4 The record articles

Appendix W Revision: The New Era of Air Quality Modeling – (Re)Introducing MMIF, an Alternative Option for Met Data

Posted: February 9th, 2017

Authors: All4 Staff 

As a self-described computational meteorologist, I get pretty obsessed with weather models, especially models of severe storms like hurricanes or big winter storms.  If the coverage on the local news or weather networks is any indication, I don’t think I’m alone in this.  My love for weather modeling makes me particularly excited for the Appendix W revision, and there is a specific Appendix W revision that should also pique your interest.  Read on and I will tell you why!

In order to do any kind of air quality modeling, it is necessary to have knowledge of the state of the atmosphere that pollutants are being released into.  The previous version of 40 CFR Part 51, Appendix W offered two options for meteorological input: one year of on-site monitored meteorology or five years of data from an off-site weather station with meteorological conditions representative of the project site.  Either of these two options could be a barrier to our clients due to the cost and time associated with developing on-site meteorological data, or the sacrifice that occurs when “representative” meteorological data are used to simulate the dispersion conditions at a client’s facility.  With the December 2016 revisions to Appendix W, a new option is available that allows for the use of representative prognostic model data.  The data can be processed through the mesoscale model interface program (MMIF) as input to AERMET in situations where on-site monitoring is prohibitive or infeasible, and where representative monitored data are not available or the best choice.

The basics: what’s a prognostic model?

A prognostic model is a model that can also be used for weather forecasts.  Specifically, it uses equations of fluid motion to evolve the state of the atmosphere, represented at its most basic by a three-dimensional grid of winds, temperature and pressure. In the case of a weather forecast, the state of the atmosphere is evolved to a future time.  The results of the prognostic model can also be used to give hourly meteorological information at model grid points where meteorological measurements may not have been made.  This means that if we can use the hourly results from a prognostic model as input to AERMET, then it’s possible to have meteorological data at an area representative of the project site (or right at the project site, really!).

An example of a prognostic model is the Weather Research and Forecasting (WRF) model, not to be confused with Star Trek’s Worf.  WRF is widely-used, and it’s the prognostic model recommended and used by the United States Environmental Protection Agency (U.S. EPA) and the National Weather Service (NWS). WRF can be run and specifically tailored to a project site, or a previously developed WRF data set that is appropriate to a project site may be used as input.

Let’s get into the details: how do we incorporate prognostic data?

With representative prognostic model data from the most recent three years, it’s time to set up MMIF to process the data into a form appropriate to input to AERMET.  For regulatory applications, Appendix W states that MMIF should be used to generate input for AERMET rather than straight to AERMOD, although direct to AERMOD processing is an available option in MMIF.

MMIF will generate a surface file and an upper air file that are appropriate for input to AERMET and are the analog of a surface file from a NWS station and an upper air file from a balloon sounding.  MMIF will also generate an output file with surface characteristics that is equivalent to the output from the AERSURFACE program, which is a step that is required when running AERMET for on-site or NWS meteorological data.

With the MMIF output files in hand, these files can be used as input to AERMET, and the air quality modeling approach proceeds as it would for air quality modeling applications using representative NWS data or on-site meteorological data.  The option of using prognostic model data for air quality modeling applications provides an additional choice that may be time saving and cost saving for some applications.

Between the three choices of input meteorological data for air quality modeling applications, using existing NWS data are the most cost-effective approach but only when the data are truly representative of your project site.  Using on-site meteorology is a solid answer in challenging situations where otherwise representative data are not available and where precision and certainty in modeling results is required.  However, setting up an onsite monitor and collecting data can cost between $50,000 and $250,000, and you have to wait for one year of data collection.  The use of MMIF with prognostic model data as meteorological input provides a middle ground for situations where representative NWS data are not the preferred choice and the time to collect one year of onsite data is not practical.

Wrapping up: do you need assistance?

If you are facing an air quality modeling project, on-site data are not available, and National Weather Service data may not be the best choice for your facility, you are a candidate for prognostic meteorological data.  ALL4 is here to help you evaluate any and all meteorological options for your air quality modeling project.  Our experienced meteorologists, air quality modelers, and meteorological modelers are ready to provide solutions that cover all the meteorological input data bases: off-site representativeness analysis and meteorological data processing, on-site meteorological monitoring, AND use of prognostic model data through the MMIF tool.  Should you have any questions, don’t hesitate to contact Dan Dix (ddix@all4inc.com, 610.933.5246 x118), or anyone else on ALL4’s air quality modeling team if you have any questions and want to know more.


    Sign up to receive 4 THE RECORD articles here. You'll get timely articles on current environmental, health, and safety regulatory topics as well as updates on webinars and training events.
    First Name: *
    Last Name: *
    Location: *
    Email: *