next up previous contents
Next: Auxiliary information array indices Up: Indices Previous: Dimension Indices   Contents


START and END Fields

START and END are index fields contained in the DESCRIP1, DESCRIP3, BADVAL, PROCSPEC, PROCDUP, AUXSPEC, and PAKSPEC record types. These records can apply over specific, limited ranges of the data dimensions. For example, a BADVAL record can define a bad-data flag value which applies only to a specific subset of the data: any values found in that subset which match the bad-data flag are considered to be bad or missing data, while the same values found outside the subset are treated as normal, legitimate data values.

The START and END indices are the mechanism for defining such subsets. Both START and END are one-dimensional arrays (vectors) of indices. These are indices into the virtual array defined in Section 2.3.2. If we denote the order of the Level 0, Level 1, and Level 2 dimensions by $I$, $J$, and $K$, respectively, then a PROCSPEC.START array is $I+J+K$ elements long. If a record type is to apply over a specific range of values in each dimension, the START indices specify the beginnings of those ranges, and the END indices specify their ends. (Note that an index of -1 represents the last element of a dimension.)

Some START and END records do not cover all the indices of the virtual array. Every BADVAL record, for example, must specify bad-data flags for all grid points in all Level 0 dimensions; its START and END fields refer only to ranges in the Level 1 and Level 2 dimensions and are hence $J+K$ elements long.

Likewise, DESCRIP1, which must specify which data records are described by the Level 1 dimension being defined, can cover ranges only of the Level 2 dimensions. Its START and END fields, then, are only $K$ elements long.

For examples, see Section 3.3.


next up previous contents
Next: Auxiliary information array indices Up: Indices Previous: Dimension Indices   Contents
Eric Nash 2003-09-25