle, WA 98101


                 U.S. Department of the Interior
           Bureau of Ocean Energy Management
                      45600 Woodland Road
                        Sterling, VA 20166

                            Prepared by:

                       Amec Foster Wheeler
              Environmental & Infrastructure, Inc.
               4021 Stirrup Creek Dr., Suite 100
                        Durham, NC 27703


                       RAMBOLL ENVIRON
               773 San Marin Drive, Suite 2115
                        Novato, CA, 94998

                          September 2015


Under Interagency Agreement (IA) number M12PGT00033R dated 9 August 20121 between the U.S.
Department of the Interior (DOI), Bureau of Safety and Environmental Enforcement (BSEE) on
behalf of the Bureau of Ocean Energy Management (BOEM) and the U.S. Environmental Protection
Agency (EPA), Region 10 (R10) established a collaboration in which the regulatory approved
AERMOD (AERMIC [AMS/EPA Regulatory Model Improvement Committee] Model) dispersion
program2 was evaluated for predicting near field concentration impacts (< 50 kilometers) from
emission sources located at overwater locations, e.g., outer continental shelf (OCS). The objectives
of the collaboration are to (1) replace the Offshore and Coastal Dispersion (OCD) dispersion model3
with AERMOD for sea surface based emission releases, and (2) assess the use of Weather
Research  and  Forecasting  (WRF) model4 predicted meteorology in lieu of overwater meteorological
measurements from towers and/or buoys with AERMOD. The work plan consisted of six tasks to
address the objectives of the collaboration study5. The results and findings are presented in a three
volume report. Volume 1 is a project report that discusses Tasks 1, 4, 5 and 6 in detail and provides
only summaries of Tasks 2 and 3.  Task 2 and Task 3 details are contained in Volume 2 and Volume
3, respectively. This report is a technical summary of the six tasks that comprised the collaboration

A.     Task 1. Evaluation  of Two Outer Continental Shelf Weather  Research and Forecasting
       Model Simulations  for the Arctic

       Task 1  involved the review two existing WRF datasets for the Arctic Ocean that could be
       used to provide the necessary meteorological variables for dispersion model simulations of
       OCS sources within their domains. The objective of this task was to examine the differences
       between  the two datasets, analyze model performance with  overwater measurements, apply
       the Mesoscale Model Interface (MMIF)6 Program and AERMOD-COARE (Coupled Ocean-
       Atmosphere Response Experiment) (AERCOARE)7'8 to the datasets using different options
       to  develop meteorological input files for AERMOD,  and compare AERMOD model
       predictions from the resulting datasets using simulations of typical OCS sources.

       After the  first MMIF runs were completed and reviewed, peculiar planetary boundary layer
       (PEL) height and precipitation values were observed every three hours from one of the
       datasets. The problem was traced to the version  of WRF used to perform a reanalysis.
       9Additionally, shortcomings were noted with the other dataset, namely: 1) the domain does
       not extend far enough east to cover the Beaufort Sea where many potential lease blocks
1 Environmental Protection Agency. 2012.  Interagency Agreement No. M12PG00033R with the Bureau of Safety
and Environmental Enforcement on behalf of Bureau of Ocean Energy Management, Herndon, VA. Region 10
Seattle, WA. Augusts.
2 Environmental Protection Agency. 2004.  User's Guide for the AMS/EPA Regulatory Model-AERMOD. Publication
No. EPA-454/B-03-001. Office of Air Quality Planning and Standards, Research Triangle Park, NC.  September.
3 DiCristofaro, D.C. and S.R. Hanna.  1989. OCD: The Offshore and Coastal Dispersion Model, Version 4. Volume I:
User's Guide and Volume II: Appendices. Sigma Research Corporation, Westford, MA
4 National Center for Atmospheric Research. 2014. Weather Research & Forecasting (WRF) ARW Version 3
Modeling System User's Guide.  Mesoscale & Meteorology Division, Boulder, CO.
5 Environmental Protection Agency. 2012.  Order for Supplies or Services, Contract No. EP-W-09-028.
Headquarters Procurement Operations, Ariel Rios Building, 1200 Pennsylvania Ave, NW, Washington DC.
September 6.
6 Brashers, B. and C. Emery. 2014. The Mesoscale Model Interface Program (MMIF), Draft User's Manual.
Prepared for the Air Quality Assessment Division, U.S. Environmental Protection Agency. ENVIRON International
Corporation, Air Sciences Group, Novato, CA.
7 Fairall, C and E.F. Bradley. 2003. The TOGA-COARE Bulk Air-Sea  Flux Algorithm. NOAA/ERL/ETL, 325 Broadway,
Boulder, CO. September 2.
8 Richmond,  K. and R. Morris. 2012. Evaluation of the Combined AERCOARE/AERMOD Modeling Approach for
Offshore Shores. Prepared for U.S. Environmental Protection Agency, Region 10. EPA-910-R-12-007. ENVIRON
International Corporation, Novato, CA.

