CMAQ/MCIP/WRF/emission grid definition

Hi all,

I am using WRFv3.7.1 to create met files for CMAQv5.2.1. My domain is over India, so the emissions input file are not generated by SMOKE. I know that WRF setups the grid and domain for CMAQ, but I am unclear on some of the specifics.

I have successfully run some WRF and CMAQ with the grid specifications below. My primary question is this: if I add a larger outer domain in order to nest my original smaller domain, and the coordinates of the lower-left corner change but the center stays the same, does the emission grid need to change somehow? That is could I update WRF in this manner and run CMAQ with the same, unchanged emission files?


  • e_we, e_sn = 160
  • dx, dy = 4000
  • map_proj = lambert
  • ref_lat = 28.0338
  • ref_lon = 77.2786
  • truelat1 = 26.0
  • truelat2 = 32.5
  • stand_lon = 77
  • lower-left corner lat = 25.15879
  • lower-left corner lon = 74.13348


  • NCOLS, NROWS = 147
  • XCENT = 77
  • YCENT = 28.0338
  • XORIG = -266500
  • YORIG = -294000
  • BTRIM = 5
  • WRF_LC_REF_LAT = 28.0338

I’m not sure if I’ve described my question clear enough. Please let me know if I should provide more information.


It sounds like you are looking to rerun WRF with a coarse (e.g., 12 km) outer domain, process it with MCIP, use that to drive CMAQ, and then use those time-varying coarse fields as inputs to BCON to derive boundary condition files that you will use as input in a rerun of your 4 km CMAQ simulation. For your coarse CMAQ simulation, you want to aggregate your existing fine-scale emissions onto the coarse grid.

I have not reverse-engineered a coarse grid myself, but it is possible. You of course have to ensure that your grids are aligned. To do this, you can use ref_x and ref_y in conjunction with ref_lat and ref_lon, as described in the WRF WPS user guide.
Alternatively, you can create two nested WRF domains at once, where the inner domain is specified using i_parent_start and j_parent_start. Keep playing around with rerunning geogrid and tweaking your WPS namelist until your resulting matches the original file from your standalone 4km domain.

This may just be repeating what you already know, in which case it isn’t much help (sorry). Maybe someone else can provide further specifics, or possibly on the WRF user’s forum.

When you do get this worked out, it would be a great use case for us to have documented as one of our tutorials.

The benchmark CMAQ case seems to have the same grid + coordinate system (like the NCOLS, NROWS, XORIG, YORIG) across MCIP and SMOKE output for use in CMAQ. However, I wanted to know about the capability/limitation of the grid that we can use in CMAQ. How flexible is CMAQ in accomodating MCIP and SMOKE outputs in different grids (different row/column number, x/y-origin, etc). If I had WRF output files from collaborator 1 in grid 1, and SMOKE output files from collaborator 2 in grid 2, what is the easiest or cleanest way to proceed to make CMAQ to accept both WRF and SMOKE outputs in different grids?

CMAQ requires that all its inputs have the same coordinate system and grid.

If the coordinate systems are the same and the grids are of the same resolution and aligned (so that the grid nodes coincide – are (XORIG1-XORIG2)/XCELL and (YORIG1-YORIG2)/YCELL integers?), then you can use M3Tools programs m3wndw and bcwndw to window to a common domain within the intersection of the grid-coverages.

If not, then you can interpolate or re-aggregate using M3Tools programs m3cple or mtxcple to generate outputs on a common CMAQ grid. Be aware, however: (1) you “fuzz” the features in the output files to at best an effective (XCELL1+XCELL2) resolution (the Nyquist Theorem), and (2) be careful how you re-scale emisions data, since they are not in interpolatable MKS units (the actual units have an unstated “per grid cell” in the units, and you need to adjust the output by the ratio of the grid-cell areas.