Time dependence of noise characteristics in continuous GPS observations from East Africa

A noise model for the regional continuous GPS (cGPS) timeseries in East Africa (Ethiopia and Eritrea) was computed using the maximum likelihood estimation (MLE) method. Using this method and assigning different noise models for each cGPS site and each component (north, east and vertical) may bias the noise level of the velocity solutions due to the non-uniformity of the length of the timeseries. Within the whole regional network, the length of the timeseries varies from one to seven years. We compute a preferred regional noise model for the whole data sets using a stacked maximum likelihood values for the different power - law indexes (between −2 and 0 with a time step of 0.1), presuming that there is only one noise model that exists in the regional cGPS timeseries. Therefore, a single power - law index (flicker plus white noise) was assigned for the whole regional network irrespective of the length of the timeseries. This approach is more robust and “realistic” to determine the noise characteristics of the regional GPS network.