Some thoughts after a quick look
- The streaking artefacts come from differing Cd and density volume extents, add Cd (and CdFuel?) to Extra Resize Fields in you Gas Resize node
- The particle to volume settings don't necessarily match between CdFuel (done in SOPs) and density (done in DOPs), so there might be some less-colored density created
- Some of the color loss might be due to opengl preview..
- The differently coloured smoke parcels will unavoidably mix as the simulation advances. You can probably mitigate that by using more substeps, and maybe by using a higher order Advection Method (HJWENO?). Remember to use the same method in your color advect node as you use in the smoke solver.
This all would need quite a bit of testing to see if it actually helps, so take it with a grain of salt