eofproj [ parameter=value ... ] [ inputfile... outputfile ]
Parameters are: eigen_file, variable, eigen_name, eigen_dim, modes.
eofproj projects the first modes orthogonal vectors onto a variable to obtain the associated expansion coefficents. The eigenstructure is read from the eigen_file which must be an output dataset from eof, svd or cca. This file must contain variables "X_vec", "X_amp" and "X_var", which contain the eigen structure for eigen_name "X". These variables represent eigenvectors, expansion coefficients and eigenvalues (or mode variances), respectively.
The input variable that is to be projected must have all the same dimensions as "X_vec", except for eigen_dim, which for the case of the eigen_file the corresponding dimension will be the dimension "\FImodes" (representing the number of modes in the eigen structure), and in the case of the input file is arbitrary and represents the number of "instances" of the input variable along the dimension eigen_dim. The output variable contains the expansion coefficients and is two-dimensional. The first dimension is "modes" and has length modes, the second dimension has size equal to the number of "instances" of the input variable.
Specifies the file containing the eigen structure. This must be an output dataset from eof, svd or cca.
There is no default.
Specifies the input variable to project the eigen structure onto.
There is no default.
Specifies the prefix name of the eigen structure in the eigen_file (e.g. "X" in the description above).
There default is the name entered for variable.
Specifies which dimension of the input variable represents the different "instances" of the input data. If there is only one instance and thus no explicit dimension, then no dimension name is entered.
The default is none.
Specifies the number of modes to project onto the data.
Valid values are [ > 0 and <= size of eof_file dataset dimension modes]. There is no default.
The example in eof computes the EOFs for a time series of spatial patterns of sea surface temperature (sst) and retains the principle 10 patterns in the dataset sstsweof.tdf. Assuming that these 10 patterns represent the "canonical" modes of the data, any newly observed pattern of sst can be projected onto these modes to determine the degree each of these empirical modes is represented in the new sst sample. In this example the dataset newobserv.tdf contains one instance of an sst pattern (Note: if it contained N patterns along dimension "time" for example, then eigen_dim would be specified as "time").
% eofproj in/out files : char(255) ? newobserv.tdf observproj.tdf eigen_file : char(255) ? sstsweof.tdf variable : char( 31) ? sst eigen_name : char( 31) ? [sst] eigen_dim : char( 31) ? [] modes : int ? 10 % contents observproj.tdf printout : char( 3) ? [no] Contents of File: observproj.tdf Page 1 Dimension Size Coord Scale Offset mode 10 ? 1 0 Attribute Type Units Value history byte Variable Type Units sst_amp float Variable Dimension Size sst_amp mode 10 Variable BadValue ValidMin ValidMax Scale Offset sst_amp -3.4028e+38 -3.4028e+38 3.4028e+38 1 0
datasets, eof, spectral, eof, eoffilt, svd, cca, emath, laminate, dimavg, burst, xcorrel, anomaly.
Last Update: $Date: 1999/05/10 20:13:37 $