how can I read multiple netcdf file using ncread in matlab?

I have 6000 files netcdf files.In each file there are four variables like latitude, longitude,precipitation and time.

I made a variable

filename=dir(location of folder in which files are)

now I want to read data from each 6000 files for every variable I write

for j=3:length(filename)
a=ncread('\folder',filename(j).name,variable);
end

but files are not being read. can u tell me how can I read these files?

the format of file is 3B42_Daily.20020315.7.nc4

[KSSV]

Are Looking the variable names in all the files same?

[Deep Shah]

yes

Answer 1

[Walter Roberson]

ncvars =  {'latitude', 'longitude', 'precipitation', 'time'};

projectdir = 'C:\some\location';
dinfo = dir( fullfile(projectdir, '*.nc4') );
num_files = length(dinfo);
filenames = fullfile( projectdir, {dinfo.name} );

lats = cell(num_files, 1);
lons = cell{num_files, 1);
precips = cell(num_files, 1);
times = cell(num_files, 1);

for K = 1 : num_files
  this_file = filenames{K};
  lats{K} = ncread(this_file, ncvars{1});
  lons{K} = ncread(this_file, ncvars{2});
  precips{K} = ncread(this_file, ncvars{3});
  times{K} = ncread(this_file, ncvars{4});
end

[Farshid Daryabor]

thanks

[Michelle De Luna]

Hi @Walter Roberson! I know this is a long shot, but I was hoping to see if you could help me adapt this code to a situation where each of my .nc files has a 4x1 column vector. That is, there are four timesteps in each of the .nc files, and I am trying to combine all of the corresponding data values from all of my .nc files into one large file that I can work with in the future.

[Walter Roberson]

Your .nc file has only one variable of interest, and it is always a 4 x 1 result in each file?

How are your files named? In particular if they are numbered files, then we need to know if the numbers have leading zeros -- are the files data1.nc data2.nc ... data10.nc data11.nc ..., etc., or are they data001.nc, data002.nc, data003.nc, ... data010.nc, data011.nc, and so on ? If there are no leading zeros but they are numbered, then we have to compensate a bit to arrange them in the correct order.

[Benjamin Osborne]

Hi Walter, how would one adapt this code to enable it to plot the 'lats' 'lons' and 'precips' data as individual subplots in a figure using the pcolor function?

[Walter Roberson]

@Benjamin Osborne

ntimes = max(cellfun(@length, times));

for K = 1 : num_files
    these_times = times{K};
    for T = 1 : length(these_times)
        %plots are numbered along rows, reverse of usual way!!
        subplot(num_files, ntimes, sub2ind([num_files, ntimes], T, K));
        pcolor(lons{K}, lats{K}, precip{K}(:,:,T));
        title( filenames{K} + " " + string(these_times(T)) );
        xtitle('long'); ytitle('lat');
    end
end

[James Wyatt]

@Walter Roberson this has been a great help, thank you!

Do you know of a way to have this program go through multiple folders to search for and combine the .nc files? I have similar data spread over multiple folders that I would like joined.

[Walter Roberson]

@James Wyatt

The modification to the code I posted several years ago would be

projectdir = 'C:\some\location';
dinfo = dir( fullfile(projectdir, '**', '*.nc4') );
num_files = length(dinfo);
filenames = fullfile( {dinfo.folder}, {dinfo.name} );

The '**' I used means "all sub-folders at any depths". If you want to use all the folders that are immediate children of the parent directory then use '*' instead of '**'

[Bram Loef]

@Walter Roberson

And how do I modify this in a way that I only substract the precips with timeseries for a location (thus one value for lat and lon) for all different .nc files?

[Tsehaye Negash]

This program is not working.

[Walter Roberson]

@Tsehaye Negash

What error message are you encountering?

Answer 2

[Dushantha Sandaruwan WIJENDRA NAIDHELAGE]

This code can be used to substract the specific region from the set of netcdf files

clear all;clc
av_file=dir('*.nc');
ncdisp(av_file(1).name);
lon1=ncread(av_file(1).name,'lon');
lat1=ncread(av_file(1).name,'lat');
% temp1=zeros(141,121,365);
b=[1982:2021];
count=0;
for i=1:40
    a=['sst.day.mean.',num2str(b(i)),'.nc'];
    
    sst0=double(ncread(a,'sst'));
    
    a1=find(lon1>=40&lon1<=110);
    b1=find(lat1>=-10&lat1<=30);
    lon_ind=lon1(a1);lat_ind=lat1(b1);
    arb=sst0(a1,b1,:);
    [~,~,nt]=size(lhflx1);
    sst_final(:,:,count+1:count+nt)=arb(:,:,:);
    count=count+nt;
end

[DGM]

How is this an answer to the question?