[G02-P] Poster

[G02-P-08] Gross-Error Detection using Artificial Neural Networks for the Gravity Database in Egypt

^○Hussein Abd-Elmotaal, Mostafa Ashry (Minia University, Minia, Egypt)

The paper attempts to detect the gross-errors in the Egyptian gravity database using a technique based on the artificial neural networks. The gravity database of Egypt has been built from different sources for data measured at different decades with, sometimes, lack of precision information. The gravity measurements may also have contained gross-errors (biases) due to faulty sensors. Hence, it is important to detect the gross-errors before using this database for geodetic applications. The objective of the gross-error detection scheme is to identify the measurements that possibly contain gross-errors so that they can be removed. The total number of gravity data in Egypt is too less compared with Egypt's surface area. Accordingly, the gross-error detection technique should be smart enough to eliminate only the real blunders. A gross-error detection technique using artificial neural networks is developed. The artificial neural networks constitute the input layer, one or more hidden layers and output layer. The input units represent the input data, whereas the output units represent the output data. These hidden neurons enable the network to learn complex tasks by extracting progressively more meaningful features from the input patterns. The developed technique has been applied to the gravity database in Egypt, and a number of points were found to have blunders. The results are given and extensively discussed.