spectrogram x-axis match with data

Hello, I am plotting a spectrogram from an earthquake signal, but the time length of the signal (~250 seconds) does not match the time length displayed in the spectrogram (~100 seconds). Could you help me understand why this discrepancy occurs and how to fix it? Thank you!

[Mathieu NOE]

that is quite strange , but I cannot tell the reason as you don't show the code

to be sure to have same x axis display you can use xlim (with same values) on both subplots

[Jorge Luis]

Thank you. I describe the code here:

load fildat.mat % loading signal

Tp=10.1938; % Pwave arrival

Ts=18.5094; % Swave arrival

T_end_coda=165.55; % end Coda wave

component=’NS’;

% Parameters for the spectrogram

windowLength = 100; %Length of each segment (adjust as needed)

overlap = 80; %Overlap between segments (adjust as needed)

nfft = 100; %Number of FFT points

% fildat: is the signal

fs=200; % sample frequency

time = (0:length(fildat)-1) / fs; % Time vector in seconds

[thresh_spec,Traf, traf] = spectrogram(abs(fildat),windowLength,overlap,nfft,fs,'yaxis');

thresh_spec1 = abs(thresh_spec); % Magnitude of the spectrogram

% Convert power to decibels

P_dB = 10*log10(thresh_spec1);

fil=[0.20 0.35 0.58];

a1=[0.10 0.38 0.93];

a2=[0.91 0.1 0.1];

a3=[0.17 0.61 0.10];

a4=[0.97 0.38 0.93];

a5=[0.99 0.73 0.43];

fsz=17;

color_text=[0.09 0.27 0.68];

lg_loca='northeast';

lg_size=9;

line_width=3;

fh=figure;

% Create a tiled layout

tiledlayout(2, 1, 'TileSpacing', 'compact', 'Padding', 'compact');

% First subplot (row 1, column 1)

nexttile;

plot(time,fildat,'Color',fil,'LineWidth',1,'DisplayName',join([string(component),' - Acc Time Series'])); % Earthquake waveform

xlabel('Time (s)');

ylabel('Acc (m/s^2)');

title({name1},{[component,' - Acceleration Time Series']},'FontSize',fsz+2,'color',color_text,'FontWeight', 'bold');

xlim([0, max(time)]);

grid on;

% Mark important time points (adjust these values to match your data)

hold on;

xline(time(1), '--','LineWidth',line_width,'Color',a1,'DisplayName', 'T_{initial}'); % Initial time

xline(Tp, '--','LineWidth',line_width,'Color',a2,'DisplayName', 'T_{P-arrival}'); % Hypothetical P-wave arrival

xline(Ts, '--','LineWidth',line_width,'Color',a3,'DisplayName', 'T_{S-arrival}'); % Hypothetical S-wave arrival

xline(T_end_coda, '--','LineWidth',line_width,'Color',a4,'DisplayName', 'T_{coda-end}'); % End of coda

xline(time(end), '--','LineWidth',line_width,'Color',a5,'DisplayName', 'T_{end}'); % End of the signal 'Z_{TOR}=50km

legend('Location',lg_loca,'color',[0.97 0.96 0.96],'TextColor', [0 0.35 0.74],'EdgeColor',[0 0.35 0.74],'FontSize',lg_size,'LineWidth',1, 'NumColumns', 1)

grid on

set(gca, 'FontName', 'Times New Roman', 'FontSize', fsz-3, 'FontWeight', 'bold', ...

'XColor', color_text, 'YColor', color_text, 'XTickLabelRotation', 0, ...

'LineWidth', 0.25, 'GridColor', [0.5, 0.5, 0.5], 'GridAlpha', 0.5);

% Plot the spectrogram

nexttile; % Bottom plot for spectrogram

imagesc(Traf, traf, P_dB); % Display spectrogram

axis xy; % Flip Y-axis

xlabel('Time (s)');

ylabel('Frequency (Hz)');

title({name2},{[component,' - Spectrogram']},'FontSize',fsz+2,'color',color_text,'FontWeight', 'bold');

colormap(jet); % Set colormap

colorbar; % Display color scale

set(gca, 'FontName', 'Times New Roman', 'FontSize', fsz-3, 'FontWeight', 'bold', ...

