Closed llx206 closed 3 years ago
Hi, this usually happens when your time step is too large. Use a CFL approach to choose your time step in o.run considering the magnitude of the currents and the grid size of your original model.
hank you! I will try it according to your suggestion. In addtion,Is the resolution of landmask fixed ?
Yes, the automatic landmask is now always used at the highest resolution (GSHHG-full).
From your plots it seems like the time_step is on the order of a day, but this should normally not be more than 1 hour, ideally 15-30 minutes for oil simulations. You may still use a coarser output timestep: https://opendrift.github.io/tutorial.html#running-the-model
Thank you! And what about the first question?
This is probably related to too large time step, combined with seeding over a time interval.
Do you still see the same when using a smaller time step (e.g. 30 minutes)? What are your seed- and run-commands?
stime = datetime(2020, 6, 22, 7, 0, 0) time = [stime,stime + timedelta(hours=24)] mass_oil = 2000 oiltype = 'MEDIUM (brage)' o.seed_elements(lon, lat, radius=50, number=3000, time=time, mass_oil=mass_oil, wind_drift_factor=0.02, oiltype=oiltype) o.run(end_time=datetime(2020, 6, 23, 7, 0, 0), time_step=1800,time_step_output=3600)
i find that if the timedelta is set to 0 then the phenomenon in question 1 no longer exists .
This is probably due to different calculation- and output-intervals (0.5hr vs 1 hr). Thus the position of elements which are seeded later than the initial time, are only recorded at a later step when they have moved for up to one hour. This artefact will also disappear if you use same output as calculation time step (e.g. time_step=time_step_output=1800), even if you seed elements over a time interval.
thanks, i get it.
HI, I have some questions
figure 2