Filling the area between curves (loglog)

Hello,

I would like to fill the area between the two lines (on the interval 10^-2 & 10^-1):

I tried creating my own polygon using "patch" but I wondered if there is any better and more straightforward option ?

I tried using "fill" with different options but I couldn't make it work.

load('Area_glaciers_scandinavia.mat');

% Create log bins that span from 0.01 to 10000 km2
cc = 10.^(-2:0.1:4);

% Create the histogram of those bins (count nb of glacier/bin)
H = hist(area,cc);

figure()
loglog(cc,H)


idx(1) = find(roundn(cc,-4)==0.1259); %put the boundaries of the linear part of the curve)
idx(2)= find(roundn(cc,-2)==19.95);
%use polyfit
%we use log10 because we are looking for the exponents
P = polyfit(log10(cc(idx)), log10(H(idx)),1);
cf =  polyval(P,log10(cc));
hold on 
loglog(cc,10.^cf);

[KSSV]

Show us the code how you have plotted this. So that people can pick up that code and help you.

[V.D-C]

Sorry, I changed it

Answer 1

[V.D-C]

Not very elegant, but I found that the best solution was to create manually:

x = [cc(1:idx(1)) fliplr(cc(1:idx(1)-1))];
y = [10.^cf(1:idx(1)) fliplr(H(1:idx(1)-1))]; 
patch(x,y,'g')

Answer 2

[David Wilson]

I too had some difficulty to do this elgantly. I changed the name of your data file, and commented out the material that didn't work.

%% patch in loglog plots 
load('glacier_Data.mat');

%{
% Create log bins that span from 0.01 to 10000 km2
cc = 10.^(-2:0.1:4);
% Create the histogram of those bins (count nb of glacier/bin)
H = hist(area,cc);
figure()
%}

loglog(cc,H)
idx(1) = find(roundn(cc,-4)==0.1259); %put the boundaries of the linear part of the curve)
idx(2)= find(roundn(cc,-2)==19.95);
%use polyfit
%we use log10 because we are looking for the exponents
P = polyfit(log10(cc(idx)), log10(H(idx)),1);
cf =  polyval(P,log10(cc));

loglog(cc,H, ... 
       cc,10.^cf);
grid on 

Now try on a linear plot. (Just for testing)

%% Now do the shading on a linear plot
logcc = log10(cc); logH = log10(H); 
idx = find(cc<1e-1); 
xp = [logcc(idx), fliplr(logcc(idx))]; 
yp = [cf(idx), fliplr(logH(idx))]


plot(logcc, logH, ...
     logcc, cf, '--')
 
patch(xp, yp, 'y') 

Now try, as requested on a log plot. It's certainly not elegant, but does work.

%% Now try on log scale 
loglog(cc,H, ... 
       cc,10.^cf);
h=patch(10.^xp, 10.^yp, 'y', 'faceAlpha', 0.5)    
grid on 

This gives the following:

[V.D-C]

Hi, thanks for your answer but I managed to do it without changing the code as you did, and it goes to the intersection between the two plots.

You showed me the way though ! Thank you for that !