20) 
rom 
the 
be 
any 
ial 
21) 
be 
Tor 
ers 
vey 
23) 
mal 
coefficient matrix is needed for computations. If. only a few 
points would be included within the constrained solution, then 
it would be necessary to extract from M, only that part which 
relates to points upon which the constraint is involved. The 
influence of the constraínt can be computed on a point. by point 
basis. Under the case where multiple observations were made on 
the same point, 1.e., distances and angles to a number of 
adjacent stations, them correlation could be present and 
equation (23) would be formed completely. Again, because of 
the sparsity nature, no "significant computational burden would 
be required. SDarsitv. is preserved because measurements 
normally involve adjacent points and seldom go across sections 
of land. This is not true for geodetic values though. In 
cases such as this, proper ordering could lead to the creation 
of a banded/bordered matrix in which methods such as recursive 
partitioning can be employed. 
KILUSSGUCK TEST AREA 
A test adjustment was performed on the Killsbuck Area located 
in Holmes County, Ohio [Burtch, 1983]. The area encompasses 
approximately 30 square kilometers. It‘ was flown with a Zeiss 
RMKAR aerial camera with a focal length of 152.02 mm at à 
height of approximately 2100. meters from. datum. Ihe camera 
utilizes a 23x23 cm reseau pattern. The block was flown in two 
strips with -a total of 32 photographs with 60% overlap and 
sidelap. A total of 31 . ground points were used of which 38 
were full control points, 31 were only vertical control points, 
3 were only horizontal control points, and 9 were unknown 
survey points. 
Three different ajustments utilizing Case TITI theory (Merchant, 
1973] were. performed. on. the data. The. third adjustment 
involved fifteen distance constraints. Eight control points 
associated with some of the constraints were treated as 
non-control..points.. The „standard deviations of the points 
which were changed from control .to non-control status . are 
predominantly below the 5-cm level after the adjustment. One 
of the advantages in using a sequential adjustment is that the 
influence of the constraint is computed and this information 
can sBe:usedc.to evaluate. the: integrity of the survey data. 
Table:l.-shows the influence .of the constraints on the survey 
parameters. One can see that ip. the X-direction, a very high 
influence occurs with points 1022 and 1024 which involved the 
shortest distance constraint used in the adjustment, Because 
of the large influence and since no perceptual change in the 
residuals for the photo observations can be found, i1t..may,be 
inferred that one or both have an error in their control value. 
A snallc-error:forsitheclength. of a. shont line would have a 
greater influence in the localized... area.. Therefore,, One can 
sec that this::is a:convenient mechanism by which the validity