This can certainly be done, however it is definitely not trivial!
Try this —
LD = load('Miguel Albuquerque azimuth3.mat');
[M,C] = contourf(azimuth3,25, 'EdgeColor','none');
Levels = C.LevelList
Levels =
1.0e+06 *
0.0001 0.0657 0.1314 0.1970 0.2626 0.3282 0.3939 0.4595 0.5251 0.5908 0.6564 0.7220 0.7876 0.8533 0.9189 0.9845 1.0501 1.1158 1.1814 1.2470 1.3127 1.3783 1.4439 1.5095 1.5752 1.6408
idx = find(M(1,:) == Levels(k));
ValidV = rem(M(2,idx),1) == 0;
StartIdx{k,:} = idx(ValidV);
VLen{k,:} = M(2,StartIdx{k});
title('Original Filled Contour Plot')
[HiLvls,idx] = maxk(Levels,9);
cm = turbo(numel(Levels));
idxv = StartIdx{k1}(k2)+1 : StartIdx{k1}(k2)+VLen{k1}(k2);
patch(xv, yv, cm(k1,:), 'EdgeColor','none')
title('Highest Level Contours')
Make appropriate changes to work as you want it to with your data.
EDIT — (26 Jul 2022 at 21:04)
Increased the number of contours to be plotted in the second plot.
.