基于ARIMA模型的地球自转变化预报

Prediction of Earth Rotation Using ARIMA

准确预报地球自转变化对于精密定位、空间飞行器的跟踪与正常运行,具有重要的科学意义和实用价值。根据周日变化(UT1-UTC)和极移变化(PM)的特性,用最小二乘法,建立了适合于UT1-UTC和PM趋势项和周期项观测数据的拟合模型。对于UT1-UTC残差序列采用差分自回归移动平均(ARIMA)模型进行预报,对于PM残差采用季节性自回归移动平均(SARMA)模型进行预报。实例结果表明:我们的UT1-UTC预报结果比地球自转服务(IERS)产品好,而PM比IERS要差一些。当大气角动量(AAM)和海洋角动量(OAM)数据参与计算后,对UT1-UTC的预报有细微改善,对PM无改善。

Based on the spectra of the UT1 or polar motion （PM）, we fit the observed data to a combination of polynomial and sinusoids using least-squares techniques, and then predict by extrapolating using the coefficients estimated. We use both the observed data and the differences between the successive data in the least square fit. The residuals are handled using the ARIMA model. This way, the predictions are made in several steps. We have also attempted to incorporate the atmospheric （AAM） and oceanic angular momentum （OAM） in the prediction algorithm. The results are compared with the IERS predictions.