How to use fill or patch when some vertices are NaN?

Hi All,

I have a structure of water body outlines read in from a shapefile defining 1000+ water bodies. Each element of the structure contains (usually) one but sometimes more water bodies with the last row of each body being NaN. Both Patch and Fill plot only the outlines and won't put in the color because of the nan. For some of them I can just patch/fill values from 1:end-1 and that works, but when there are two water bodies with a NaN in the middle. How can I patch/fill these water bodies on a regular plot axes (not a map axes)?

water(1)
ans = 
       Geometry: 'Polygon'
    BoundingBox: [2x2 double]
              X: [3.0289e+05 3.0294e+05 3.0299e+05 3.03e+05 3.0298e+05 3.0292e+05 3.0292e+05 3.029e+05 3.0289e+05 NaN]
              Y: [4.2084e+06 4.2084e+06 4.2084e+06 4.2084e+06 4.2083e+06 4.2083e+06 4.2084e+06 4.2084e+06 4.2084e+06 NaN]
fill(water(i).Lon(1:end-1), water(i).Lat(1:end-1),[0.5 0.5 1],'FaceColor',[0.5 0.5 1]) % Works and plots with color
fill(water(i).Lon, water(i).Lat,[0.5 0.5 1],'FaceColor',[0.5 0.5 1]) % Plots only the outline

water(8)
ans = 
       Geometry: 'Polygon'
    BoundingBox: [2x2 double]
              X: [1x89 double]
              Y: [1x89 double]
water(8).X
ans =
  Columns 1 through 10
   3.4428e+05   3.4427e+05   3.4427e+05   3.4429e+05   3.4427e+05   3.4425e+05   3.4424e+05   3.4424e+05   3.4425e+05   3.4425e+05
  Columns 11 though 20
   3.4428e+05          NaN   3.4419e+05   3.4419e+05    3.442e+05    3.442e+05    3.442e+05   3.4418e+05   3.4416e+05   3.4415e+05
  Columns 21 through 30
   3.4413e+05    3.441e+05   3.4408e+05   3.4407e+05   3.4409e+05    3.441e+05   3.4412e+05   3.4413e+05   3.4415e+05   3.4417e+05
  Columns 31 through 40 
   3.4419e+05          NaN   % Etc etc.
i = 8; fill(water(i).Lon(1:end-1), water(i).Lat(1:end-1),[0.5 0.5 1],'FaceColor',[0.5 0.5 1])  % Fails because of NaNs in the vectors

[Vaidyanathan Thiagarajan]

Hello Emily,

You can remove the NaNs and then use 'fill' as shown in the following code snippet :

XY = [0 0; 1 0; 1 1; NaN NaN;0.5 0.25];

X = XY(:,1);
Y = XY(:,2);

%remove all NaN's from both X and Y
X = X(~isnan(X));
Y = Y(~isnan(Y));

fill(X,Y,[0.5 0.5 1],'FaceColor',[0.5 0.5 1])

The above code assumes that there is a one-to-one mapping of NaNs in X and Y. If that is not the case, you might have to suitably modify the code. For more details on 'isnan' and not (~) function, kindly refer to the following links :

https://www.mathworks.com/help/matlab/ref/isnan.html

https://www.mathworks.com/help/matlab/ref/not.html

However, it is best to give some thought about why you are getting NaNs in the first place. It suggests a possible problem in the underlying algorithm. Hence, it would be best to address this issue first rather than explicitly removing NaNs (as in the above code).

Hope this helps.

Vaidyanathan

[the cyclist]

You could do

keepIndex = ~isnan(X) & ~isnan(Y);

X = X(keepIndex);
Y = Y(keepIndex);

if the NaNs do not always appear in both X and Y.