c. Performing partially constrained adjustments takes advantage of the inherent higher-accuracy
GPS data relative to the existing network control. Less warping of the GPS data (due to poor existing
networks) will occur.
d. A partial constraint also lessens the need for performing numerous trial-and-error constrained
adjustments in attempts to locate poor external control points causing high residuals. Fewer ties to the
existing network are needed if the purpose of such ties is to find a best fit on a fully constrained
adjustment.
e. When connections are made to the NAD 83, relative accuracy estimates of NGRS stations can be
obtained from the NGS. Depending on the type of adjustment software, these partial constraints may be
in the form of variance-covariance matrixes, error ellipses, or circular-accuracy estimates.
5-52. Rigorous Least Squares Adjustment. Adjustment of GPS networks on PCs is typically a trail-
and-error process for both the free and the constrained adjustments. When a least squares adjustment is
performed on a network of GPS observations, the adjustment software will provide 2D or 3D coordinate
accuracy estimates, variance-covariance matrix data for the adjusted coordinates, and related error
ellipse data. Most software programs provide relative accuracy estimates (length and azimuth) between
points. Analyzing these various statistics is not easy, and they are also easily misinterpreted. Arbitrary
rejection and readjustment to obtain a best fit should be avoided. The original data-reject criteria must
be established and justified in a final report document.
a. When a series of loops are formed relative to a fixed point or off another loop, redundant
conditions are formed. This is comparable to loops formed in conventional differential level nets.
These different loops allow forward baseline-vector position computations to be made over different
paths. From the different routes (loops) formed, different positioning closures at a single, fixed point
result. These variances in position misclosures from the different routes provide additional data for
assessing the internal consistency of the network, in addition to checking for blunders in the individual
baselines. The number of different paths or conditions is partially related to the number of degrees of
freedom in the network.
b. Multiple baseline observations provide additional redundancy or strength to a line or network
since they are observed at two distinct times of varying satellite geometry and conditions. The amount
of redundancy required is a function of the accuracy requirements of a particular survey.
c. Performing a free adjustment on a complex network containing many redundancies is best
performed using the least squares methods. Adjustment methods are difficult to evaluate when complex
interweaving networks are involved. Least squares adjustment software will output various statistics
from the free adjustment to assist in detecting blunders and residual outliers in the free adjustment.
Most commercial packages display the normalized residual for each observation (for example, GPS,
EDM, angle, or elevation), which is useful in detecting and rejecting residual outliers. The variance of
unit weight is also
5-47
EN0593
Previous Page
