Temperature blowing up near boundary while use nudge

[zhiguoMEI]

Hello everyone! I was run a west pacific model ,and I want to use nudge to imporve the performance of bottom temp and salt
I made the nudge.nc file by myself. I set the tracer_NudgeCoef as 1day at the bottom level and decrease to 1/30 day at the 5th level.
But after the 5-month run, an abnormal increase/decrease in temperature at the edge (please see the figure below).

I check my boundary file there is no promble.
could anyone give me some advice?

[kate]

If the boundary is at the edge of the plot, you want to mask out those two purple points. Otherwise you should unmask the point just north of them. At the boundary, the algorithm is using values a few gridpoints in from the boundary so they all have to be wet for it to be well-behaved.

[zhiguoMEI]

If the boundary is at the edge of the plot, you want to mask out those two purple points. Otherwise you should unmask the point just north of them. At the boundary, the algorithm is using values a few gridpoints in from the boundary so they all have to be wet for it to be well-behaved.

Thanks Kate for your advice , and yes my boundary is at the edge of the plot,I'll try to mask the error points to see if it works well.

[zhiguoMEI]

If the boundary is at the edge of the plot, you want to mask out those two purple points. Otherwise you should unmask the point just north of them. At the boundary, the algorithm is using values a few gridpoints in from the boundary so they all have to be wet for it to be well-behaved.

Hi Kate! I mask the three error points near the boundary and the error temperature still appears after 5-month run! Are there any problems while nudging the tracers (temp/salt) at the bottom level!

微信截图_20201208125251.png (10.62 KiB) Viewed 2453 times

[kate]

So this is only in the bottom temperature and salinity? What are the velocities doing? Rather than nudging to cover it up, it would be better to understand where it is coming from.