       exist and several have already been leased, and 2) the dataset covers only the open-water
       period and could not be used in dispersion modeling assessments of permanent sources.

       The peculiar results and shortcomings of the datasets made both datasets unsuitable for the
       needs of this collaboration study.

B.     Task 2. Evaluation of Weather Research and Forecasting Model Simulations for Five Tracer
       Gas Studies with AERMOD

       Task 2 included the comparison of WRF-driven AERMOD results to the concentrations
       measured during five offshore tracer gas studies (i.e., Cameron, LA; Carpinteria, CA; Pismo
       Beach, CA; Ventura, CA,  and Oresund, Denmark). Four of the five tracer gas studies,
       involving experiments conducted over North American waters (coastal waters of California
       and the Gulf of Mexico), were  used previously to evaluate COARE and AERCOARE. The
       fifth tracer gas study was  the Oresund Nordic Dispersion Experiment conducted during 1984
       in the strait that separates Denmark and Sweden and was also used to evaluate CALPUFF,
       Version 69. Volume 2 provides details of the methodology and analyses used in this task.
       The report also presented results of the AERMOD  performance evaluations, and analyzed
       how modeling options affect model performance.

       The results of this task suggested small differences in the key meteorological variables can
       cause large differences in predicted tracer gas concentrations for a given hour. Although
       many of the WRF simulations  perform reasonably well when compared to regional surface
       observation of winds and  temperatures, relatively small differences near the overwater point
       of release can have a large effect on stability which is a function of the "sign" of air-sea
       temperature difference. Spatial gradients of sea surface temperature (SST) near the coast
       and wind direction play key roles in the simulation of the stability and PEL heights over the

       The model performance analysis of the field experiments examined  in this study
       demonstrated WRF based AERMOD simulations can result in estimates of concentration  as
       good as or better than AERMOD simulations using observations - but not in all cases - and
       is dependent upon the reanalysis and PBL scheme.

C.     Task 3. Analysis of AERMOD  Performance using Weather Research and Forecasting Model
       Predicted and Measured  Meteorology in  the Arctic

       Task 3 analyzed alternative methods for supplying meteorological variables to AERMOD for
       regulatory air quality modeling of sources located overwater. It is hypothesized, given an
       appropriate overwater meteorological dataset, that AERMOD can be applied for New Source
       Review (NSR) demonstrations following the same modeling procedures as used for sources
       over land. That is, a combined modeling  approach  where the meteorological variables are
       provided by WRF and processed by MMIF and optionally AERCOARE. In this task,
       AERMOD is run using predicted meteorology and four observational datasets over the
       Chukchi and Beaufort Seas along the Arctic coasts of AK with the same hypothetical
       emission source scenarios for years 2010-2012 during the open water/ice free season. Year
       2009 was excluded because of the lack of temperature profiler data. The modeling
       methodologies and analyses conducted under this task are discussed in detail in Volume  3.

