You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Add exclusion for marine zones with high traffic #20
Is your feature request related to a problem? Please describe.
Our potentials for offshore area are likely going to be a bit optimistic as we do not have a method to exclude ports with and areas with high ship traffic.
It details AIS positions between Jan 2015 and Feb 2021 (~6 years)
PyPSA-Eur implements this exclusion with the following heuristic:
$$Yearlysightings * DaysInYear * TotalYears = 400 * 8760 * 6 = 21,024,000$$
It's a bit of an ugly number, but it should be fully compatible with the module's methods.
An important detail is that this dataset should only be used if specified, as the unzipped file is 9 GB in total.
Example
I've been playing around with adding this, and it seems possible with relatively minor changes. One difference is that PyPSA uses nearest resampling, which is fast but not as good (I think...) as the average resampling method we are using in the module.
Here are some results I got after playing around with a local implementation.
I'd say they are mostly equivalent, although nearest leads to overly pessimistic results (higher drops).
Signed difference looks like resampling noise, mostly.
Counts and fractions of valid maritime area:
nearest excluded 24,836,038 cells 21.877278%
average excluded 20,706,773 cells 17.964159%
mask disagreement 4,318,521 cells 4.088983%
nearest only 4,223,893 cells 4.001051%
average only 94,628 cells 0.087932%
both excluded 20,612,145 cells 17.876227%
neither excluded 91,787,487 cells 78.034790%
Activity
irm-codebase commentedon Aug 13, 2026
Here are some results I got after playing around with a local implementation.
I'd say they are mostly equivalent, although
nearestleads to overly pessimistic results (higher drops).Signed difference looks like resampling noise, mostly.