Perhaps this —
opts = detectImportOptions('tempDataTrollhFlygpl.xlsx', 'VariableNamingRule','preserve', 'HeaderLines',9);
opts = setvartype(opts, 4, 'double');
D=readtable('tempDataTrollhFlygpl.xlsx', opts);
M = D(:,[3 4])
Lv = month(M{:,1}) == 7;
M7 = M(Lv,:)
D = unique(day(M7{:,1}));
Y = year(M7{:,1});
Yu = unique(Y);
Yidx = Y-Y(1)+1;
Year_7c = accumarray(Yidx, (1:numel(Yidx)).', [], @(x){M7{x,2}}); % July For Each Year As Separate Cell Element
Year_7m = cat(2,Year_7c{:}); % July Matrix (Columns = Years Corresponding To The 'Y' Vector)
A = mean(cat(2,Year_7c{:}),2);
Asem = std(cat(2,Year_7c{:}),[],2)/sqrt(numel(Year_7c));
cv = tinv([0.025 0.975], numel(Year_7c)-1);
hold on
for k = 1:numel(Year_7c)
hp(k) = plot(D,Year_7c{k}, 'DisplayName',string(Yu(k)));
end
hpm = plot(D, A, '-k', 'LineWidth',2, 'DisplayName','Mean');
hpci = plot(D,Asem*cv+A, '--k', 'LineWidth',2, 'DisplayName','95% CI');
hold off
xlabel('Date');
ylabel('Temperature');
title('Juli')
legend([hp hpm hpci(1)], 'Location','eastoutside')
% A = mean(M, 4)
% plot((M.Var3),(M.Var4))
The code first identifies the July entries and the corresponding years. It then uses accumarray to aggregate the days in every July as elements of a cell array, converts them to a numeric matrix, and then plots them as a function of the days. It also calculates the daily mean, and 95% confidence intervals based on the critical values of the t-distribution.
EDIT — Corrected typographical errors.
.
