Comparison Between Gross Errors Detection Methods in Surveying Measurements

Abstract

The least squares estimation method is commonly used to process measurements. In practice, redundant measurements are carried out to ensure quality control and to check for errors that could affect the results. Therefore, an insurance of the quality of these measurements is an important issue. Measurement errors of collected data have different levels of influence due to their number, measured accuracy and redundancy. The aim of this paper is to examine the detection of gross error capabilities in vertical control networks using three methods; Global Test, Data Snooping and Tau Test to compare the effectiveness of these three methods. With the least squares’ method, if there are gross errors in the observations, the sizes of the corresponding residuals may not always be larger than for other residuals that do not have gross errors. This makes it difficult to find (detect) it. Therefore, it is not certain that serious errors should be detected by just examining the magnitudes of the residuals alone. These methods are used in conjunction with developed programs to calculate critical values for the distributions (in real time) rather than look for these in statistical tables. The main conclusion reached is that the tau (τ) statistic is the most sensitive to the presence gross error detection; therefore, it is the one recommended to be used in gross error detection.