LOCAL ERRORS IN RELATION TO ANALYTICAL METHODS OF AIR TRIANGULATION |
| |
| Abstract: | AbstractIn view of the limitations of the use of plotting machines in air triangulation a detailed investigation of the practical possibilities of the analytical methods has become necessary, especially now that inexpensive means of reducing the labour of calculation is in course of realisation.In analytical methods it is practically possible to correct the measured photographic coordinates for determinable errors. Also, the use of a reseau as developed by the Ordnance Survey facilitates the reduction of several distortions. It is however argued and demonstrated that some errors of a local character remain without effective checking.A discussion of the nature of local errors shows that the number of points usually employed for relative setting is too small to be effective in disclosing them. An increase of the number of observed points is therefore suggested, and should be expected to improve the accuracy in the meanwhile.An assessment of the value of doubling the number of minor control points and of the improvement resulting from the use of reseau is made from a statistical analysis of data kindly provided by the Ordnance Survey.Comparison of 89 overlaps carrying one centimetre reseau and 100 overlaps without the reseau shows that the use of reseau reduced the root mean square error of evaluation of want of correspondence by 13 per cent., which just passes the 10 per cent. point of significance. The use of double minor control on the reseau overlaps gave an improvement which is about equal to that due to the use of the reseau. Doubling the number of minor control points is shown to increase the effective number of degrees of freedom from one to an average of four, which should reduce effectively the chances of large errors passing unnoticed through the stages of relative setting.Work on the frequency of local errors of various sources is in hand, and will probably be published in later issues of the Review. |
| |
| Keywords: | |
|
|
