Abstract:In order to choose the most robust estimation methods, observations are simulated by generalized Gaussian distribution (GGD),which can perfectly describe the real distribution of observations. Leveling and trilateration networks with equal weighted and independent observations are taken as examples, and simulation experiments are used to compare the robustness of 12 commonly used robust estimation methods. Our results indicate that when observations obey GGD, the L1, Danish and German-McClure methods, and the IGGⅢ scheme are relatively more efficient robust estimation methods in leveling and trilateration networks with equal weighted and independent observations; this is similar to results in simulation experiments.