procnias

LSA without control points

Discussion created by procnias on Apr 25, 2014
Hello,
     This is my first post to the forum. I'm a biologist and by no
means a GIS expert, so you'll have to forgive me if my description of
my problem is inexact or lacks sufficient detail. I study a species of
bird in which the males form small, spatially stable aggregations
during the breeding season (known as leks). The average spacing
between male territories is on the order of 3-5 m. I am attempting to
create "accurate" (~1 m accuracy) maps of the males on these leks. I
have attempted to use consumer-grade GPS to obtain lat/long
coordinates (using waypoint averaging) but have not been able to
achieve the level of accuracy I need. I have since started conducting
spatial surveys by using a theodolite for measuring the coordinate
geometry (COGO) of traverses.
     I have played around with both the COGO toolbar and parcel
editing to enter my traverse data, and it's here that I am
encountering my primary problem. Because I am working in a remote
rainforest in South America I do not have accurately georeferenced
control points for use in the least squares adjustment (LSA) of the
parcel editing environment to correct for the misclosure error in my
traverses. My goal is to have a map of each lek that is relatively
accurate, so having each point accurately georeferenced is far less
important to me than having the points accurately placed with respect
to one another. The LSA in parcel editing clearly requires a network
of high-accuracy control points, so unless there is a way around this,
this may not be the way to go. The COGO toolbar presents a different
problem; Although I can do relative adjustments for each traverse
(i.e. a Bowditch/transit/Crandall adjustment) without the need for
control points, I believe each traverse represents a single closed
polygon (or multiple vertex line) not a network of polygons. This
means that I will have to make adjustments sequentially for each
closed traverse,  and the redundancy of data I have for point location
estimation from multiple traverses passing through the same point
cannot be factored into the adjustment (i.e. I cannot conduct a single
adjustment operation on the entire network of traverses). Clearly this
is also problematic since the order in which I enter the traverse data
will strongly affect the end result, and the data redundancy would
vastly improve the accuracy of my point network if it were used in the
adjustment.
     So, my question is whether anyone has ideas for achieving a
relative LSA (or Bowdtich/transit adjustment) on a network of unjoined
parcel points created from traverse data. Thanks in advance for your
help!
Cheers,
Julian

Outcomes