Spatial-Temporal Contour plot from multi excel files

5 views (last 30 days)
Hello there,
I have four groups of datasets (in excel files, attached). Let’s say they are: data_A1, data_A2, data_A3, data_A4. Each of them contains 3 parameters, namely ax, ay, and az. To make it easily imagined in space-time perspective, let’s say ax is the variable of time and ay is the variable of depth and az is the variable I'd like to plot its spatial-temporal variability. My intention is to get a contour plot of those datasets, representing the spatial-temporal evolution of variable az. So, the contour will plot data ax in x-axis, ay in y-axis, and az value presented by the colormap.
Note that:
  • There may be a possibility of overlapping/similar values of ay (depth) between data_A1, data_A2, data_A3, data_A4. In this case, I just need the first record to be considered; or maybe if any other proper handling you could suggest, please welcome.
  • There also may be blank values of az.
Thanks!

Accepted Answer

Star Strider
Star Strider on 18 Oct 2024
I am not certain what you want.
As a first approach, this presents them as surfaces —
files = dir('*.xlsx');
for k = 1:numel(files)
filename = files(k).name
T{k} = readtable(filename);
T{k}
[Tstart,Tend] = bounds(T{k}.ax)
dn{k} = datenum(T{k}.ax);
DN{k} = linspace(min(dn{k}), max(dn{k}), numel(dn{k}));
AY{k} = linspace(min(T{k}.ay), max(T{k}.ay), numel(T{k}.ay))
end
filename = 'data_A1.xlsx'
ans = 619x3 table
ay az ax ______ ______ ____________________ 738.4 6.103 12-Feb-2000 16:50:00 737.94 6.1639 12-Feb-2000 17:00:00 737.97 6.2012 12-Feb-2000 17:10:00 738.03 6.2016 12-Feb-2000 17:19:59 737.97 6.2027 12-Feb-2000 17:30:00 737.91 6.2093 12-Feb-2000 17:39:59 737.98 6.2048 12-Feb-2000 17:50:00 738 6.179 12-Feb-2000 18:00:00 738 6.1685 12-Feb-2000 18:10:00 737.91 6.1394 12-Feb-2000 18:20:00 737.93 6.1043 12-Feb-2000 18:29:59 737.86 6.0869 12-Feb-2000 18:40:00 737.81 6.0737 12-Feb-2000 18:49:59 737.83 6.0665 12-Feb-2000 19:00:00 737.77 6.0663 12-Feb-2000 19:09:59 737.79 6.0705 12-Feb-2000 19:20:00
Tstart = datetime
12-Feb-2000 16:50:00
Tend = datetime
16-Feb-2000 23:50:00
AY = 1x1 cell array
{[736.2120 736.2302 736.2484 736.2666 736.2847 736.3029 736.3211 736.3393 736.3575 736.3757 736.3938 736.4120 736.4302 736.4484 736.4666 736.4848 736.5030 ... ] (1x619 double)}
filename = 'data_A2.xlsx'
ans = 616x3 table
ay az ax ______ ______ ____________________ 998.8 4.7603 12-Feb-2000 16:40:00 997.14 4.7809 12-Feb-2000 16:50:00 997.08 4.8233 12-Feb-2000 16:59:59 997.15 4.828 12-Feb-2000 17:10:00 997.24 4.8172 12-Feb-2000 17:19:59 997.26 4.8216 12-Feb-2000 17:30:00 997.26 4.8499 12-Feb-2000 17:39:59 997.22 4.8599 12-Feb-2000 17:50:00 997.26 4.8561 12-Feb-2000 18:00:00 997.26 4.8569 12-Feb-2000 18:10:00 997.26 4.8593 12-Feb-2000 18:20:00 997.22 4.8634 12-Feb-2000 18:29:59 997.13 4.89 12-Feb-2000 18:40:00 997.13 4.8933 12-Feb-2000 18:49:59 997.14 4.9276 12-Feb-2000 19:00:00 997.07 4.9166 12-Feb-2000 19:09:59
