w=warning("off",'MATLAB:table:ModifiedAndSavedVarnames'); % turn of nag warning on variable naming
tWD=readtable('weather_data_01_11_25.csv'); % use generic variable names
warning(w)
head(tWD) % see what's in it since we don't know
tWD.Properties.VariableNames(2:4)={'WindSpeed','WindDirection','Temperature'}; % more convenient names
tWD=addvars(tWD,deg2rad(tWD.WindDirection),'After','WindDirection','NewVariableNames','WindRadians');
figure
hPax=polaraxes('ThetaDir','clockwise','ThetaZeroLocation','top');
hold(hPax,'on')
ALPHA=0.05; % face alpha for background -- set as result pleases
BKGD='gcbm'; % character string for background color cycle -- adapt to suit...
edges=[0 1/2]; % initial 90 degree wedge
MAX=40; % arbitrary max from initial windrose--have to fix to match other data
for i=1:numel(BKGD) % iterate over colors
polarhistogram('BinEdges',pi*edges,'BinCounts',MAX,'FaceColor',BKGD(i),'FaceAlpha',ALPHA,'EdgeColor','none','displayname','')
edges=edges+(1/2); % next quadrant
end
edges=deg2rad(0:10:360); % reset for real thing
FACES={'#A2142F','#D95319','#EDB120','#0072BD','#4DBEEE'};
BINS=[15 10 8 4 2 0];
for i=1:numel(BINS)-1
LABEL=sprintf('%d - %d m/s',BINS(i),BINS(i+1));
polarhistogram(tWD.WindRadians(tWD.WindSpeed<BINS(i)),edges,'FaceColor',FACES{i},'displayname',LABEL)
end
Try something like the above for starters -- you have to draw the background as more sections because the axes color is only a single property.
The scaling for the background will have to be set dynamically from the data; I just looked at the original figure to get the value "40" above.
I also took the opportunity to make the code more generic so it can be applied to any similar dataset without the baggage of the date -- that's in the dataset; don't carry metadata around with the filenames.
I the also showed how to make more convenient names in the table to suit and made the second dataset of the wind direction in radians to only do the conversion once.
Another nicety that could be done would be to set the bin magnitudes in a data array along with their associated color values and then iterate over the number of bins in a loop rather than duplicating the same line of code over and over. Generically, this might mean adding another function that set bin sizes/magnitudes to be able to handle any data set with different ranges.
And, of course, finish up the original task by picking another pair of colors to draw in the backgrounds...there again, a loop construct would be good way to implement.
ADDENDUM:
Oh! Forgot to add -- you'll want to save an array of handles to the histogram objects so that when you add the legend (another place where the string could/should be generated dynamically from the binning values instead of having hardcoded) you can pass only the actual data handles; otherwise it's going to have extra labels for the background areas; the 'Annotation' property is read-only so even if the 'DisplayName' is blank, you'll still get the legen colored box, it just won't have a label associated with it. Pass the array of wanted legend objects to legend only.
