Choice of hydrogen masers rate prediction model

Forecasting atomic clock rates is an important part of time scale algorithm. The prediction model and observation interval are optimized using characteristics of prediction accuracy - the systematic predicting error and the standard deviation of the predicting error. Using various pairs of frequency measures as an example, it is shown that complicating the prediction model, on the one hand, reduces the systematic predicting error, and on the other hand, increases the standard deviation of the predicting error due to an increase in the number of free parameters in the model. For various pairs of frequency measures from State Time and Frequency Standard GET 1-2022, involved in the transfer of frequency units from the primary frequency standard to frequency and time measures, recommendations for choosing a forecasting model and the duration of the observation interval are given.

time scale, units of time and frequency, hydrogen maser time and frequency standards, prediction error