I'm trying to convert a 1 dimension lat,lon,data to 2d data . Is this possible?

I have a one dimensional latitude,longitude, data. Essentially all are in one dimension. However I want to create a 2D plot with the data. So, what I want is the data should be in 2D format without using the griddate which uses interpolation. So, in 2D format of data there should be original value of 1D data present at every original latitude, longitude point and other points it will be NaN or zero.

Answer 1

[Chunru]

Updated solution:

T1 = readtable('chiranjit Das DAY_1.xlsx');
dg = 1;
longrid = (-180:dg:180);
latgrid = (-90:dg:90);
z = nan(length(latgrid), length(longrid));
for ilat=1:length(latgrid)
    for ilon=1:length(longrid)
        ii = T1.LATITUDE-latgrid(ilat)>=-dg/2 & T1.LATITUDE-latgrid(ilat)<dg/2 & ...
            T1.LONGITUDE-longrid(ilon)>=-dg/2 & T1.LONGITUDE-longrid(ilon)<dg/2;
        z(ilat, ilon) = mean(T1.DATA(ii));
    end
end
figure
imagesc(longrid, latgrid, z);
axis xy
xlabel('lon');
ylabel('lat');
colorbar

figure;
cmap = turbo(1024); cmap = cmap(257:768,:);
ax = axesm('MapProjection','apianus','MapLatLimit',[-90 90], 'MapLonLimit', [-180 180]);
[lon, lat] = meshgrid(longrid, latgrid);
geoshow(ax, lat, lon, z, cmap , 'DisplayType', 'image')

Your description of the problem is not detailed. Is this what you want?

n = 100;
lat = rand(n, 1);
lon = rand(n, 1);
z = rand(n, 1);
figure
stem3(lon, lat, z);
xlabel('lon'); ylabel('lat'); zlabel('z');

[Deepshikha Dadhwal]

Hello Chunru,

I tried plotting TROPOMI data using the code you provided here but I am unable to store the data in the 'z' variable. TROPOMI data is originally in 2D format and I converted it into 1D to fit into the code.

I tried regridding with "Griddata" function as well but while going for coarser resolutions, the griddata function gives fewer and fewer data points. Both the codes are attached and the a single datafile could be downloaded from the following link as I could not attach my own due to size issues.

Tropomi data for methane

My aim is to regrid the data from original resolution to a desired resolution (ex. 2.5x2.5 degree) and count the number of data points falling per grid.

Answer 2

[Steven Lord]

My aim is to regrid the data from original resolution to a desired resolution (ex. 2.5x2.5 degree) and count the number of data points falling per grid.

Based on this comment you made on the answer by @Chunru I think you may not need or want to convert your data from a list of points to something approximating a grid. Instead I think you just need to call histogram2.

[Deepshikha Dadhwal]

Hello Steven,

I used histogram2 function and I could count the number of datapoints within the specified equally no. of grids, but it only works when my desired no. of Latitude-Longitude coordinates are equal, but that still solves problem number 2 but problem no.1 still remains as

1.) I am not able to change the resolution in "Griddata" method, the coarser resolution gives fewer data points.

2.) Unable to store the same data using @Chunru's method.

Basically, if I could get the mean of the data falling into those grids, my problem would be solved.

Answer 3

[Bruno Luong]

T = readtable('chiranjit Das DAY_1.xlsx');

A = table2array(T);
lat=A(:,1);
lon=A(:,2);
data=A(:,3);

longrid = -180:180;
latgrid = -90:90;

[~,~,~,i,j] = histcounts2(lat,lon,latgrid,longrid);
B = accumarray([i,j],data,[length(latgrid),length(longrid)]-1,@mean,NaN);
imagesc(longrid,latgrid,B);

[Deepshikha Dadhwal]

Thanks Bruno.

The plot is flipped though with the excel sheet data and Tropomi data as I mentioned before is 2D. So when I am converting it into 1D and then proceeding with the code, it's messing up the results.

[Bruno Luong]

Nah the plot is flipped by default when using imagesc, you only need to call

set(gca,'YDir','normal')

This is only visual. The underline data is correct, meaning bin i=1 corrspond to (-90,-89), etc...

I did not (and sillt do not) read what you wrote before sorry.

[Deepshikha Dadhwal]

This is the data I am working on. I am attaching the code as well which I tried to run for it. (I did the griddata and the histogram2/histcounts2 separately).

I could count the no. of data points inside the grids using Histcounts2.

I could not

a) Regrid to a coarser resolution or

b) collect and take the mean in this case and count the data falling into those grids together in the same code.

Your code works well for 1D data that I work with but it seems to create errors while working with this data.