       The analyses suggest WRF was able to produce hourly meteorological datasets that
       compared favorably to over-water measurements.  Most AERMOD concentration results
       using WRF meteorology were adequate, falling within the fraction-factor-of-two lines and
       producing robust highest  concentration values that corresponded well with  observation-
9 Earth Tech. 2006. Development of the Next Generation Air Quality Models for Outer Continental Shelf (DCS)
Application. Final Report: Volume 1. Prepared for the U.S. Department of the Interior, Minerals Management
Service, Office of International Activities and Marine Minerals.  January.

       driven AERMOD results. Furthermore, the results suggest WRF extracted meteorology can
       be used as an alternative to offshore observations for air permitting in such areas. However,
       there was little discernable advantage in using AERCOARE to process meteorology
       extracted from WRF using MMIF to develop the input meteorology for AERMOD. The results
       also indicate that WRF extracted meteorology be filtered by limits on minimum wind speed,
       mixing height, and Monin-Obukhov length to avoid extreme conditions not typically observed
       over water.

D.     Task 4. Evaluation of Predicted and Measured Mixing Heights

       Task 4 provided a comparison of mixing heights derived from WRF to observations in the
       Arctic.  Available sources were surveyed for in-situ upper air observations in the Arctic
       marine environment. These sources included Shell's Kipp & Zonen passive microwave
       profiler installed at Endeavor Island in the Beaufort Sea, the radiosondes launched as  part of
       quality assurance procedures for the profiler, and radiosondes launched during Japanese
       Agency for Marine-Earth Science and Technology (JAMSTEC) research cruises10, currently
       only available for 2009 in the Chukchi Sea.  Predicted heights from the WRF solutions were
       compared to measured temperature profiles (i.e., profiler and radiosonde). The analyses
       conducted under this task are discussed further in Volume 1.

       Observed mixed layer heights  were analyzed from the Kipp & Zonen profiler11, JAMSTEC,
       and National Weather Service  soundings from Barrow, AK by two methods:  1) subjective
       hand analyses (i.e., a visual inspection of the plotted profiles by a trained meteorologist) and
       2) objective analyses using a Critical Bulk Richardson Number (CBRN)12 approach. The
       WRF mixing height output was analyzed twice using MMIF: 1) the 'pass-through' option and
       2) the CBRN calculation. Comparisons of the WRF and MMIF mixing heights to the hand
       analyses and CBRN approach, as well as mixing heights derived by AERMOD's
       meteorological preprocessor (AERMET)13 at Endeavor Island generally showed little
       correlation (r), with the best comparisons to the CBRN of r =  0.65 to 0.70.

       Profiles generated by WRF captured the essence and variation of the structure of the
       JAMSTEC soundings remarkably well.  In comparing the mixing heights estimated using the
       CBRN  method for both the WRF/MMIF and JAMSTEC soundings, 75 percent of the
       estimates fell within a fraction-factor-of-two with a correlation of about 0.6.

E.     Task 5. Evaluation of AERSCREEN for Arctic OCS Application

       The objective of Task 5 is to develop and evaluate a systematic approach for generating
       screening meteorology representative of overwater conditions that can be incorporated into
       AERSCREEN in lieu of MAKEMET14, the overland meteorological processor for
       AERSCREEN, or used directly by AERMOD in place of observed data. Three exploratory
       screening datasets were developed for input to AERCOARE to generate the input
       meteorology for AERMOD by:  1) reviewing and analyzing meteorological data  collected for
10 International Arctic Research Center (IARC) and Japan Agency for Marine-Earth Science Technology (JAMSTEC)
Annual Report. For the period April 1, 2009 to March 31, 2010. IARC, Fairbanks, AK
11 Hoefler Consulting Group. 2010. Quality Assurance Project Plan for the Endeavor Island (Endicott)
Meteorological Monitoring Program, Prudhoe Bay, AK.  Prepared for Shell Offshore, Inc., Anchorage, AK. April.
12 Gryning and Batchvarova. 2003. Marine atmospheric boundary-layer height estimated from NWP model
output.  International Journal of Environment and Pollution, 147-153.
13 Environmental Protection Agency.  2004.  User's Guide for the AERMOD Meteorological Preprocessor (AERMET).
Publication No. EPA-454/B-03-002. Office of Air Quality Planning and Standards, Research Triangle Park, NC.
14 Environmental Protection Agency.  2011.  AERSCREEN User's Guide. EPA-454/B-11-001.  Office of Air Quality
Planning and Standards, Research Triangle Park, NC. March.

