Patch not working properly with 2d latitude and longitude

Hi all,

i am trying to plot the sea surface temperature from a netcdf file. when i upload it , i have this

i tried to use patch in this way

patch(longitude ,latitude , mean(tos,3,'omitnan') , 'FaceColor','interp', 'EdgeColor' , 'interp' )

but the results is not good , i have temperature over the land , so i think something is wrong with my grid, but

I can't understand why. If you have any suggestion to help I would very appreciate it.

Thanks

Michele

Answer 1

[Voss]

Try using surface instead of patch.

[MICHELE GNESOTTO]

Thank you for your suggestion but I have the same problem , this is the plot that i obtain.

[Voss]

Hmm, the plots I get look a little different:

unzip('adaptor.esgf_wps.retrieve-1652740104.757926-1767-15-d7ea9cac-5f2f-4b52-9b1f-0ba84ab0122b.zip')
longitude = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','longitude');
latitude = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','latitude');
tos = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','tos');

% patch
figure();
patch(longitude,latitude,mean(tos,3,'omitnan'), ...
    'FaceColor','interp', ...
    'EdgeColor','interp');

% surface
figure();
surface(longitude,latitude,mean(tos,3,'omitnan'), ...
    'FaceColor','interp', ...
    'EdgeColor','interp');

[MICHELE GNESOTTO]

yes , the one that i posted is with surface, i think i did not interpolate the edges , for the color , but is still not good you can see that there are of south america is colored when is not supposed to be since there is no sea in there. actually i reshape it with cdo and it seems to work fine, but i would still like to find a solution with matlab. I think the problem is that tos is depending on i and j as the latitude and longitude and i don't know how to correlate them properly

[Voss]

Looks like you're right. Here's using i and j:

unzip('adaptor.esgf_wps.retrieve-1652740104.757926-1767-15-d7ea9cac-5f2f-4b52-9b1f-0ba84ab0122b.zip')
i = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','i');
j = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','j');
tos = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','tos');
whos

  Name          Size              Bytes  Class     Attributes

  ans           1x117               234  char                
  cmdout        1x33                 66  char                
  i           292x1                1168  int32               
  j           362x1                1448  int32               
  tos         362x292            845632  double              

figure();
surface(j,i,tos.', ...mean(tos,3,'omitnan'), ... % tos is 2D in this case
    'FaceColor','interp', ...
    'EdgeColor','interp');
axis equal
xlim([1 362])

[MICHELE GNESOTTO]

yea thanks looks good , do you think is possible to plot coastlines in these case?

[Voss]

Here it is using latitude and longitude. It appears to have better proportions now.

[The longitude goes from 72.5 up to 360 then wraps around back to 0 and up to 73.5, but there is a little dip around 271 where it decreases for one sample, so to try to get it right, I find where there is the big decrease from 360 to 0 and add 360 to samples after that, so that from beginning to end it's ~72.5 to ~433.5. Then I label the xticks with mod(longitude,360). I'm sure you can get something better with stuff from the mapping toolbox, if you have it.]

unzip('adaptor.esgf_wps.retrieve-1652740104.757926-1767-15-d7ea9cac-5f2f-4b52-9b1f-0ba84ab0122b.zip')
longitude = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','longitude');
latitude = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','latitude');
tos = ncread('tos_Omon_CMCC-ESM2_ssp585_r1i1p1f1_gn_21000116-21000116_v20210126.nc','tos');

longitude = longitude(:,1);
disp([longitude(195:205) [NaN; diff(longitude(195:205))]]);

  267.1299       NaN
  268.0960    0.9661
  269.0615    0.9655
  270.0265    0.9650
  270.9909    0.9644
  271.9549    0.9640
  270.7897   -1.1652
  271.7916    1.0019
  272.7956    1.0040
  273.8017    1.0061
  274.8098    1.0081

idx = find(diff(longitude) < -350,1);
longitude(idx+1:end) = 360+longitude(idx+1:end);

figure();
surface(longitude(:,1),latitude(1,:),tos.', ...
    'FaceColor','interp', ...
    'EdgeColor','interp');
axis equal
xticklabels(sprintfc('%d',mod(xticks(),360)))

Regarding plotting coastlines, I'm not sure.