The study aims to take an advantage of the systematic combination of statistical procedures namely Principal Component analysis (PCA), Linear Model to develop a local climatological model for ionospheric Total Electron Content (TEC) irregularities. Further, Auto Regressive Moving Average (ARMA) model, Neural Network (NN) model are used to forecast the ionospheric irregularities over Bengaluru region, India. Bengaluru International GNSS Service (IGS) station dataset (geographic lat.- 13.02°N, long.77.57°E; geomagnetic latitude: 4.4°N) of an 8-year period has been used to implement the proposed algorithm. Retrieval of the main components of the times series using Principal Component Analysis, identifying and analyzing the trends and cycles in the PCA residual to model the un-modelled irregularities using Linear model and finally fitting the residual series using ARMA/NN models to forecast the irregularity es has delivered successful results. The ionospheric TEC irregularities, measured in TECU/hour, are investigated during the 24th solar cycle ascending and descending phases. The MAE and RMSE values are used for the validation of the proposed models in forecasting the ionospheric TEC irregularities during both geomagnetic quiet and storm days (12–14 October 2016). It is observed that MAE value of NN model is 0.5 TECU/hour during geomagnetic quiet period whereas it is 0.98 TECU/hour during geomagnetic disturbed period. Moreover, the RMSE value of ARMA model is 0.73 TECU/hour and 0.67 TECU/hour for NN model which reveals that NN model is comparatively good in forecasting the ionospheric TEC irregularities during geomagnetic quiet period. The proposed model can be useful to develop an ionospheric irregularity climatological tool for low latitude regions.

中文翻译：

---

低纬度 GNSS 站电离层 TEC 不规则性的建模和预测

该研究旨在利用统计程序的系统组合，即主成分分析 (PCA)、线性模型，为电离层总电子含量 (TEC) 不规则性开发局部气候模型。此外，自回归移动平均（ARMA）模型、神经网络（NN）模型用于预测印度班加罗尔地区电离层的不规则性。8 年周期的班加罗尔国际 GNSS 服务 (IGS) 站数据集（地理纬度 - 13.02°N，经度 77.57°E；地磁纬度：4.4°N）已用于实现所提出的算法。使用主成分分析检索时间序列的主要成分，识别和分析 PCA 残差中的趋势和周期以使用线性模型对未建模的不规则性进行建模，最后使用 ARMA/NN 模型拟合残差序列以预测不规则性 es 已取得成功的结果。在第 24 个太阳周期的上升和下降阶段，研究了电离层 TEC 不规则性，以 TECU/小时为单位进行测量。MAE 和 RMSE 值用于验证提议的模型，用于预测地磁平静日和风暴日（2016 年 10 月 12-14 日）电离层 TEC 不规则性。观察到NN模型的MAE值在地磁安静期为0.5 TECU/小时，而在地磁扰动期为0.98 TECU/小时。此外，ARMA 模型的 RMSE 值为 0.73 TECU/小时和 0。NN 模型为 67 TECU/hour，说明 NN 模型在预测地磁静默期电离层 TEC 不规则性方面较好。所提出的模型可用于开发低纬度地区的电离层不规则气候学工具。