STAR*NET V6 - Circe
STAR*NET V6 - Circe
STAR*NET V6 - Circe
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Chapter 5 Preparing Input Data<br />
In general, you should always use standard errors, global project defaults or otherwise,<br />
that reflect the conditions and equipment used to collect your field data. If you believe,<br />
for example, that your angles are “good” to within plus or minus 4 seconds, then do not<br />
use a standard error of 2 seconds to get a “better” adjustment. It will not help. You will<br />
likely get a Total Error Factor in your adjustment statistics that is very different than 1.0,<br />
and the adjustment will fail the Chi Square test.<br />
Except for direction sets, <strong>STAR*NET</strong> assumes that all input observations are<br />
uncorrelated. This means, for example, that there is no interdependence between two<br />
distance measurements or between angular and distance observations. Note that the<br />
output coordinates from the program will almost certainly have correlation between their<br />
Northing and Easting coordinates. Refer to the section “Ellipse Interpretation,” in<br />
Chapter 8, “Analysis of Adjustment Output” for further information on output coordinate<br />
correlation.<br />
Why do I want to use weights?<br />
Independent weights assigned to observations allow <strong>STAR*NET</strong> to solve the survey<br />
network more rigorously. Less precise measurements (larger standard errors, smaller<br />
weights) will have less of an influence on the solution, while more precise values<br />
(smaller standard errors, larger weights) will have more of an impact. For example, if<br />
you are combining some old survey data using taped measurements with a new survey<br />
using modern EDM measurements, you may want to give the taped distances higher<br />
standard errors (lower weight) than the EDM measurements.<br />
In order to introduce constraints into a network, it becomes possible to fix an observation<br />
by giving it a very high weight (very small standard error). For example, you may wish to<br />
have a certain azimuth along a line - simply fix the input value with a small standard<br />
error by using the “!” symbol. Conversely, some input values may be only<br />
approximations that you would like not to influence the adjustment. They can be left free<br />
by giving them low weights (high standard errors) with the “*” symbol.<br />
<strong>STAR*NET</strong> has two special codes (“!” and “*”) to allow you to indicate fixed and free<br />
observations, in order to simplify data input. You should normally use these special<br />
codes rather than creating your own very low or very large standard error values to fix or<br />
free observations.<br />
99