5

First all, my question was different from the question Converting grid resolution from degrees to kilometers.

In chemical transport modeling, the pollutant source inventory was an essential input which contain residential, industrial, agriculturial, vehicle emission in the area of interest.

The source inventory can be built independently or downloaded from the websites of several institutions which support some global/regional source information dataset.

For example:

  1. INTEX-B

    It contain different emission species like SO2, NOx, CO, NMVOC in the resolution of 0.5 degree x 0.5 degree

  2. Tracer-P

  3. REAS -> Regional Emission inventory in ASia
    Spatial resolution: 0.25 degree by 0.25 degree

  4. EDGAR The Emissions Database for Global Atmospheric Research
    spatial resolution: 0.1deg x0.1deg and 0.5deg x0.5deg

These source inventories are all gridded in geographic coordinate. But for regional/small scale research, the simulation grid was set in the unit of kilometers(For example, 36 km for regional simulation, 4 km for urban simulation).

Then, I want to re-arrange the source inventory into km x km cell. Due to different distance corresponding to the same degrees gap along the latitude, the degree-cell can't be transform into kilometer-cell directly.

If directly, the width of each cell

$$\Delta y=2\pi R_e\frac{\Delta lat}{360^\circ}$$

The length of each cell:

$$\Delta x=2\pi R_e\cos(lat)\frac{\Delta lon}{360^\circ}$$

So, every cell has its own size in kilometers. The uneven grid networks enhance the difficulty for simulation. I'm wondering to transform the inventory into uniform grid in kilometers(for example: 0.25 degree x 0.25 degree -> 36 km x 36 km)

How to achieve it? I'm familiar with Python language.
Any advice would be appreciate!

Community
  • 1
Han Zhengzu
  • 2,459
  • 1
  • 13
  • 25
  • Which projection do you use for the 36 km x 36 km grid? – daniel.heydebreck Jul 05 '16 at 11:00
  • 1
    Remark 1: The approach, which was described here, to approximate the grid cell size is based on the assumption of ideal spherical Earth. This is fine for a first order guess of the grid cell size but it should not be used when we want to know the 'real' size of the grid cell (because the Earth is not ideal sphere). – daniel.heydebreck Jul 06 '16 at 14:18
  • Remark 2a: The Earth can be differently projected onto a plane. One can preserve some metrics (geometric length, angles, area) of the original curved Earth surface when it is projected onto a plane but not all metrics can be preserved. Thus, there are projections (see Wikipedia) that preserve the area and other projections that preserve angles -- for example. Preserving on metric might yield considerable distortions of other metrics. – daniel.heydebreck Jul 06 '16 at 14:31
  • 1
    Remark 2b: Preserving two metrics is generally problematic and not possible for the whole globe. However, a projection might approximately preserve these metrics in a predefined region. See for example the Lambert conic equal area projection. We, for example, use these type of projection for our chemistry transport model simulations. – daniel.heydebreck Jul 06 '16 at 14:38
  • Remark 3: If we have decided on a projection (e.g. equal area projection = preserve area), we can interpolate our data from the lat-lon grid onto our new grid (in the new projection). The emissions should have the dimensions 'mass/(time * area)' and not 'mass/time' (instead of 'mass' it might be 'number') to avoid some extra work ... . – daniel.heydebreck Jul 06 '16 at 14:45
  • Thanks for your detailed reply. So, once deciding the projection for specific area(left low corner -> right up corner), I should (1) extract the corresponding emission data from the source inventory(Regional/Global), (2) design the source grid network(as, 2 km x 2 km) and all grids center point in lat-lon, (3) interpolate these center point from the extracted emission data. Is that right to go? – Han Zhengzu Jul 07 '16 at 07:20
  • 1
    sounds reasonable :-) – daniel.heydebreck Jul 07 '16 at 08:03
  • Thanks for your reply! By the way, did you need to pre-treat the source inventory like this? If yes, what tools did you use? – Han Zhengzu Jul 07 '16 at 08:37
  • 1
    A colleague of mine does the emission preparation for our whole working group. He uses his own interpolation routines. I use cdo (cdo remap) when I do interpolations. Maybe we should switch to the chat, if you have more questions? – daniel.heydebreck Jul 07 '16 at 09:25
  • Thanks for your patience! If I met some trouble, I'll consult you for help. – Han Zhengzu Jul 07 '16 at 23:52

0 Answers0