Tstart = datetime
12-Feb-2000 16:40:00
Tend = datetime
16-Feb-2000 23:10:00
AY = 1x2 cell array
{1x619 double} {1x616 double}
filename = 'data_A3.xlsx'
ans = 634x3 table
ay az ax ______ ______ ____________________ 1407.9 3.2265 12-Feb-2000 16:40:00 1408.4 3.2062 12-Feb-2000 16:50:00 1408.8 3.1958 12-Feb-2000 16:59:59 1409 3.1159 12-Feb-2000 17:10:00 1409.1 3.0674 12-Feb-2000 17:19:59 1409.1 3.1072 12-Feb-2000 17:30:00 1409.2 3.1402 12-Feb-2000 17:39:59 1409.2 3.1925 12-Feb-2000 17:50:00 1409.2 3.093 12-Feb-2000 18:00:00 1409.2 3.1666 12-Feb-2000 18:10:00 1409.2 3.1888 12-Feb-2000 18:20:00 1409.1 3.1839 12-Feb-2000 18:29:59 1409.1 3.1962 12-Feb-2000 18:40:00 1409.1 3.2176 12-Feb-2000 18:49:59 1409.1 3.1954 12-Feb-2000 19:00:00 1409.1 3.2809 12-Feb-2000 19:09:59
Tstart = datetime
12-Feb-2000 16:40:00
Tend = datetime
17-Feb-2000 02:10:00
AY = 1x3 cell array
{1x619 double} {1x616 double} {1x634 double}
filename = 'data_A4.xlsx'
ans = 817x3 table
ay az ax ______ ______ ____________________ 1808.7 2.4468 12-Feb-2000 16:40:00 1808.7 2.4444 12-Feb-2000 16:50:00 1809.2 2.438 12-Feb-2000 16:59:59 1809.5 2.4302 12-Feb-2000 17:10:00 1809.7 2.4162 12-Feb-2000 17:19:59 1809.7 2.4082 12-Feb-2000 17:30:00 1809.8 2.3998 12-Feb-2000 17:39:59 1809.8 2.4149 12-Feb-2000 17:50:00 1809.8 2.4166 12-Feb-2000 18:00:00 1809.9 2.422 12-Feb-2000 18:10:00 1809.9 2.432 12-Feb-2000 18:20:00 1809.8 2.4315 12-Feb-2000 18:29:59 1809.8 2.448 12-Feb-2000 18:40:00 1809.7 2.4747 12-Feb-2000 18:49:59 1809.7 2.4836 12-Feb-2000 19:00:00 1809.8 2.4857 12-Feb-2000 19:09:59
Tstart = datetime
12-Feb-2000 16:40:00
Tend = datetime
18-Feb-2000 08:40:00
AY = 1x4 cell array
{1x619 double} {1x616 double} {1x634 double} {1x817 double}
for k = 1:numel(files)
Fcn{k} = scatteredInterpolant(dn{k}, T{k}.ay, T{k}.az);
[AX{k},AY{k}] = ndgrid(DN{k},AY{k});
AZ{k} = Fcn{k}(AX{k},AY{k});
end
% figure
% hold on
for k = 1:numel(files)
figure
surfc(AX{k}, AY{k}, AZ{k}, 'EdgeColor','none')
colormap(turbo)
colorbar
xlabel('ax')
ylabel('ay')
zlabel('az')
title(extractBetween(files(k).name,'_','.'))
end
% hold off
It is necessary to use datenum here because scatteredInterpolant does not work with datetime arrays.
If plotted on the same axes, these appear as flat ribbons with respect to ‘az’, and lose their depth (at least in a relative sense). You can of course plot them as contour plots, however I am at a loss as to how to present them in a ‘spatio-temporal perspective’. Animating them will not show here, and it would be difficult to interpolate them over a smooth time scale in any event, in part because their sizes are not the same (although that could be standardised). The other option is to subtract them serially.
.
  8 Comments

Sign in to comment.

More Answers (0)

Categories

Find more on Migrate GUIDE Apps in Help Center and File Exchange

Products


Release

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!