dougcahl / eddy_identification_winding Goto Github PK

Hi dougcahl,

I'm using your code for the winding angle detection approach and realize that there're probably a bug in the eddy_subroutine function.

Around line 170, you detect if the direction of the velocity vector changes from the previous components. I notice that you convert the ang_diff to a reasonable range ([0, pi] or [-pi, 0]) here after this check but I guess this should before this check. Otherwise, the ang_diff could reach a huge number like 330 in my case.

        if ii > 2 % only after first winding angle
            newdir = sign(ang_diff);
            if newdir ~= dir % if winding changes direction
                baddir = baddir + ang_diff;
                if abs(baddir) > baddir_thres % break after this many degrees bad dirs
                    break
                end
            end
            if newdir == dir % reset if goes back to correct direction before baddir_thres
                baddir = 0;
            end
        end




        if ang_diff > 180
            ang_diff = ang_diff - 360;
        end
        if ang_diff < -180
            ang_diff = ang_diff + 360;
        end  

I switch the sequence of this section and it works well but let me know if you are trying to do something else.

dear dougcahl
Thank you for providing the code for eddy extraction and tracing.
When I run the program (streamplot_dc.py, line 96,
cbook._check_in_list(['both', 'forward', 'backward'],
integration_direction=integration_direction)), the following problem occurs: AttributeError: module 'matplotlib.cbook' has no attribute '_check_in_list'.
What is the reason for this? Is it the matplotlib version?
Thanks.