Short-term earth orientation parameters predictions by combination of the least-squares, AR model and Kalman filter

► We employ a combination of least squares (LS), AR model and Kalman filter (LS+AR+Kalman) in short-term prediction of the EOP. ► The LS+AR+Kalman method works better than the LS+AR model. ► LS+AR+Kalman shows general better results compared to the EOP PCC. This study employs a combination of the least-squares, an autoregressive (AR) model and a Kalman filter (LS+AR+Kalman) in short-term prediction of the earth orientation parameters (the length-of-day (LOD), UT1-UTC and polar motion). Compared to least-squares and AR model (LS+AR), the combination of least-squares, AR model and Kalman filter performs better in the prediction of UT1-UTC and LOD, and shows a significant improvement in prediction of the polar motion.