Gross Errors Detection for the African Gravity Database

To establish the gravity database for the African geoid, it is needed to remove the blunders from the available gravity data set. As the available gravity data for Africa is very limited, including large data gaps, the gross errors detection technique should be smart enough to eliminate only the real blunders. A smart gross error detection technique has been adopted. It is based on the least squares prediction algorithm. The technique works first to estimate the gravity value at the data station using other values than the current data point. It thus compares the estimated value to the data value for possible blunder detection. Hence the technique measures the influence of removing the data value of a current point on the neighbourhood stations. Only if the value of a certain station proves to be blunder, it is then removed from the data base. Another effective technique to estimate the blunder in the gravity database of Africa is designed using the Artificial Intelligence. The results of both gross errors detection techniques are compared and analyzed in order to give a proper judge on both algorithms.