Errors in Data-Collection Methods
Survey errors usually arise from three sources: operational errors, instrument errors, and external
errors. Operational errors include operator errors and limitations of the surveying procedure.
Instrument errors result from limitations of instruments or devices with which measurements are
taken. External errors arise from variations in natural phenomena such as temperature, humidity,
wind, and gravity. These errors, along with how they were identified and removed, are discussed
Operational errors of the Zeiss survey system resulted from improper leveling of the instrument,
mis-aiming the instrument at the center of the prism cluster while taking the measurement, an
error in positioning the instrument in coordinate space (particularly in elevation), and movement
of the tripod during the survey. Improper leveling of the instrument affected all the points
measured and was not always easy to detect. Incorrectly aiming the instrument also resulted in
an error but only on individual points. This kind of error occurred when the instrument was
triggered to begin a reading prior to properly aiming at the prisms. This resulted in sampling the
azimuth and zenith angles at the time of triggering. These incorrect angles were then used with
the distance to the prisms to calculate the CRAB location. Errors affecting single points were
usually easy to detect and remove. Errors from mis-positioning the instrument and from
movement of the tripod were eliminated after August 1985 when two permanent
fixed-instrument mounts were established on the pier and on the roof of the facility.
A different type of operational error resulted when topographically important points, such as the
bar crest or trough, were missed. Survey points were selected based on the timed travel of the
CRAB, with more points taken close to shore where the profile shape is more complex. While it
was possible for the Zeiss operator to follow the vertical movement of the CRAB as it moved,
small features and some peaks of significant features were sometimes missed or inadequately
Use of the Geodimeter 140T greatly reduced operational errors. The auto-tracking system
eliminated errors caused by missing topographically important points and errors caused by
improperly aiming at the prisms.
The Zeiss has not been subject to systematic instrument errors, except for requiring occasional
servicing. However, the Geodimeter has a number of idiosyncracies that were not fully
understood during the early Geodimeter surveys.
One of the instrument errors of the Geodimeter results from the separation of the tracking unit
and the angle measurement unit. For accurate vertical measurements, the two units must be
parallel. Instead of attempting to fine adjust the parallelism, the angular error was computed and
applied to the measured vertical angle in the collection software. Unfortunately, the angle can
change during a survey, apparently resulting from temperature changes. Although checks of the
vertical angle correction were made, they were not always made frequently enough to fully
remove the error. The associated vertical error increases with distance from the instrument. The
vertical angle change was usually less than 30 sec. Over a distance of 1,000 m, a 30-sec angle
error can result in a maximum elevation error of 16 cm.
Another error associated with the vertical angle correction resulted from the instrument shifting
out of level caused by the slight shifting of the pier and the FRF building, probably by differential
heating. 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 compensator.
However, this applies only for the horizontal compensation, not for the vertical. Consequently,
even if tilted, the instrument continued to collect erroneous vertical data with no indication,
except for the bubble level on the instrument being off-center. Because the operator watched a
PC screen, and not the instrument, 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 in a
"dome shelter" located on the roof of the FRF building. Frequently sighting and re-sighting a
prism of known location was also used as a setup check and for computing corrections.
Unfortunately, during the first year of use, some of the out-of-level errors were wrongly corrected
for by re-computing the vertical angle correction. This unfortunately added to the error. Once
the tilt error was fully understood, a program of frequent level checks was instituted.
The other instrument error resulted from oscillations of the telescope caused by an improperly
adjusted tracker amplitude as it searches to locate the center of the prism. This resulted in jagged
data with an oscillatory amplitude of a few centimeters. Although the data appeared to follow
the true profile shape somewhat, it was difficult to remove the oscillations since they were not
centered about the true profile. A final problem was interference from sunlight, which could
overwhelm the tracker power. This problem affected early morning surveys of lines 188 and
190. The problem was eliminated by surveying the northern profile lines, 58 and 62, first.
Electronic survey instruments are sensitive to atmospheric and climatic variation since they use
the speed of light to determine distance and optical aiming to measure the angles. The
instruments allow for rough adjustment for these variables. During the summer months heat
shimmer and the temperature gradients near the land-sea interface may also affect the accuracy of
the angular measurements. The authors have no quantitative feel for these sorts of errors other
than they are usually negligible relative to the other types of errors. These errors may explain
some systematic offshore changes amounting to several centimeters during the summer months
that do not appear to be related to the dominant processes.
An additional, but un-quantified source of error resulted from the thermal expansion and
contraction of the CRAB frame and its liquid-filled tires. Although the CRAB's height was
routinely measured on land, there was no way to measure its height fully submerged. No
adjustment to the data has been made to account for this effect. The variation is suspected to be
on the order of 3 to 6 cm (0.1 to 0.2 ft) between summer and winter extremes, but it occurs
gradually. This amount of variation, combined with the slight uncertainty of the over 300
different instrument setups, results in a survey noise level that obscures small bed-level changes
along the offshore reaches of the profiles.
Error Identification and Correction
Example of typical survey errors and their effects
Examples of typical nearshore survey errors and their impact are shown in the figure above.
Errors in the data were most easily recognized through comparison plots. All data were
compared to at least the previous survey of the profile. Usually the data were error free. If not,
this comparison showed where possible errors occurred. The suspect points were then inspected
more carefully. A decision was made as to whether the point or points were in error or
represented real changes. The figure above shows the utility of comparison. Questionable data,
where no clear error could be discerned, were noted and compared to the next survey as well.
This provided a double-check. Errors were also identified using the measured location of the
reference prism checked during the instrument setup and throughout the survey.
Corrections to the data consist of two primary types: deletion of points and subtraction or
addition of biases. Data points that were obviously erroneous were removed from the data.
These erroneous data points usually resulted from incorrect targeting of the Zeiss to the prism
cluster, improper adjustment of the Geodimeter tracker amplitude, or the use of points that were
unaccountably, but obviously, wrong. Biases were either constant or gradual (distance
dependent) due to improper leveling of the instrument. Vertical errors due to leveling became
increasingly evident with increasing distance.
Most often, constant vertical offsets were the result of either improper stationing of the
instrument (setting its elevation incorrectly), or rarely, improper entry of the elevation of the
prism cluster on the CRAB. These errors could be traced back to the data through the recorded
setup procedure. If there was no evidence of a mistake in stationing the instruments or entry of
the prism height, removal of the suspected bias was dependent on three factors. The bias had to
extend over the entire profile (past the normal closure point); the bias was restricted to only one
or two of the four profile lines; and there had to be no reason to expect evidence of profile
activity at depth. For instance, if the measured profile showed significant erosion at a depth
below the extreme profile closure depth during a period of below normal wave activity, errors
were strongly suspected. However, the bias was removed from the data only after a second
survey of the profile confirmed that an error had been made.
Gradual biases, or rotations of the data, which resulted from calibration or leveling errors rather
than stationing errors, were more difficult to discern in the data. If the shift in the data could be
directly attributed to a mis-calibration of the internal level compensation, or to a leveling error
during or after setup of the instrument, then the rotation needed to correct the data was
determined by re-calibration of the level.
All changes made to the data, with the exception of the tilt adjustment of the CRAB, were
recorded in processing and data collection logbooks and were coded into the data file.