Double checking a for loop

Can someone just double check my for loop. Specifically the bolded line. I wanted to calculate the integrated chlorophyll and then use it to find the relative amount by dividing temp by the intchl. Technically speaking, if I add up all my relative chlorophyll values for each column they should equal 1. However, they are coming out totalling a little over one so I think I messed up.

for i=1:length(databin)
    temp=databin{i};
    temp=temp(:,Varnum);
    VARfield(1:length(temp),i)=temp;
 
  intchl(i) = nansum(0.5.*(temp(1:end -1) + temp(2:end)));         %MARKED
  %integrated chlorophyll should just be one number for each cast
 relchl = temp/intchl(i);
  VARfieldrel(1:length(relchl),i)=relchl;
end 

Answer 1

[Sergey Kasyanov]

Hello,

sum(temp/intchl) -> sum(temp)/intchl means that you divide sum of all values by sum of mean values. Lets write it in detail:

sum(temp) = x1 + x2 + x3 + ... + x(n-1) + x(n)

intchl = (x1+x2)/2 + (x2 + x3)/2 + ... + (x(n-2) + x(n-1)/2 + (x(n-1) + xn)/2 = x1/2 + x2 + x3 + .... + x(n-1) + xn/2 = sum(temp) - (temp(1) + temp(end))/2

It means that intchl is lower than sum(temp).

I think you should to calculate intchl as sum(temp).

[Sergey Kasyanov]

Do you need to integrate temp over the bincenters?

In that case you can use next code:

intchl(i) = nansum((temp(1:end-1) + temp(2:end))/2.*diff(bincenters'))

[Jacqueline Chrabot]

actually I think I do. because the data is binned by the depth which is the bincenters. That might work!

[Jacqueline Chrabot]

Sergey, my bincenters is 1 by 20 .. I still get the error, "matrix dimensions don't agree".

[Jacqueline Chrabot]

VARfield=NaN(totalrow(1),length(databin));

Varnum=11; %chloro

for i=1:length(databin)

temp=databin{i};

temp=temp(:,Varnum);

VARfield(1:length(temp),i)=temp;

%intchl(i) = nansum((temp(1:end-1) + temp(2:end)))

intchl(i) = nansum((temp(1:end-1) + temp(2:end))/2.*diff(bincenters'));

%integrated chlorophyll should just be one number for each cast

relchl = temp/intchl(i);

VARfieldrel(1:length(relchl),i)=relchl;

end

Sergey I'm stuck here and bolded line still gives me the error 'Matrix dimensions don't agree'

[Sergey Kasyanov]

I cant understand why error still exist. Maybe you should to use diff(bincenters) instead of diff(bincenters')?

[Jacqueline Chrabot]

databin= cell(length(datafi),1)

for i = 1:length(datafi);

%for i =1;

temp=datafi{i};

minval = 0;

maxval = max(temp.Var7);

step = 0.25; %increase to 25, was at 0.10

bins = minval:step:maxval;

bincenters = bins + step/2;

bincenters = bincenters(1:length(bincenters)-1);

DATAavg_input = ones(length(bincenters),size(temp,2)).*NaN;

for xx = 2:length(bins);

nn = find(temp.Var7 >= bins(xx-1) & temp.Var7 < bins(xx));

DATAavg_input(xx-1,3:end) = nanmean(table2array(temp(nn,3:end)),1);

end

DATAavg_input(:,1)= datenum(datetime(temp.Var1(1), 'InputFormat', 'MM/dd/yyyy', 'Format', 'MM-dd-yyyy hh:mm:ss.SSSS') + temp.Var2(1));

%make sure to convert it back for contour using datetime

databin{i}=DATAavg_input;

end

DataBin = DATAavg_input;

Sergey, for the time being do you know how to convert back the time from the bolded line?? I tried using xticks later on in my script and realized all the date and times were the same so xtick was saying 00:00 for everything??