10

The University of Washington runs the Advanced Research Weather Research & Forecasting (WRF) model in predictive mode, and then tests the skill of the model anywhere from 12 - 84 hours out from the initialization time. The accuracy of the wind and the temperature profile are considered to be the most important indicators of the model's skill. The CMAQ Eulerian grid model is then be used to represent air pollutant dispersion, using the WRF data and grids.

I've been told that the models don't do well with grid-spacings any finer than 1/3 km. In fact, some people have said that the 4/3 km spacing can perform just as well, and that we've reached a point where Eulerian grid models don't gain much more by going to finer grids.

I know that for modeling pollutant dispersion at fine horizontal scales, Lagrangian methods have historically been used instead of Eulerian grid-cell methodology. However, with recent developments in computing, gridded modeling domains have been able to use much finer grid-spacings. What is the finest grid spacing that Eulerian dispersion grid-cell models should realistically use? Why?

f.thorpe
  • 13,583
  • 2
  • 46
  • 89
  • This is not a criticism. I find a scale of 1/3 km intriguing because 1/3 doesn't produce a neat number in digital computing. A scale of 0.25 km or 0.5 km would result in neat numbers for calculations. Why would a scale of 1/3 km (0.333 km) be used in such modelling? – Fred Dec 07 '15 at 01:40
  • @farrenthorpe - the point Fred is making is not about the grid resolution. It can be as high as is currently possible. He I believe is asking why is it not a neat number such as 0.25 or 0.5 km ? –  Dec 07 '15 at 02:06
  • 1
    @Fred sorry I missed your point. In our case, the reason 1/3 is used is (I believe) is due to grid nesting and maximum computational capability for forecast due date. That is, the nested domains break up into 9 pieces rather than 16, which is a big computational difference, but the gain is nearly the same. And they just go to two decimal places... 1.33 – f.thorpe Dec 07 '15 at 02:54
  • 1
    @farrenthorpe - Presuming you are operating in a LES domain - the answer to your question is in this reference - http://journals.ametsoc.org/doi/pdf/10.1175/JAS3435.1. Primarily EDM is an eddy diffusivity approach and does not simulate properly short range dispersion –  Dec 07 '15 at 03:17
  • @gansub FYI grid resolution is really not an appropriate term for modeling as it is more specific to remote sensing. A model has grid-spacing which varies by nest/domain. – f.thorpe Dec 07 '15 at 15:35
  • @farrenthorpe - sorry my bad. –  Dec 07 '15 at 16:21
  • Please forgive me but I am having some trouble understanding your question. Are you talking about truncation errors associated with discretization in the Eulerian framework which give rise to numerical diffusion and dispersion? – Isopycnal Oscillation Dec 07 '15 at 20:32
  • @IsopycnalOscillation - Not really. The question is whether the EDM framework can capture some short range(spatial and temporal) events for that fine grid spacings ? –  Dec 08 '15 at 01:09
  • At what point, the resolution is not only a question of computer: do you have the terrain data at this precision ? (not only z, but also the surfaces properties, including plants and soil, for instance). Do you also resolve cloud precisely enough to account for accurate lighting of that terrain, and thus predict evaporation accurately ? On lakes and oceans, how fine is your coupling ? – Fabrice NEYRET Feb 21 '16 at 01:21
  • Ok course the third dimension is a different story: only the bottom is involved in precisions issues concerning the boundary real condition. Above, you do need as much resolution as you can to treat correctly the thin inversion layer, for a start. – Fabrice NEYRET Feb 21 '16 at 01:23

0 Answers0