GHOST's Features

Last revised 19 April 2006

Development Language

• Fortran 95 for main programs and some utility software

• Fortran 77 for other utility software

• Operating system scripts for controlling programs/utilities

Supported Platforms

• MS Windows (via Command Prompt)

• Linux (Red Hat, Fedora)

• Unix (HP-UX, Sun OS, Solaris) -- no longer supported by GSD

• VAX (VMS) -- no longer supported by GSD

• Macintosh (Mac OS X) -- no longer supported by GSD

Features

• Mathematical algorithms: Adjusts 3D, horizontal and vertical networks using the same mathematical algorithms as GeoLab, based on an adjustment in the local geodetic coordinates system of each point.

• Helmert blocking: For handling very large networks by dividing (partitioning) a network into blocks to be adjusted separately and subsequently combined -- more numerically efficient than a simultaneous adjustment.

• Correlations between position/baseline data: Allows for use of full covariance matrix for correlated data (e.g., simultaneously observed GPS baselines).

• Correlations between station constraints: Allows for weighted station constraints using full covariance matrix to account for usually high correlations that between control stations, reflecting their relative accuracy much relative accuracy as compared to their absolute accuracy (very important for the proper integration of surveys!).

• Baseline residuals: Computes position and baseline residuals in both geocentric Cartesian (XYZ) and local geodetic (NEU) components. It is important to check the residual of the vertical (U) component to check for the common occurance of antenna height problems.

• Variance factor tests: Performs statistical test of the estimated variance factor for testing the adjustment model, including constraints, data weighting and outliers.

• Individual outlier tests: Computes outliers tests using more appropriate Tau test, based on in-context testing and the Tau distribution (following Vanicek and Krakiwsky [1986] and A.J. Pope [1978])

• Goodness-of-fit tests: Computes separate and combined goodness-of-fit tests (and histograms) of residuals for different observation groups -- to check that the data are normally distributed

• Reliability estimation: Estimates internal reliability (redundancy numbers and minimally detectable biases) for quantifying potential biases/blunders that would be undetectable by the outlier tests -- probability levels of Type I and II errors are both selectable.

• Graphics: Generates files for use in MapInfo for displaying points, labels and colour-coded observation connections.

• Source code: Fortran source code to all of GHOST's programs is available for user's to add their own features and customize for their own organizations.