Interpolate both dataset on a same grid

Hi, i have the two different datasets (PFA the excel file) as following

Dataset A size (119*177) with each lat and long file of same size

Dataset B size (145*73) with long size of (145*1) and lat size of (73*1)

The steps according to my understanding are-

• Make a lat, long file for dataset B of size (145*73) - question how to do that?
• Interpolate the dataset A to the size of Dataset B, which i am guessing using interp2

[Jan]

What are your inputs? "PFA the excel file" is not clear. Are the "data set" simply matrices? Do you really need to "make a file"?

[Sophia]

Sorry, i forgot to attach the file. PFA the file now!

[Jan]

Is importing the Excel file a part of your problem? If not, wouldn't it be less confusing to start with: A = rand(199, 177), B = rand(145, 73)? Does it matter that I do not know what "PFA" means? But it does matter, on which grid the data are recorded. Maybe the X positions of A go from 1 to 177, while the positions of B go from -35 to 37. Before we can suggest a method to apply the grids to A and B, you have to explain the details.

[Sophia]

Hi, The reason why i attached the file was to give a detailed idea about the matrix sizes.

The simplest way to define my problem is i have the results from two different datasets which are on two different lat/long grids. I want to interpolate them on to a common grid and find the normalized two-dimensional cross correlation.

I tried this for my first question ( Make a lat, long file for dataset B of size (145*73) - question how to do that?), and it seem to work but not sure until i have the final result

%%change the long and lat from 1D to 2D
  [mm,nn] = size(skw_16); %skw_16 is of the size 119*117

    %long is 1*145, lat is 1*73 %change them to 145*73

    nlong = zeros(mm,nn); nlat = zeros(mm,nn);

    for i = 1:mm  

        nlong(i,:) = lat(:)'; %nlong(i,:) = long(longstart:longend)';

    end

    for j = 1:nn

        nlat(:,j) = long(:)'; %nlat(i,:) = lat(latstart:latend)';

    end
  %The result %nlong 145*73, nlat 145*73

Now i am onto my second question- *Interpolate the dataset A to the size of Dataset B, which i am guessing using interp2*

skw_mtn_ncep = interp2(long,lat,skw_ids_2016,nlong,nlat);

%long 119*177, lat 119*177, skw_ids_2016 119*177

or should i interpolate the 145*73 onto 119*73?

[Jan]

An abbreviation for the first part:

nlong = repmat(lat, mm, 1);
nlat  = repmat(long(:), 1, nn);

But you do not need to create matrices with the grid coordinates, because interp2 accepts vectors directly.

See my answer...

Answer 1

[Jan]

I assume that this helps:

B  = rand(145, 73);
A  = rand(119, 177);
y  = linspace(1, 145, 199);
x  = linspace(1, 73, 177);
BB = interp2(1:73, (1:145).', B, x, y.');

Now BB is the linear interpolation of B with the same size as A.

Note: I still find the order of the coordinates for interp2 counter-intuitive:

 Vq = interp2(V,Xq,Yq) assumes X=1:N and Y=1:M where [M,N]=SIZE(V)

I'd expect the 1st input to be correlated to size(V, 1) and the 2nd to size(V,2), but MathWorks decided to swap the order. This is at least consistent with e.g. gradient.

[Sophia]

Okie, here comes the part why i attached the excel file. From my understanding of your code A and B are two datasets. But x and y, i am not sure what they are representing here. In my case i have lat and long for each A and B. And the max and min lat/long values in both case are different.

I am not sure if i am explaining my question well enough or not. let's assume for example you have the data for United States at a grid of 25km*25km. And another dataset in which the data might be extended upto Canada but still has the data for United States, at a grid of 2.5 degree.

The idea is to plot them on a common grid, and find the 2D cross-correlation between them for just the common areas.

so in that case-

y  = linspace(1, 145, 199);
x  = linspace(1, 73, 177);

I am wondering, if its the right way?

Answer 2

[Sophia]

I just received this e-mail. I am worried how come MATHWORKS can share my personal email with any random person. The e-mail is as following-

This message was sent to you by this author via your author page on MATLAB Central.

Hello, I read your question in Matlab Central and would like to offer you my instant 1-to-1 programming service. An affordable, quick and tailored matlab-programming service to your needs. Check it out at: www.bit.ly/******** Thank you*

[Jan]

@sophia: MathWorks did not publish your personal mail address. On your profile page, you have enabled the possibility to contact you through the forum's interface: https://www.mathworks.com/matlabcentral/profile/authors/5688027-sophia - see the "Contact" button below you profile picture on the left. On the setup of your profile you can deselect the "Allow users to contact me securely" checkbox, if you do not want this.

Unfortunately having this contact service enable means, that you get a lot of time wasting messages: "Solve my homework ASAP!", "Didn't you read my question?", "I offer 5$ for solving my final project" and "Send me code for simulating an X-boson vector search for anti-matter delta distributions in the co-valent subspace using the most modern KI system."

[Stephen23]

"I am worried how come MATHWORKS can share my personal email with any random person"

Just to make it clear: that person does NOT have your email address! They sent a message via your user profile, and the email was sent by TMW on their behalf. If you reply to that email then that user will then have your email address.

[John D'Errico]

I'm sorry to see you had this spam sent to you. As Jan and Stephen said, your address is NOT made public. But they do offer a way for people to contact you, and now it seems that someone is using that service as a means to spam people.

It seems this person is now beginning to harass people with their spam. You are the second person to have reported the offer of work for pay, which now implies they are doing it to many people. After all, if two people are so upset that they complain online, there are hundreds of people who will have been affected.

I'll report the problem to the proper authority, as I did the last time I saw someone complain. I hope we can help to prevent more such spam being generated from people's accounts.

[Jan]

Offering a programming service in a Matlab forum is not completely off-topic, such that I'm not sure if this should be classified as spam. But of course the nature of the forum is sharing solutions in public, therefore I do not like such advertisements.

But I strongly recommend not to click on any bit.ly link, which is sent without a request by mail or through the contact form of this forum. This could be a cheap trick to let you open an infected page.

[Stephen23]

"Offering a programming service in a Matlab forum is not completely off-topic"

However it a clear violation of the terms of use for this forum:

1.e "You agree ... nor to use the communication systems provided by the Site (e.g., comments) for any commercial solicitation purposes."

2.a.iv "You may not use the Site to sell or market your products or services to others."

[Sophia]

thanks guys it helped, the Matlab Central send me the mail assuring none of my information has been compromised. Thanks for the flag.