Bilinear interpolation of gridded data

I have GCM data for a region with grid size of 2.81 x 2.79 degree(lat x lon) where starting latitude range is 4.9 degree to 41 degree and longitude range is 59 degree to 110 degree. So i need to interpolate this data on a 2.5 X 2.5 degree with latitude range 5-40 degree and longitude range 60-100 degree. how do i use the meshgrid aND LOOPS to do the interpolation

[darova]

Can you attach your data?

[Diljit Dutta]

here I have attached the tabular data of variables.. which is a 3D array of 23x14x3355. where 23 is number of longitudes from 59.06 - 120.93 degree with spacing of 2.81 degree and 14 is number of latitudes from 4.18 - 40.46 degree and 3355 is the time scale daily data for 3355 days. Now it is needed to be interpolated on 2.5 x 2.5 scale with longitude range 60-100 and latitude range 5-40 degree.

Answer 1

[darova]

Hi! This how i do this

clc,clear

load Variables.mat
[m,n,k] = size(predictor);

lon = linspace(59.06,120.93, m);
lat = linspace(4.18,40.46, n);
lon1 = 60:2.5:100;                      % new range
lat1 = 5: 2.5:40;

[LAT, LON] = meshgrid(lat,lon);         % original mesh
[LAT1, LON1] = meshgrid(lat1,lon1);     % new mesh

predictor1 = zeros([size(LAT1) k] );            % preallocation
for i = 1:k
    pred = predictor(:,:,i);
    predictor1(:,:,i) = interp2(LAT,LON,pred,LAT1,LON1);
end

    % DRAW SOMETHING !
R = predictor(:,:,1);
R1 = predictor1(:,:,1);
s = pi/180;
[x,y,z] = sph2cart(LAT*s,LON*s,R);
[x1,y1,z1] = sph2cart(LAT1*s,LON1*s,R1);
%%

plot3(x,y,z,'b')        % original data: blue lines
hold on
surf(x1,y1,z1)          % interpolated data: color surface
hold off
axis equal

[Diljit Dutta]

why is it necessary to use spherical coordinate and then converting to cartesian for plotting?

[darova]

Because MATLAB doesn't support spherical system of coordiantes

Only cartesian and polar(2D)

[darova]

You can plot something like this

surf(LAT,LON,R)

BUt what is it? How to understand it?

[Diljit Dutta]

ok I thought surf(LAT,LON,R) would give a variation of the parameter over a region given by the coordinates. So on this data I further need to do bias correction by Quantile matching with a observed field data. So for that how do I proceed. Can u suggest.

[darova]

I know nothing about such correction. Sorry

Answer 2

[Nicolas B.]

Hi,

may I recommand you to take a look at the function griddedInterpolant to solve your case? It will be definitely faster than a solution with loops.

Regards

[Diljit Dutta]

for that i need to convert tabular data to some where everything is assigned to particular latitude and longitude. I am having problem with that

[Nicolas B.]

Do you have any example of typical input data?

[Diljit Dutta]

here I have attached the tabular data of variables.. which is a 3D array of 23x14x3355. where 23 is number of longitudes from 59.06 - 120.93 degree with spacing of 2.81 degree and 14 is number of latitudes from 4.18 - 40.46 degree and 3355 is the time scale daily data for 3355 days. Now it is needed to be interpolated on 2.5 x 2.5 scale with longitude range 60-100 and latitude range 5-40 degree.