← Back to the tool

Help & methodology

Entering data

  • Units. Distances and coordinates are US survey feet (1 ft = 1200⁄3937 m exactly).
  • Angles are angles right — turned clockwise from the backsight to the foresight. Both traverse directions work for a loop; enter stations in the order you occupied them.
  • DMS entry. A dotted angle is always read as ddd.mmss — 123.4530 means 123°45'30", never decimal degrees. Extra digits are decimal seconds (123.45305 → 123°45'30.5"). Spaced (123 45 30), dashed (123-45-30), and symbol (123°45'30") forms also work. Watch the live echo under each field.
  • Directions (first-leg azimuth, azimuth marks) also accept quadrant bearings such as N 45°30'20" E.
  • Closed loop. One row per station; the last row's distance closes back to station 1. The starting coordinate and first-leg direction fix the traverse in space (they are held, not adjusted).
  • Link traverse. First and last rows are the control stations. Their angles are turned from (at the start) or to (at the end) the azimuth marks whose azimuths you enter in the setup panel.

CSV format

The Download input (CSV) button writes this format, and Import CSV reads it back — use it to save and share jobs. Lines starting with # are comments. A file containing only station rows is also accepted.

mode,loop
start,10000.00,5000.00
first_leg,106.5956
station,angle,distance
1,83.5157,617.08
2,119.4829,564.12
3,100.0033,507.23
mode,link
start_control,1000.00,1000.00
end_control,1186.50,1223.00
opening_azimuth,180.0000
closing_azimuth,90.0000
station,angle,distance
R,240.0000,200.00
U,150.0000,100.00
S,240.0100,

How the adjustment works

  1. Closure check. Angles are carried around the traverse to get the angular misclosure, balanced equally, and coordinates are run forward to get the linear misclosure and precision ratio — the familiar compass-rule numbers, shown before any least squares.
  2. Least squares (default). Every angle and distance becomes a weighted observation equation (Ghilani, Adjustment Computations, chs. 14–16). The unknowns are the north/east coordinates of every non-control station. The system is linearized at compass-rule coordinates and iterated to convergence (typically 2–3 iterations).
  3. Compass rule (optional method). If your deliverable requires the traditional Bowditch distribution, switch the method next to the Adjust button — angles balanced equally, linear misclosure distributed in proportion to leg length. Statistical measures are only produced by least squares.
  4. Blunder hints. Before adjusting, the classic closure-geometry checks run automatically: a misclosure that parallels one leg points at that distance; a misclosure perpendicular to the chord from one station points at that angle. Flagged least-squares residuals offer one-click exclude & re-run to test an observation's effect (the excluded list appears above the results; restore any time).
  5. Weights. Angle σ = spec × √2 ÷ √sets (an angle is the difference of two pointings); distance σ = √(a² + (ppm·D)²) from the EDM spec. Centering errors are not modeled separately.
  6. Datum. Loop: the start station is held fixed and the first-leg azimuth is constrained (σ = 0.001″). Link: both control stations are held fixed and the mark angles act as azimuth observations. Control is never adjusted.

Reading the statistics

  • S₀ (standard deviation of unit weight) should be near 1.0 when the work matches the instrument settings. Persistently above 1 → the settings are optimistic or something is wrong; below 1 → the settings are conservative.
  • χ² test checks S₀ against its expected range at 95% confidence (two-tailed). A high failure usually means a blunder or optimistic settings — it does not necessarily mean bad fieldwork.
  • Standardized residuals (Baarda w-test, v⁄√qvv) compare each residual to its a-priori precision. Values over 3 flag likely blunders; a single blunder often drags neighboring observations over the line too, so chase the largest one first.
  • Error ellipses are reported at 95% confidence (2.4477 × the 1-σ axes, the large-redundancy value; exact small-sample F-scaling is not applied). Ellipse azimuths are clockwise from north.
  • Observations are treated as uncorrelated (diagonal weight matrix) — standard practice for traverse work.
  • Coordinate standard deviations are scaled by the a-posteriori S₀, matching common adjustment software output.

SPJ point extractor

The SPJ point extractor recovers point coordinates (names, N/E/elevation, lat/lon) from Trimble / TDS .spj data-collector job files — parsed entirely in your browser, so the file never leaves your machine. Filter and select the points you want, then download them as CSV in US survey feet and meters. Records the parser cannot read confidently are flagged rather than guessed; total-station raw observations (angles/distances) are not yet supported.

Validation

The engine reproduces the published worked examples of Ghilani, Adjustment Computations: Spatial Data Analysis (6th ed., Wiley 2017) — including the Example 16.1 link traverse, the Example 16.2 horizontal network with its deliberate outlier, and the Chapter 19 error-ellipse computations — in an automated test suite. Grid and ground are your responsibility: enter grid distances with grid azimuths, or ground with ground.