|
|
|
This figure shows the two major physical record types that are recorded on the field tapes: the volume header and the lidar data record. Dashed lines on these figures represent logical groupings of defined data structures and solid lines depict actual defined data structures.
Note: Areas enclosed by dotted lines indicate logical groupings of physical data blocks. Areas enclosed by solid lines indicate physical, defined data blocks.
Exabyte tape drives go through a self test and then write a beginning of tape record when they are first asked to record on a new tape. This process can take up to several minutes, if this were to occur in-flight during a ping-ponging action of the tape drives, data would be lost. Since there is not enough buffer space to carry over that long of time interval, a dummy end-of-file (EOF) mark is written on the start of each field tape. A write EOF command is simple to give, and it causes the drive to perform all of its initialization functions as well as completely dumping its own buffers. Therefore, software written to read a field tape must handle a EOF at the very beginning of each tape.
A physical data tape may consist of multiple files (volumes). The first record of every file is a volume header. This record will then be followed by multiple data records. These records may consist of different data types, at the present time, however, only lidar data rays are being recorded. The first long word of every physical record on the tape identifies that record by containing four unique ASCII characters. If post processing software is not using a particular type of data, it can simply read a record and check the first word. If it is the wrong type of data, the next record can be read. Whenever any of the parameters that are described in the volume header change, the old volume header is rewritten to tape, a file mark is written on tape, and the new volume header is written to the tape. In this way, each file on the tape will consist of any number of data records sandwiched between two identical volume headers.
The volume header format is shown in the attached document labeled Volume Header Description. The control processor (SUN workstation) is responsible for building this header. It will send every new header to the Vxworks processor which will read the header and know how to act. System operation should be suspended before the new header is sent out, as the Vxworks processor will have to calculate new offsets etc. as a result of the header.
The volume header begins with a volume descriptor that is identical to that defined by Lee. Following this descriptor, will come the sensor descriptor block for the SABL lidar. This sensor descriptor block will completely describe the operation of the SABL lidar. The block begins with a radar descriptor following the exact format as defined by Lee. Most things recorded in this block are appropriate for both radar and lidar, so it is used to keep things uniform. This descriptor is then followed by the field lidar info. This is the detailed information needed to keep track of the lidar calibration in the field. The field lidar info is followed by a cell spacing descriptor to define the spacing of the data cells for each lidar. This descriptor is used in place of the cell range vector, described by Lee, to save space in the header. Following the cell spacing descriptor are as many parameter descriptors as there are recorded parameters for the lidar being described. In the first SABL there will be two parameters: returned power from infra-red and returned power from green. These parameter descriptors are in the same format as those described by Lee. Additional sensor descriptors might follow those specifically for the lidar. If used, these blocks will describe the format of various other sensors that are recorded on the tape.
The lidar data record has a maximum length as described in the volume descriptor. It contains an integer number of field data rays. These data blocks will come from the lidar that is described in the volume header. The field data ray starts with a ray info block followed by a platform info block. Both of these blocks are defined as described by Lee. Following these blocks comes the lidar field data. This block is different than the parameter data as defined by Lee. Due to the various data reduction algorithms, each lidar data record may contain a different number of field data rays each with a different length. The field data ray is the format in which the data are passed out of the lidar Vxworks processor, over the Ethernet, and into the display and analysis processors.
Below are definitions for the various data elements (gate, cell ray, sweep, volume etc.) and format components (block, descriptor, header, etc.).
The field data tape for the SABL closely follows the DOppler RAdar Data Exchange Format (DORADE)[1], as described by Wen Chau Lee. This is a very efficient tape format since the housekeeping information is recorded only when it changes and only essential information such as time, aircraft position, etc. are recorded with each data ray. Data rays are blocked together into large physical records to allow maximum tape usage. Changes to major lidar system parameters, such as the number of gates, scan speed, photomultiplier tube (PMT) gain, etc., cause the tape system to write an end of file. Therefore, a tape may consist of multiple files. These files are the volumes as described by Lee and will always start with a volume header and end with the same volume header.
File marks (with headers on both sides) are periodically (every 95 MBytes of tape) placed on the tape for redundancy and ease of using a large data tape. The volume number (written in the header) will not change across redundant tape headers.
Basically the SABL field tape is a super set of the DORADE format. For example there are several extra field blocks that describe the basic operation of the lidar more specifically. There are, however, two major differences between the field tape and the exchange format.
One major difference will be that the sweep record will not be used. Physical lidar data records are created from multiple data rays, but are not grouped into sweeps.
The other major difference is that the descriptor that describes the spacing of the data cells is different than that used in the DORADE format.
Doppler Radar
Data Exchange Format (DORADE)