search

MIDOSS / analysis-rachael

Apache License 2.0
0 stars 0 forks source link

WA_origin, WA_destination #2

Open SusanEAllen opened 4 years ago

SusanEAllen commented 4 years ago

Make sure all choices add to 1. Don't leave some as all zeros as this is brittle. Just make a best guess.

RachaelDMueller commented 4 years ago

@SusanEAllen, what you’re seeing (and I know you know this...I’m just spelling it out for communal clarity) is the heterogeneity of traffic between locations. Some locations receive oil from ATBs, some do not. The zeros reflect places where there is no oil transfer for particular vessels. This is fixed by the DOE data. Instead of introducing guesses to the DOE data file, my preference would be to handle the case of zero values separately. This would look something like: If the sum of fuel fractions is not one then use the more generic fuel allocations, which we have set as “US”, “Pacific”. I’m leaving CAD out b/c we don’t have data for CAD and don’t have this issue for those origins/destinations. This option effectively says: there is a mistake in our origin-destination methodology but instead of throwing out this case and selecting a new location/vessel-type to fix the mistake, we are going to keep the selected location and vessel (which would be tanker, ATB or tug) and use a generic fuel allocation as cargo.

This is what I have been planning, and I think this approach addresses your concern. Are you okay with this approach instead of the “fix the data” approach?

SusanEAllen commented 4 years ago

All zeros are too brittle. Not adding to 1 is a nice easy check for consistency in the data and using it for this purpose instead will not allow that check. Given the difficulty in attributing types of vessels to ATB and tank barge, I think it would make sense to propagate that information into the file. You can add a note to the file that you are doing that. It is probably a better guess than the generic case.

RachaelDMueller commented 4 years ago

I think we resolved this. I’m going to leave the data as is with 0s for cases of no gallons of oil transfers and will use “if sum of fuel fraction equals zero” to attribute fuel based on the generic US allocation (which combines both import and export). Yes? @SusanEAllen, can I close this or would you like to leave the thread open for discussion/comment?

SusanEAllen commented 4 years ago

Fine to close.