Organisation of input files for SAMPLER program

At this point, the reader may wish to examine the demonstration files bnd.dig, cont.dig, and sound.nod with a text editor and read the format specifications for DIGIT and NODE files in Part III. Some points to notice are:

• The outer boundary data in bnd.dig is organised in 6 blocks, which correspond to the alternate stretches of land and open sea boundary A-B, B-C, C-D, D-E, E-F and F-A in Fig. 2.1. The distinction between land and sea parts of the outer boundary is maintained so that, if desired, different sub-sampling rates can be used for each when sub- sampling with the SAMPLER program. The beginning of each continuation block is signalled by a record consisting of two values: 88 88. The SAMPLER works on the assumption that the first block of outer boundary data corresponds to a land boundary and subsequent blocks are sea and land alternately.
• The format of the soundings file sound.nod is a degenerate case of the NODE format defined in Part III. Since all the soundings become interior nodes for the depth grid, there are no boundary nodes and so the second line of the file has a zero, indicating the file contains neither outer boundary nor island boundary information, and the total number of nodes in line 1 equals the number of interior nodes in the third line.
• All TRIGRID programs accept free format input, with numeric fields separated by blanks or commas but not tabs. To reduce disk storage requirements, fixed formats are used for output. To improve readability, only I and F formats are used. Since TRIGRID uses single precision arithmetic, this means that coordinate and depth values must be scaled at the outset to suit the output formats listed in Part III, otherwise data may be accidentally truncated. For instance, longitude and latitude values to the accuracy used in many hydrographic data banks will lose significant figures if input unchanged to TRIGRID. Referring the coordinates to a datum close to the domain being modelled will avoid this problem and also leave the data better conditioned for the numerous difference calculations carried out in TRIGRID programs.