I want to know the method for removing excessive angle data.

I want to remove data, if current direction larger or smaller about mean current direction +15, - 15(degree), respectively.

First figre, I can remove data by up and down line(mean current direction +15 or -15)

But case of Figure2, some datas exceed 360(degree), So I don't how to remove unneedful data....

[Mann Baidi]

Hi,

Can you share the code and the data, it will be helpful to understand the issue.

[재혁]

for i = 1:2:(length(SoundSpeed)-2)
    U_vec2(i,:) = nanmean([U_vec(i,:) ;U_vec(i+1,:)],1); % 5분간격으로 되어있지만 10분간격으로 만들기 위해 2간격으로 평균을 하였음
    V_vec2(i,:) = nanmean([V_vec(i,:) ;V_vec(i+1,:)],1);
end
U_vec2(find(U_vec2(:,1) == 0),:) = []; V_vec2(find(V_vec2(:,1) == 0),:) = [];
Current_dir = (atan2(-U_vec2(:,1:19),-V_vec2(:,1:19)) + pi)*180/pi;
Magvel2 = sqrt(U_vec2.^2+V_vec2.^2); % magniture^2 = U^2 + V^2
mean_Current_dir = nanmean(Current_dir,2);
%%
for i = 1:length(Current_dir)
    indind{i} = find(Current_dir(i,:) > mean_Current_dir(i) + 15 | Current_dir(i,:) < mean_Current_dir(i) - 15);
end
for i = 1:50:length(Current_dir)
    figure(i);
    scatter(log(Depth(1:19)),Current_dir(i,:)); hold on
    yline(mean_Current_dir(46));
    yline(mean_Current_dir(46)+15);
    yline(mean_Current_dir(46)-15);
    ylabel('Current direction (^o)','fontname','arial','fontweight','bold','fontsize',20);
    xlabel('Depth (mab)','fontname','arial','fontweight','bold','fontsize',20);
end

here to my code,

and my current direction data is degree(0 ~ 360), so if mean current direction is 340 then, my code can not identify 0 ~ 10

My data length is approximately 3000

[Mann Baidi]

for i = 1:length(Current_dir)
    indind{i} = find(Current_dir(i,:) > mean_Current_dir(i) + 15 | Current_dir(i,:) < mean_Current_dir(i) - 15);
end

What is the significance of 'indind' variable?

[재혁]

This is a variable used to identify values that exceed or fall short of 15 degrees from the mean flow direction.

Answer 1

[Mathieu NOE]

hello @재혁

IMO , there is no need to use a for loop to compute 'indind' variable

also nanmean is obsolete, you can use mean with 'omitnan' option instead.

here a very simplified code to show how to remove (replace) values exceeding your limits with NaN's

Current_dir = 45*rand(100,1); % random numbers between 0 and 45 

% mean_Current_dir = nanmean(Current_dir,2); obsolete
mean_Current_dir = mean(Current_dir,'omitnan');

%% No need for a for loop here 
% for i = 1:length(Current_dir)
%     indind{i} = find(Current_dir(i,:) > mean_Current_dir(i) + 15 | Current_dir(i,:) < mean_Current_dir(i) - 15);
% end

indind = (Current_dir > mean_Current_dir + 15) | (Current_dir < mean_Current_dir - 15);

% create data and remove values above and below threshold
Current_dir_cor = Current_dir;
Current_dir_cor(indind) = NaN; 


scatter(1:numel(Current_dir),Current_dir); hold on
scatter(1:numel(Current_dir),Current_dir_cor,'*'); 
yline(mean_Current_dir);
yline(mean_Current_dir+15);
yline(mean_Current_dir-15);