I cannot plot discontinuous X data.

I have the data as shown below.

1.36667  105.403

1.375    105.6288

1.38333  105.8527

1.39167  106.049

11.7  88.2319

11.70833  88.1568

11.71667  88.2712

11.725  88.3003

21.26667  999

21.275  92.3620

21.28333   92.2410

21.29167  92.3133

The x axis expresses time and the y axis shows the observed data.

I try to use the code below to plot dicontinuous data.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

data = xlsread('G07.xls');

time = data(:,1);

tec  = data(:,3);

max_step = 31/3600;

segends = [find(diff(time) > max_step) length(time)]; % assume x non-decreasing
nsegs = length(segends);

plotargs = cell(3, nsegs);

segend = 0;

for segment = 1:nsegs

      segstart = segend+1;

      segend = segends(segment);

      plotargs{1,segment} = time(segstart:segend);

      plotargs{2,segment} = tec(segstart:segend);

      plotargs{3,segment} = plotstyle;

end

plot(plotargs{:});

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

This code works well only one gap data. But the data above has 2 gaps so there is an error message below.

??? Error using ==> horzcat
CAT arguments dimensions are not consistent.

Error in ==> plot_g at 7
segends = [find(diff(time) > max_step) length(time)]; % assume x non-decreasing

I don't know how to fix this problem and how to plot multiple gaps data.

Aike...

Answer 1

[Walter Roberson]

An easier way to do this, is that at each point you want the break, insert a NaN in both the x and y data. Then use a single plot() call instead of a loop.

Answer 2

[Aike]

Yes, Walter Roberson, However, I have around 11,680 data so I cannot insert NaN to every files one by one.

[Image Analyst]

What identifies a "gap"?

And what does 11,680 data points have to do with anything? As I read it you have only 1 file and the data array from that file has 11,680 rows. This is a very small amount of data and you should not worry about the size of it at all. You may think it's large, but trust me - it's not. Far from it.

[Walter Roberson]

I didn't say you should insert the NaN manually. You can insert the NaN by way of program. Doing so is easier than maintaining the segments.

Answer 3

[Aike]

Sorry, I didn't make it clear. The data above comes from parts of the GPS data from one GPS satellite (G07). In fact there are 32 GPS satellites for one day. That means I have 32 files for each day. For all data, I have to analyze one year data so I have 32*365 = 11,680 files for one GPS receiver in one year. Actually, there are around 1000 GPS receivers from GEONET over Japan so I cannot insert NaN to 11,680*1000 GPS files.

For the data gap, I mean the discontinuous x data which expresses Time in my data. The x data below was shown in universal time (UT).

1.36667 1.375 1.38333 1.39167

11.7 11.70833 11.71667 11.725

21.26667 21.275 21.28333 21.29167

The data is discontinuous, 1 UT --> 11 UT --> 21 UT The code on M.file above can plot only 1 UT --> 11 UT. In the case of 1 UT --> 11 UT --> 21 UT, there is an error with the program. Thus, I need some help to plot whole data.

Thank you very much for all kindly reply messages.

Regards, Aike...

[Walter Roberson]

Start with a data array. Scan backwards from the end. At each point determine whether a gap exists. If it does, insert a NaN at that point, such as

X(K:end+1) = [NaN; X(K:end)];
Y(K:end+1) = [NaN; Y(K:end)];

Keep scanning backwards until you get to the beginning of the array. Then

plot(X, Y);

Just the one plot().

Notice you have no segment arrays, and you plotted it in one statement instead of having to loop doing multiple plots for the same data set. Notice the low probability of dimension problems. Notice too that you had no need of modifying the data files.

After you have tested this out, you can consider even more efficient ways of inserting the NaN if you feel that is worth while.