The essence of the system is a common data format in which time series polarimetry data can be represented. This format is based on the Starlink standard data format described in SGP/38. In particular the polarimetry example given in that document is the basis for the TSP data format. The hierarchical structure of a typical TSP dataset is shown below.
| Component | Type | Description |
| DATA_ARRAY(N,M) | _REAL | Stokes I array |
| VARIANCE(N,M) | _REAL | Variance on I array |
| LABEL | _CHAR | Label for data |
| UNITS | _CHAR | Units for data |
| AXIS(2) | Structure Array | Axis information |
| AXIS(1) | AXIS | Wavelength Axis |
| .DATA_ARRAY(N) | _REAL | Wavelength Axis Data |
| .LABEL | _CHAR | Wavelength Axis Label |
| .UNITS | _CHAR | Wavelength Axis Units |
| AXIS(2) | AXIS | Time Axis |
| .DATA_ARRAY(M) | _DOUBLE | Time Axis Data |
| .LABEL | _CHAR | Time Axis Label |
| .UNITS | _CHAR | Time Axis Units |
| MORE | EXT | Extension structure |
| .POLARIMETRY | EXT | Polarimetry Extension |
| .STOKES_Q | NDF | Q Stokes parameter structure |
| .DATA_ARRAY(N,M) | _REAL | Stokes Q array |
| .VARIANCE(N,M) | _REAL | Variance on Q array |
| .STOKES_U | NDF | U Stokes parameter structure |
| .DATA_ARRAY(N,M) | _REAL | Stokes U array |
| .VARIANCE(N,M) | _REAL | Variance on U array |
The DATA_ARRAY component shown here may actually be replaced by a structure which holds the data array and other components.
The TSP structure is a special case of the NDF (Extensible N-Dimensional Data Format) described in SGP/38. The main data array of the NDF structure contains the I or Intensity Stokes parameter data. A polarimetry extension structure contains up to 3 additional NDF structures corresponding to the Stokes parameters containing the polarization information. Any number from zero to three of these additional Stokes structures may be present. The above example has the two linear polarization Stokes parameters in structures STOKES_Q and STOKES_U. For circular polarization data a structure called STOKES_V would be used.
The STOKES NDF structures contain only DATA_ARRAY and optionally VARIANCE components. The axis, label and units information pertaining to the Stokes parameters is that in the main structure.
All the data arrays used in TSP are simple or primitive arrays. Other types of data arrays such as SCALED arrays, SPACED arrays etc. described in SGP/38 are not supported at present.
The TITLE, QUALITY and HISTORY components described in SGP/38 are not currently used by TSP. If present they will be propagated from input to output files.
TSP datasets may be one, two or three dimensional. A 1D dataset represents a polarization spectrum. The axis is a wavelength axis. As well as representing spectra the wavelength axis is also used to contain the wavelengths of a small number of broad band channels for data resulting from instruments such as the Hatfield Polarimeter.
2D datasets may be either time series polarization spectra, or polarization images. For time series polarization spectra the first axis is the wavelength axis, and the second axis is the time axis. The time is represented in the form of the Modified Julian Date (MJD = JD - 2400000.5) in a double precision array. Note that special cases of the time series dataset are those with a wavelength axis of size 1, and with no additional Stokes parameters. Thus simple time series photometry can be represented in this way.
3D datasets represent time series imaging (or imaging polarimetry) data.