       and developed in Task 3 to determine typical ranges of key meteorological parameters for
       the open water ice-free Arctic (includes buoy data, WRF data); 2) using the information from
       step 1 to develop an initial screening dataset from combinations of ranges of values of the
       minimum set of meteorological parameters required by AERCOARE (wind speed, wind
       direction, air-sea temperature difference, relative humidity, and mixing height) to calculate
       meteorology needed by AERMOD; 3) refine the initial dataset to produce two additional
       datasets. The derivation of the three exploratory datasets  are explained in more detail in
       Section 6 of Volume 1.

       Hypothetical emission sources, as described in Volume 1, were modeled utilizing the three
       exploratory screening  meteorological datasets. The same sources were used for refined
       modeling driven by 1)  meteorological data extracted from WRF using MMIF and input to
       AERCOARE or directly to AERMOD, 2) observations from buoy data input to AERCOARE,
       and 3) AERMOD-ready datasets for the Beaufort and Chukchi Seas provided by EPA Region
       10 to support permit applications for offshore exploratory drilling.

       The one-hour high-first-highest (H1H) concentrations from each of the three screening
       datasets were compared to concentration estimates from refined modeling. No screening
       dataset yielded consistently conservative results across all emission sources. The
       development of the screening data sets and analyses conducted under this task are
       discussed in detail in Volume 1.

       More detailed analyses of the screening and refined modeling results are needed such as
       extracting,  reviewing, and comparing the hourly meteorological data corresponding with the
       one-hour H1H concentrations that could provide insights into possible refinements to the
       screening datasets. An evaluation of scaling parameters for the 3-, 8-, and 24-hr averaging
       periods is needed to make this a viable option for screening modeling in the Arctic.

F.     Task 6. Collaboration Study Seminar

       EPA Region 4 in Atlanta, GA hosted the DOI BOEM/EPA  Region 10 Collaboration Study
       Seminar and Demonstration on 16 and  17 September 2014. Details  of the seminar are
       presented in Volume 1.

       The status and/or results of each  of the above five tasks, including hands-on demonstrations,
       were presented by EPA R10 and  its contractor team of AMEC (Prime) and ENVIRON
       (subcontractor). For the hands-on demonstrations, the attendees were provided personal
       computers that allowed them to follow along as the presenter ran through the exercises.
       After the seminar, each EPA and  BOEM/BSEE regional office were provided a flash drive
       that included all the PowerPoint presentations, exercises,  agenda and attendance sheet.

       There were twenty (20) attendees at the seminar and demonstration. The attendees included
       two from BSEE, seven from BOEM, and eleven from EPA regional offices. In addition, a
       conference call line was setup so that technical staff from  the EPA's Office of Air Quality
       Planning and Standards (OAQPS) could listen in and ask questions.

Volumes 2 and 3 will be used  to support R10's recommendation that WRF predicted meteorology
preprocessed by MMIF and input into AERMOD satisfies the regulatory requirements as an
alternative model pursuant to Appendix W in 40 CFR 51 (Appendix W). However, subsequent
studies and experiments are needed to address peer review comments that were not fully addressed
in the three volumes. Of particular relevance for offshore emission sources is the need for a
shoreline fumigation algorithm and a platform downwash algorithm in AERMOD to replace OCD, and
updated tracer gas experiments to evaluate model  performance. In addition, R10 suggested
changes to Appendix W that addresses the use of five (5) years WRF predicted meteorology and the
MMIF program with the AERMOD.