'XColor', color_text, 'YColor', color_text, 'XTickLabelRotation', 0, ...

'LineWidth', 0.25, 'GridColor', [0.5, 0.5, 0.5], 'GridAlpha', 0.5);

%clim([100, -100]); % Adjust dynamic range for better visualization

% Match x-axis of the spectrogram to the waveform time axis

%xlim([0, max(time)]); % Set x-axis limits to match the time vector

fh.WindowState = 'maximized';

Answer 1

[Mathieu NOE]

hello again

so the problem was simply that you inverted time and frequency vectors when calling imagesc !

I guess the mistake is due to the fact that when computing the spectrogram data, you named the frequency vector Traf and the time vector traf , so that can be misleading.

I opted to change Traf into Fraf which sounds more like a frequency vector and not a time vector.

other minor modifications :

you can remove this line : axis xy; % Flip Y-axis as a few lines latter you are doing :

set(gca, 'FontName', 'Times New Roman', 'FontSize', fsz-3, 'FontWeight', 'bold', ..

so simply add : ,'Ydir','normal', and you're done with y axis flip

now I think your code is working fine

name1 = 'Test'; % just for the demo here 
name2 = 'Test'; % just for the demo here

load fildat.mat % loading signal
Tp=10.1938; % Pwave arrival
Ts=18.5094; % Swave arrival
T_end_coda=165.55; % end Coda wave
component='NS';
 
% Parameters for the spectrogram
windowLength = 100; %Length of each segment (adjust as needed)
overlap = 80;      %Overlap between segments (adjust as needed)
nfft = 100;         %Number of FFT points
% fildat: is the signal
fs=200; % sample frequency
time = (0:length(fildat)-1) / fs; % Time vector in seconds
[thresh_spec,Fraf, traf] = spectrogram(abs(fildat),windowLength,overlap,nfft,fs,'yaxis');
 %   [S,F,T] = SPECTROGRAM(...) returns a vector of frequencies, F, and a
%   vector of times, T, at which the spectrogram is computed

thresh_spec1 = abs(thresh_spec); % Magnitude of the spectrogram
% Convert power to decibels
P_dB = 10*log10(thresh_spec1);
fil=[0.20 0.35 0.58];
a1=[0.10 0.38 0.93];
a2=[0.91 0.1 0.1];
a3=[0.17 0.61 0.10];
a4=[0.97 0.38 0.93];
a5=[0.99 0.73 0.43];
 
fsz=17;
    color_text=[0.09 0.27 0.68];
    lg_loca='northeast';
    lg_size=9;
    line_width=3;
    fh=figure;
    % Create a tiled layout
    tiledlayout(2, 1, 'TileSpacing', 'compact', 'Padding', 'compact');
    % First subplot (row 1, column 1)
    nexttile;
    plot(time,fildat,'Color',fil,'LineWidth',1,'DisplayName',join([string(component),' - Acc Time Series'])); % Earthquake waveform
    xlabel('Time (s)');
    ylabel('Acc (m/s^2)');
    title({name1},{[component,' - Acceleration Time Series']},'FontSize',fsz+2,'color',color_text,'FontWeight', 'bold');
    xlim([0, max(time)]);
    grid on;    
 
