|
||||||||||
| PREV PACKAGE NEXT PACKAGE | FRAMES NO FRAMES | |||||||||
See:
Description
| Class Summary | |
|---|---|
| CatrasFile | |
| CatrasReader | |
| CatrasToTridasDefaults | |
| CatrasWriter | |
| TridasToCatrasDefaults | |
This is a binary format for software written by Aniol, first released in 1983.
Several versions of CATRAS were released over the years, the most recent we have seen is v4.35 released in 2003. It is uncertain if there have been changes made over the years. The code in this library is based on Matlab, Fortran and C code of Ronald Visser, Henri Grissino-Mayer and Ian Tyers.
Some of the bytes in CATRAS files contain character information. In the case each byte represents a letter. In java an array of bytes can be directly decoded into a string.
As a byte can only represent 256 values, whenever an integer is required CATRAS stores them as byte pairs. Each byte pair consists of a least significant byte (LSB) and a most significant byte (MSB). The order that they appear in files typically varies between platforms and is known as 'endianness'. As CATRAS solely runs of Microsoft (x86) processors we can safely assume that all CATRAS files will be using little-endian (i.e. LSB MSB). The counting in a byte pair therefore works as follows:
| LSB | MSB |
| 0 | 0 |
| 1 | 0 |
| 2 | 0 |
| ... | ... |
| 255 | 0 |
| 0 | 1 |
| 1 | 1 |
| 2 | 1 |
| ... | ... |
| 255 | 255 |
A byte pair can therefore store 256x256=65536 values (more than enough for most number fields). Matters are complicated though by the need to store negative numbers. In CATRAS pairs with an MSB<=128 are positive, while pairs with an MSB ranging from 255 to 128 (counting backwards) represent negative values.
Categories are typically recorded as single bytes as most categories have just a few possible values. They can therefore be conceptualised as being integers where 0=first option, 1=second option etc. The exception to this is for species as there are more than 256 species. In this case, a byte pair is used in exactly the same way as described for integers above. The only problem for species is that the codes are unique to each laboratory and refer to values enumerated in a separate '.wnm' file. Without this dictionary the species code is of little use.
Dates are stored as three single bytes, one for day, one for month, one for year. With only 256 values available for 'year', all dates are stored with 2 digit years e.g. 25/12/84. When converting to TRiDaS all years >70 are treated as 20th century, whereas years <70 are treated as 21st century. This is an arbitrary decision for use in this library as CATRAS does not care either way.
The first 128 bytes contain the file header information and the remainder of the file contains the ring width data and sample depth data. Our current understanding of the header bytes is as follows but I'm not convinced that these are all correct. Deciphering these requires painstaking work because we must try to ascertain how each byte is being used (e.g. as a byte pair, single byte or as a string):
The remaining bytes in the file contain the actual data values stored as integer byte pairs. It appears that older version of CATRAS included one or more padding values of -1. These values should be ignored. The end of the data values are indicated by a stop value of 999.
Following the ring width data values there are 42 bytes of unknown meaning. These are then followed by byte pairs representing the counts/sample depth for each ring if the series is a chronology.
|
||||||||||
| PREV PACKAGE NEXT PACKAGE | FRAMES NO FRAMES | |||||||||