Fluctuating corrections with tie lines
Besides the correction of systematic shifts, tie lines can be used for applying corrections that change over time (fluctuating corrections, drifts). This is essential for correcting local drifts of the trajectory solution which occur, for example, if there is a failure of GPS signal during a flight or drive. Fluctuating corrections are the most important part of improving the accuracy of mobile laser data, but they are useful for airborne data as well.
If no control measurements are involved, the software computes a guessed location for the tie line feature which corresponds to the average of the feature location between the laser data strips. If the trajectory solution provides estimates of the positional accuracy for each drive pass, these estimates translate to weight factors when computing the average location. This results in lower accuracy strips getting a bigger correction and better accuracy paths getting a smaller correction.
If a control measurement is used, the observed correction is the difference between the tie line feature location in the laser data and the location of the control measurement.
The actual correction for the laser data is applied as a correction curve which changes over time. The correction curve is built by computing a correction for each tie line observation. The correction is a linear interpolation between two consecutive observations.
Exaggerated xyz correction vectors for one strip.
The user can specify how much the correction curve is smoothed. The smoothing of the correction curve involves the accuracy estimates for trajectory positions. Bigger corrections values are applied for positions where the accuracy estimate is worse. A factor determines, how fast the corrections curve changes can be. A small factor results in a smoother curve. This should be used if the tie lines are not so good and may contain outliers. A bigger factor results in a less smooth curve and the single tie line observations get more influence in the final solution. This should be used for good tie lines.
If no smoothing is applied, each tie line observation fully effects the final correction values. This is recommended for checking the tie lines and finding out erroneous tie lines. Otherwise, a smoother correction curve is less sensible to remaining inaccurate tie line observations. No smoothing may also be used if correction values are computed based on accurate tie points collected from ground control points for which the exact xyz coordinate values are known.