    % Mark important time points (adjust these values to match your data)
    hold on;
    xline(time(1), '--','LineWidth',line_width,'Color',a1,'DisplayName', 'T_{initial}');  % Initial time
    xline(Tp, '--','LineWidth',line_width,'Color',a2,'DisplayName', 'T_{P-arrival}');    % Hypothetical P-wave arrival
    xline(Ts, '--','LineWidth',line_width,'Color',a3,'DisplayName', 'T_{S-arrival}');    % Hypothetical S-wave arrival
    xline(T_end_coda, '--','LineWidth',line_width,'Color',a4,'DisplayName', 'T_{coda-end}');    % End of coda
    xline(time(end), '--','LineWidth',line_width,'Color',a5,'DisplayName', 'T_{end}'); % End of the signal  'Z_{TOR}=50km
    legend('Location',lg_loca,'color',[0.97 0.96 0.96],'TextColor', [0 0.35 0.74],'EdgeColor',[0 0.35 0.74],'FontSize',lg_size,'LineWidth',1, 'NumColumns', 1)    
    grid on
    set(gca, 'FontName', 'Times New Roman', 'FontSize', fsz-3, 'FontWeight', 'bold', ...
        'XColor', color_text, 'YColor', color_text, 'XTickLabelRotation', 0, ...
        'LineWidth', 0.25, 'GridColor', [0.5, 0.5, 0.5], 'GridAlpha', 0.5);
 
    % Plot the spectrogram
    nexttile; % Bottom plot for spectrogram
    imagesc(traf, Fraf, P_dB); % Display spectrogram
    xlabel('Time (s)');
    ylabel('Frequency (Hz)');
    title({name2},{[component,' - Spectrogram']},'FontSize',fsz+2,'color',color_text,'FontWeight', 'bold');
    colormap(jet); % Set colormap
    colorbar; % Display color scale
    set(gca,'Ydir','normal', 'FontName', 'Times New Roman', 'FontSize', fsz-3, 'FontWeight', 'bold', ...
        'XColor', color_text, 'YColor', color_text, 'XTickLabelRotation', 0, ...
        'LineWidth', 0.25, 'GridColor', [0.5, 0.5, 0.5], 'GridAlpha', 0.5);
    %clim([100, -100]); % Adjust dynamic range for better visualization
     % Match x-axis of the spectrogram to the waveform time axis
     xlim([0, max(time)]); % Set x-axis limits to match the time vector
     fh.WindowState = 'maximized';

●

[Jorge Luis]

Thank you very much.

[Mathieu NOE]

as always, my pleasure !

Answer 2

[Mathieu NOE]

this can also helps you :

%%Method 1: equate axis sizes for all axes after adding colorbar
% Size of axes without a colorbar will be set to the size of axes with a colorbar.
% Create Demo
ax = gobjects(2,3); 
for i = 1:5
    ax(i) = subplot(3,2,i);
    imagesc(ax(i),rand(100,1000))
    colorbar(ax(i))
end
ax(end) = subplot(3,2,6);
plot(ax(end), 1:1000,(1:1000).*rand(1,1000)*2)
% Set the width and height of all axes to the min width & height
allAxPos = vertcat(ax.Position);
allAxPos(:,3:4) = min(allAxPos(:,3:4)).*ones(numel(ax),1);
set(ax,{'position'},mat2cell(allAxPos,ones(numel(ax),1),4))

%%Method 2: Restore axis position and set colobar positions
% The axes maintain their original sizes prior to adding colorbar.
%  The colorbar position and width is adjusted.

function addOutsideColorbar(ax)
pos = ax.Position; 
cb = colorbar(ax,'EastOutside'); 
ax.Position = pos; 
cb.Position(1) = sum(ax.Position([1,3]))+.01;
cb.Position(3) = cb.Position(3).*.5;
end

ax = gobjects(2,3); 
for i = 1:5
    ax(i) = subplot(3,2,i);
    imagesc(ax(i),rand(100,1000))
    addOutsideColorbar(ax(i))
end

ax(end) = subplot(3,2,6);
plot(ax(end), 1:1000,(1:1000).*rand(1,1000)*2)
%%Method 3: Use tiledlayout
ax = gobjects(2,3); 
tlo = tiledlayout(3,2);
for i = 1:5
    ax(i) = nexttile(tlo);
    imagesc(ax(i),rand(100,1000))
    colorbar(ax(i))
end
ax(end) = nexttile(tlo);
plot(ax(end), 1:1000,(1:1000).*rand(1,1000)*2)