How can i interpolate this data?

Greetings to you all.

I'm trying to analise some data from an ADCP (Acoustic Doppler Current Profiler). For that, i must process the data.

My matrix works on this way:

Each Row has intensity values

Each column has a certain height in relation to the bottom of the water column.

example:

0 NaN 0.5 0.3 0.7 0.9

0 NaN 0.7 0.4 0.5 0.8

0 NaN 0.9 0.1 0.4 0.5

0 NaN 1.0 0.4 0.3 0.2

The second column is equivalent a 1m height from the bottom and the instrument has not registered those values because of an error. I want to interpolate them.

Appreciate any kind of help.

Answer 1

[Star Strider]

If I understand the problem correctly, use the fillmissing function across the rows —

A = [0     NaN     0.5     0.3     0.7     0.9
     0     NaN     0.7     0.4     0.5     0.8
     0     NaN     0.9     0.1     0.4     0.5
     0     NaN     1.0     0.4     0.3     0.2];

B = fillmissing(A, 'linear', 2)

B = 4×6

0 0.2500 0.5000 0.3000 0.7000 0.9000 0 0.3500 0.7000 0.4000 0.5000 0.8000 0 0.4500 0.9000 0.1000 0.4000 0.5000 0 0.5000 1.0000 0.4000 0.3000 0.2000

The fillmissing function was introduced in R2016b.

.

[Gabriel Luca Pugliese Borges]

Hello again!

Thanks for the help, i appreciate it.

Tried using this function but it seems that it was introduced later than 2016

[Star Strider]

My pleasure!

It was introduced in R2016b.

Alternatively:

A = [0     NaN     0.5     0.3     0.7     0.9
     0     NaN     0.7     0.4     0.5     0.8
     0     NaN     0.9     0.1     0.4     0.5
     0     NaN     1.0     0.4     0.3     0.2];

B = [interp1((0:1),A(:,[1 3]).',linspace(0,1,3),'linear').' A(:,4:end)]

B = 4×6

0 0.2500 0.5000 0.3000 0.7000 0.9000 0 0.3500 0.7000 0.4000 0.5000 0.8000 0 0.4500 0.9000 0.1000 0.4000 0.5000 0 0.5000 1.0000 0.4000 0.3000 0.2000

For whatever reason, using the entire matrix to do the interpolation did not work as I expected it to, so interpolating the first three columns and then concatenating it with the last three columns is necessary here.

.

[Gabriel Luca Pugliese Borges]

interesting! In my real matrix i have more than 1 page, how should i proceed?

A = 112254x12x6

[Star Strider]

I was away doing other things for a few minutes.

I thought of a more efficient way to do this, as well as being able to do more than one page in one operation —

A = [0     NaN     0.5     0.3     0.7     0.9
     0     NaN     0.7     0.4     0.5     0.8
     0     NaN     0.9     0.1     0.4     0.5
     0     NaN     1.0     0.4     0.3     0.2];
A = cat(3, A, A.*(1+randn(size(A))/100))

A = 
A(:,:,1) =

         0       NaN    0.5000    0.3000    0.7000    0.9000
         0       NaN    0.7000    0.4000    0.5000    0.8000
         0       NaN    0.9000    0.1000    0.4000    0.5000
         0       NaN    1.0000    0.4000    0.3000    0.2000


A(:,:,2) =

         0       NaN    0.5088    0.2997    0.7131    0.8894
         0       NaN    0.7053    0.4002    0.4915    0.7957
         0       NaN    0.8982    0.1012    0.3961    0.4958
         0       NaN    0.9938    0.4026    0.3011    0.2003

B = A;                                                      % Copy 'A' To 'B'
B(:,2,:) = mean(B(:,[1 3],:),2)                             % Column 2 Is The Mean Of Colums 1 & 3

B = 
B(:,:,1) =

         0    0.2500    0.5000    0.3000    0.7000    0.9000
         0    0.3500    0.7000    0.4000    0.5000    0.8000
         0    0.4500    0.9000    0.1000    0.4000    0.5000
         0    0.5000    1.0000    0.4000    0.3000    0.2000


B(:,:,2) =

         0    0.2544    0.5088    0.2997    0.7131    0.8894
         0    0.3526    0.7053    0.4002    0.4915    0.7957
         0    0.4491    0.8982    0.1012    0.3961    0.4958
         0    0.4969    0.9938    0.4026    0.3011    0.2003

The second ‘page’ or ‘A’ is a slightly altered version of the first page to demonstrate that this works, providing that a linear interpolation is desired, and the column 2 of every page is the NaN column.

.

Answer 2

[Keegan Carvalho]

I'd assume "fillmissing" function would be best since you want to inteprolate the data row-wise (and this is not gridded interpolation).

Try this:

mydata = [0     NaN     0.5     0.3     0.7     0.9
          0     NaN     0.7     0.4     0.5     0.8
          0     NaN     0.9     0.1     0.4     0.5
          0     NaN     1.0     0.4     0.3     0.2];

mydata=fillmissing(mydata,"linear",2) 

% linear is one of the inteprolation methods you can use. There are others like spline, nearest, etc.
% 2 means inteprolation of data in each row of mydata. 1 - column

Hope this helps!

[Gabriel Luca Pugliese Borges]

Hello.

Thanks for the help, i appreciate it.

Tried using this function but it seems that it was introduced later than 2016

maybe you know another function