Аннотации:
© 2020 International Multidisciplinary Scientific Geoconference. All rights reserved. This work focuses on the possibility of practical application of two approaches - classic least square method (LSM) and Huber's M-estimator (HME) method - to the analysis of time series of celestial objects' measurements. Using those methods, the values of dynamic selenophysical parameters were produced from the processing of observations of Mösting A lunar crater in relation to the lunar limb. The reason for the use of the two approaches - LSM and HME - at observations processing is that LSM in some cases does not allow eliminating erroneous observations from the analysis. The presence of such observations might lead to unreliable estimates of desired parameters. In this paper, we therefore applied HME method that allows analyzing long-term time series of observations and assessing desired parameters most plausible with considering the impact from the erroneous observations on final results. An advantage of HME method compared to the other ones of robust analysis is that it is simple and reliable. HME method had previously shown good results at processing of observations that contain many erroneous measurements. The use of the two methods - LSM and HME - is justified by the fact that if the errors are normally distributed, parameters' estimates produced from LSM and HME should coincide. The difference in the estimates points at the presence of erroneous observations. In this paper, we obtained various estimates when using LSM and HME. We may therefore conclude that the observations are not equally accurate, and there are abnormal deviations from the normal distribution of errors. In our case, the use of HME method provides more accurate estimates of the desired parameters compared to LSM. As a result, the values of selenographic parameters were obtained with the highest reliability of their estimates.
