fixed prism of known elevation. This vertical angle correction was then 
applied to the measured vertical angle in the collection software. It wasn't 
learned until later that the correction angle changes during the day, especially 
on hot days. Although checks of the vertical angle correction were made, they 
were not made frequently enough to fully remove the error. This vertical error 
increases with distance from the instrument. Over a distance of 1,000 m, a 
20-s angle error, if uncorrected for, would result in a vertical error of 10 cm 
(elevation checks were also made with the ELTA-2S, but there were no vertical 
problems with the Zeiss data) . 
Another error associated with the vertical angle correction occurred when 
there was slight shifting of the instrument out of level (probably due to 
differential heating) during full days of surveying. Some of this movement is 
automatically compensated for by the Geodimeter, and it is designed to stop 
acquiring data if the instrument tilts outside the range of the internal 
compensator. However, it was determined well after DELILAH that this 
applies only for the horizontal compensation, not for the vertical. 
Consequently, even when tilted, the instrument continued to collect slightly 
erroneous vertical data with the only indication being an off-centering of the 
bubble level. Because the operator watches a computer screen, and not the 
bubble level, a tilted instrument could go undetected. From the beginning, this 
error was minimized by sheltering the Geodimeter with an umbrella or by 
setting it up ina "dome shelter" located on the roof of the FRF building. 
Unfortunately, during the first year of use, some of the out-of-level errors were 
wrongly corrected for by recomputing the vertical angle correction, thus 
compounding the error. This type of error manifests itself as an offset of the 
survey data that increases with distance from the instrument. In several cases, 
the survey data for one or more profile lines collected in sequence shifted 
during one survey and then unnaturally shifted back for the next and later 
- surveys. When this shift was significant, the data were adjusted accordingly. 
A further source of error was vertical oscillations of the Geodimeter which 
occurred because of an improperly adjusted tracker amplitude. This resulted in 
jagged data with an oscillatory amplitude of a few centimeters. Although the 
data follow the true profile shape, it is difficult to remove the oscillations since 
they are not centered on the true profile. At its worst, the tracker would drift 
completely away from the prism and temporarily lock onto a horizontal plane 
above the actual position of the moving prism for a short distance before 
reacquiring the prism center, resulting in a “step function” appearance to the 
data. These problems affect the nearshore data of some of the profile lines 
from the September 1990 CRAB survey that is part of the 13-m-deep survey 
data and affected several of the minigrid surveys. These errors were not a 
problem in most of the minigrid data and are not to be confused with the 
megaripples, which were reliably surveyed on many of the profile lines. Data 
affected by these oscillations are characterized by small oscillations along the 
entire profile line with no smooth regions. Major changes caused by the 
tracker drift were identified by overlaying all surveys of each profile line and 
Appendix A Surveying A3 
