MATLAB: Direction of Arrows in Quiver (2D)

quiver

this is related to the question asked here:
http://www.mathworks.com/matlabcentral/answers/48618-how-to-set-colors-in-quiver-2d
so I did make to seperate quivers for the inwards and outwards arrows. the new problem is, that some of my arrows that i think should be pointing outwards are pointing inwards.
as can be shown here: http://tinypic.com/view.php?pic=6zvdwk&s=6
here is how i created this, tell me where you think i am wrong:
–
–
% find slope of profile
m_profile(2:45, :)=diff(z1)./diff(x1);
m_profile(1,:)=1.7612;
m_profile(29,:)=0;
–
–
%find arrow slope according to m2=-1/m1
m_arrow=-1./m_profile; theta=atand(m_arrow);
–
%create mark vector – indicating which arrows should point inweard(negative cp values)
for j=1:num
for i=1:45
    if cp_norm(j,i)<0
          mark(i,j)=1; %mark arrows pointing inwards
end end end
    % sort flagged control points - create 2 matrixes - one for inwards(flag) and one for outward (nflag) arrows
                k=1; l=1;
                flag=[];
                nflag=[];
                for i=1:45
                    if mark(i,j)==1
                        flag(k,1)=x1(i,j);
                        flag(k,2)=z1(i,j);
                        flag(k,3)=cp_norm(j,i)*cosd(theta(i,j));
                        flag(k,4)=cp_norm(j,i)*sind(theta(i,j));
                        k=k+1;
                    else
                        nflag(l,1)=x1(i,j);
                        nflag(l,2)=z1(i,j);
                        nflag(l,3)=cp_norm(j,i)*cosd(theta(i,j));
                        nflag(l,4)=cp_norm(j,i)*sind(theta(i,j));
                        l=l+1;
                    end
                end
still the result is some of the arrows are pointing inwards as you can see, where am i wrong?
thanks,
Noa

Best Answer

I'm gonna venture a wild guess. We don't know what your data looks like (well, except from the plot). However, this line:
m_profile(2:45, :)=diff(z1)./diff(x1);
would indicate that you are taking some sort of slope. However, you are only considering the forward difference. Maybe an average, the central difference, would give the results you want.
Related Question