Understanding UHDAS heading correction files

• Whether the heading correction is done during averaging or afterwards, quick_adcp.py will write out a file, “ens_hcorr.asc”:

% headings listed are original heading device
% sign is correct for use with "rotate"

% enddday, mean head, last head, dhfix, stddh, ndh, badmask(bad==1) persist_win
85.0893999     250.81    250.57       1.10    0.24     55     0      0
85.0928790     250.80    249.99       1.04    0.24     73     0      0
85.0963699     250.53    250.85       0.97    0.29     54     0      0
85.0998491     250.25    248.86       0.88    0.33     78     0      0
85.1033283     249.63    249.04       0.97    0.29     95     0      0
85.1068077     249.61    248.37       0.99    0.22    105     0      0
85.1102870     248.74    246.99       1.05    0.26     40     1     10
85.1137781     248.22    248.26       1.05    0.26     43     1     10
85.1172572     247.83    246.84       0.74    0.28     53     0      0
85.1207364     247.21    247.10       1.04    0.26     36     1     11
85.1242273     248.48    248.92       0.95    0.22     45     0      0
85.1277066     249.44    250.04       0.95    0.32     72     0      0
85.1311741     250.53    250.55       0.98    0.25     65     0      0
85.1346649     250.37    250.23       0.98    0.22     33     1     14
85.1381439     250.27    251.26       0.87    0.22     45     0      0
85.1416348     250.17    249.00       0.98    0.21     28     1     15
85.1451138     249.08    247.05       0.98    0.29     15     1     15
85.1485930     248.35    248.60       0.98    0.15     17     1     15
85.1520838     248.10    247.69       0.97    0.18      4     1     14
85.1555513     247.55    247.85       0.97    0.21     16     1     14
85.1590425     247.21    248.45       0.97    0.30     15     1     14
85.1625217     247.79    248.38       0.97    0.22      8     1     14
85.1660010     248.65    248.29       0.97    0.25     37     1     13
85.1694922     248.84    248.65       1.02    0.23     70     0      0
85.1729714     248.82    248.30       0.90    0.27     83     0      0
85.1764624     247.92    246.79       0.86    0.22     97     0      0
85.1799298     247.60    247.94       0.97    0.24    168     0      0
85.1834090     247.34    246.68       0.93    0.24    121     0      0

   (1)          (2)        (3)         (4)     (5)     (6)   (7)    (8)


% columns are valid for the ensemble in question

(1): zero-based decimal day, at the time of the ensemble end
(2): mean gyro heading
(3): last gyro heading
(4): median heading correction (see note below)
(5): standard deviation of good heading correction values
(6): number of good  heading correction values
(7): flag indicating whether median heading correction was acceptable
(8): NEW: number of previous values used for a median ("persistence")

• If “–ping_headcorr” is chosen for the heading correction, the rotation is done IN the ensembles, and the time_angle file “ens_hcorr.ang” is extracted from “ens_hcorr.asc” for debugging purposes. NO explicit “rotate” step is performed by quick_adcp.py
• if “–post_headcorr” is chosen for the heading correction, another pair of files is written:

• “hcorr.asc” : one-minute averages of heading correction (columns 1-7)
• “hcorr.ang” : values from “hcorr.asc” gridded onto the actual ensemble times

Persistence

Let’s use an Ashtech as an example. For a given ensemble, the Ashtech values are screened according to the instrument’s own QC criteria. If more than than 15% of the values are “good”, then the median is used as the heading correction estimate for this ensemble. If fewer than 15% of the values remain, the estimate is considered to have low reliability.

Prior to March 2006, low-reliability heading correction estimates were left as zeros so as to be identifiable and reasonable. (reasonable, but not very good).

After March 2005, low-reliability estimates of heading correction are replaced by the median of the hig-reliability values of the previous N ensembles (N=24, usually 2 hrs). This should only come in to effect when the Ashtech in question has dropouts.

This is an attempt to make underway data more reliable if is collected with a flaky heading-correction device. It is in your best interests to scrutinize the heading correction used in your data, and redo it if necessary. There is no substitute for context, and context is limited when acquiring real-time data.