基于陆基GNSS的电离层时间序列预报研究

为讨论不同时间序列模型对电离层垂直总电子含量（VTEC）的预报效果，在平静电离层条件下，采用载波相位平滑伪距法解算单站上空的电离层VTEC值，分别利用自回归积分滑动平均模型（ARIMA）与Holt-Winters指数平滑模型进行逐站建模，通过时长为9 d的样本序列实现3 d预报，并对预报值进行系统评估.结果表明，时间序列模型能够较好地反映预报期内的电离层VTEC变化情况，均方根误差均值不超5 TECU.此外，Holt-Winters乘法模型的预报值偏差最大，加法模型次之，ARIMA模型在11个测站的相对精度都高于Holt-Winters指数平滑模型，且其均方根误差峰值最小，具有最高的预报精度.

Ionospheric time series forecast based on land-based GNSS

In order to compare the forecast performance of different time series models,the carrier phase smoothing pseudo range method is used to calculate the ionospheric Vertical Total Electron Content (VTEC) over a single station under the condition of calm ionosphere.The Auto Regressive Integrated Moving Average (ARIMA) model and Holt Winters exponential smoothing model are used for station-by-station modeling.A 3-day forecast is achieved through a 9-day sample sequence,and the forecast values are systematically evaluated.The results show that the time series models can well reflect the change in ionospheric VTEC during the forecast period,and the mean square root errors are not more than 5 TECU.In addition,the Holt-Winters multiplicative model has the largest deviation in the forecast value,followed by the additive model.The ARIMA model has higher relative accuracy at the 11 stations than the Holt-Winters exponential smoothing model,and has the smallest root mean square error peak as well as the highest forecast accuracy.