boxfilt

 boxfilt  Low-pass filters [and subsamples] using boxcar filter[s] 

  USAGE: [ulp]=boxfilt(u,nbox)
     or
         [ulp,jdlp]=boxfilt(u,nbox,jd)
     or
         [ulp,jdlp]=boxfilt(u,nbox,jd,nsub)

    Input:  u =  matrix with time series in each column
            nbox = number of points to boxfilter (must be odd)
            jd =  julian day time vector (optional)
         nsub =  subsamboxing interval for the output low-pass data (optional)
  
    Output: ulp = low-pass filtered data 
            jdlp = julian time vector corresponding to the data ulp.

    Example:   [ulp,jdlp]=boxfilt(u,[6 6 7],jd,6);   
               applies successive boxcar filters of 6, 6 and then 7,
               then subsamples the output every 6 hours
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 Version 1.0 (4/4/97) Rich Signell (rsignell@usgs.gov)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 Version 2.0 (1/10/99) Rich Signell (rsignell@usgs.gov)
 allows for successive boxcar filters to be applied by
 making NBOX a vector.  I've done this mainly for input
 to Foreman's Tidal Program, which can scale up higher
 frequency tidal coefficients if the sucessive filtering
 information is specified.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

0001 function [ufilt,jdf]=boxfilt(u,nbox,jd,nsamp)
0002 % boxfilt  Low-pass filters [and subsamples] using boxcar filter[s]
0003 %
0004 %  USAGE: [ulp]=boxfilt(u,nbox)
0005 %     or
0006 %         [ulp,jdlp]=boxfilt(u,nbox,jd)
0007 %     or
0008 %         [ulp,jdlp]=boxfilt(u,nbox,jd,nsub)
0009 %
0010 %    Input:  u =  matrix with time series in each column
0011 %            nbox = number of points to boxfilter (must be odd)
0012 %            jd =  julian day time vector (optional)
0013 %         nsub =  subsamboxing interval for the output low-pass data (optional)
0014 %
0015 %    Output: ulp = low-pass filtered data
0016 %            jdlp = julian time vector corresponding to the data ulp.
0017 %
0018 %    Example:   [ulp,jdlp]=boxfilt(u,[6 6 7],jd,6);
0019 %               applies successive boxcar filters of 6, 6 and then 7,
0020 %               then subsamples the output every 6 hours
0021 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
0022 % Version 1.0 (4/4/97) Rich Signell (rsignell@usgs.gov)
0023 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
0024 % Version 2.0 (1/10/99) Rich Signell (rsignell@usgs.gov)
0025 % allows for successive boxcar filters to be applied by
0026 % making NBOX a vector.  I've done this mainly for input
0027 % to Foreman's Tidal Program, which can scale up higher
0028 % frequency tidal coefficients if the sucessive filtering
0029 % information is specified.
0030 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
0031 
0032 nnbox=nbox;
0033 for ii=1:length(nnbox);
0034   nbox=nnbox(ii);
0035   %filter weights
0036   box=ones(nbox,1)./nbox;
0037   [m,n]=size(u);
0038   if(min(m,n)==1),
0039     m=length(u);
0040     n=1;
0041     u=u(:);
0042   end
0043   if (m<length(box)),
0044     disp('data too short for this filter!!');
0045     return
0046   end
0047   ufilt=zeros(m-nbox+1,n);
0048   for i=1:n,
0049     uf=conv(u(:,i),box);
0050     ufilt(:,i)=uf(nbox:(length(uf)-(nbox-1)));
0051   end
0052   jdf=conv(jd(:),box);
0053   jdf=jdf(nbox:(length(jdf)-(nbox-1)));
0054 end
0055 [m,n]=size(ufilt);
0056 if (nargin==4)
0057   ufilt=ufilt([1:nsamp:m],:);
0058   jdf=jdf([1:nsamp:m]);
0059 end