Controu plot from matlab table

I have a table with values extracted from a csv I want to make a contour plot from.

Let's use this table as an example

tdata.x = [1;2;1;2];
tdata.y = [3;3;4;4];
tdata.z = randn(4,1);
tdata=struct2table(tdata);
>> tdata

tdata =

  4×3 table

    x    y       z   
    _    _    _______

    1    3    0.53767
    2    3     1.8339
    1    4    -2.2588
    2    4    0.86217

I would like to pivot this such that I can use it for plotting a contour, so in principle I want a 2x2 z matrix where rows/columns are given by y and x respectively, something in this direction:

 x 1        2
y                                                                           
3  0.53767  1.8339
4  -2.2588  0.86217

Answer 1

[KSSV]

x = T.x ; 
y = T.y ;
z = T.z ;
nx = length(unique(x)) ; 
ny = length(unique(y)) ; 
X = reshape(x,nx,ny) ;
Y = reshape(y,nx,ny) ;
Z = reshape(z,nx,ny) ;
contour(X,Y,Z)

[Morten Nissov]

I am running into some problems with this solution, for example applying this to the attached dataset and the following code results in an incorrect plot which contains a lot of artifacts.

file = 'EMAG2_amsterdam.csv';
map = readtable(file);

lon = map.lon;
lat = map.lat;
z = map.upCont;

nx = length(unique(lon));
ny = length(unique(lat));

X = reshape(lon, nx, ny);
Y = reshape(lat, nx, ny);
Z = reshape(z, nx, ny);

figure(1)
contourf(X,Y,Z)
xlabel('longitude')
ylabel('latitude')

The artifacts are even more apparent when looking at the surf instead of the contourf.

[KSSV]

file = 'EMAG2_amsterdam.csv';
map = readtable(file);

lon = map.lon;
lat = map.lat;
z = map.upCont;

m = 100 ; n = 100 ; 
x = linspace(min(lon),max(lon),m) ; 
y = linspace(min(lat),max(lat),n) ; 
[X,Y] = meshgrid(x,y) ;
Z = griddata(lon,lat,z,X,Y) ;
figure(1)
contourf(X,Y,Z)
xlabel('longitude')
ylabel('latitude')

[Morten Nissov]

What about the original version was problematic?

[KSSV]

Yes..use the present given code.

[Morten Nissov]

I was asking what was wrong with the original implementation?

I am using it for much more than just this single visualization so I was curious if there was something that I will need to keep in mind. For this reason as well I'd like to avoid resampling/interpolating for this visualization.

The confusion for me is that, according to documentation, griddata is for scattered data. But the dataset is already gridded, at least as far as I'm aware.

[KSSV]

That should work...but the data arrangment should be considered.

[Morten Nissov]

Is there another way to transform a table with 3 columns to a grid? It can be done in pandas with pivot, which is how I know the data itself if fine.

[KSSV]

The gird looks fine when you use

scatter(X(:),Y(:),[],Z(:),'.')

Show me the PAndas code which worked for you.

[Morten Nissov]

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_csv('EMAG2_faaborg.csv')
df = df[['lat','lon','upCont']].pivot('lat','lon')

lon = df.columns.levels[1].values
lat = df.index.values
vals = df.values

X,Y = np.meshgrid(lon, lat)

then I can plot using

plt.contourf(X, Y, vals)