u/v versus u_eastward, v_northward

[noah.eckman]

I've come across something that has stumped me with regards to how ROMS calculates u_eastward and v_northward from the u/v velocities. I took out the data from one of my history files, and I saw this (see below).

My question is, it seems as if U is interpolated in to v_northward, and V interpolated to negative u_eastward (effectively rotating the grid by 90 degrees.) Why is this? Is this a feature or a bug? This has caused us some confusion and I look forward to any help on the matter.

Thanks.

[kate]

This is consistent with your u-velocity (i-axis) being in the northward direction and v-velocity (j-axis) being in the westward direction. How did you set up your grid? What is the value of "angle" on your grid? Close to 90 degrees?

[noah.eckman]

Hmm, it should be u (i-axis) west/east and v (j-axis) north/south. The angle on the grid is close to 0, about 1.5 (the domain is in the mid-latitudes).

The grid is set up just on a mercator curvilinear grid. (It's setup with some matlab code that originally was from ROMS.)

The lon's and lat's are as you would expect in the grid file, the with longitude changing with respect to the i, and latitude changing with respect to j. So I don't think that the whole grid is rotated somehow.

[kate]

Angle is in radians - 1.5 would explain it.

[noah.eckman]

Thanks-- I'll look in to it.