IJECT Vol. 5, IssuE spl - 3, Jan - MarCh 2014 www.iject.org InternatIonal Journal of electronIcs & communIcatIon technology 27 ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print) Forecasting of Ionospheric TEC Values Using Holt-Winters Method Over Low Latitude Station 1 Dr. D.Venkata Ratnam, 2 G.Sivavaraprasad, 3 Harshali Mane 1,2,3 Dept. of ECE, KL University (Koneru Lakshmaiah Education Foundation) Vaddeswaram, Guntur, AP, India Abstract GPS is a satellite based navigation system, provides positional and timing information for an aircraft landing and missile guidance applications. The forecasting methods are rarely perfect but precise and accurate in short-term. The precision levels of GPS are challenged by variations in Total Electron Content (TEC) of ionosphere. The variations in TEC severely affect the GPS communication over lower latitudes. The ionospheric delay, positioning error and loss of lock are few consequences. Hence predicting and forecasting of TEC is more important. Here, we propose an ionospheric forecasting model using Holt-Winters method for estimating future TEC values. Dual frequency GPS receiver data of K L University, Vaddeswaram station (16.31° N, 80.37° E), India is considered for the analysis. GPS data (1st - 8th March, 2013), before the winter equinox (March 21, 2013) is extracted and VTEC values are estimated. From the estimated values 9th&10th, March, 2013 TEC values are forecasted. The forecasted results and original values are compared. It is found that additive model is 9.79 % more accurate than multiplicative model of Holt-Winter method. Results indicate that Holt-Winters additive model would be useful for developing an early warning ionospheric disturbance indicator for an effective navigation system at lower latitude regions. Keywords GPS, Ionospheric Delay, Forecasting Model, TEC, Holt-Winters Method, Early Warning Ionospheric Disturbance Indicator I. Introduction The Global Positioning System (GPS) is a space –based radio- satellite navigation system. GPS has been very popular in military and civilian applications by providing position and timing information [1]. GPS is a complex system based on a constellation of satellites transmitting navigational information. The GPS signals faces User Equivalent Range Errors (UERE), includes User Range Error (URE) such as atmospheric delays (Ionospheric and Tropospheric delays), ephemeris and clock errors and User Equipment Error (UEE) such as receiver noise and multi path error [1]. There is a major probability of errors in broadcasted GPS signals that propagate through the ionosphere and experience several effects proportional to the total number of electrons presented in its path from GPS satellite to receiver. The Ionosphere prevalent 50-1000 km is most predominantly disturbed by time of day, location, season and solar activities, diurnals and storm [2]. The TEC (total number of electrons per unit square meter) is the restricting the positioning and tracking capabilities of GPS as a challenger. TEC is the function of solar radiation and its characteristics depends on seasonal, periodical and sudden atmospheric disturbances like storms, solar activities, Earth gravity and etc. [3]. That means the ionospheric delays or positioning errors of GPS are the functions of TEC. The properties of Ionosphere contain generally repeated trends and seasonal patterns. GPS technique given open chance to measure TEC distributions on regular basis. The low latitude ionospheric behaviour is erratic and severe so more investigation needs to be carried out into account in developing suitable ionospheric models [4-5]. Therefore, forecasting of ionospheric TEC is advantageous for mitigation of ionospheric error and extremely useful for the progress of communication, navigation, surveying and early alert warning for developing space weather alert system. Ionospheric prediction models such as Bent model, International Reference Ionosphere (IRI) model, AR model, ARMA model are available [2]. But these models may not be appropriate as ionosphere TEC existed with seasonally repeated trend patterns. In this paper, an alternative statistical method known as Holt-Winters is implemented. The Holt-Winters method effectively forecast TEC values that helps in forecasting ionospheric delay [6-7]. II. Holt-Winter Method The Holt–Winters method is a statistical exponential smoothing method, used for prediction of future values based on past values of time series data. Holt-Winters method is used to capture seasonality. The Holt-Winters seasonal method comprises the forecast equation and three smoothing equations Lt, bt, St. It has three Smoothing parameters: alpha(α), beta(β), and gamma(γ), for the estimates of the level, slope b of the trend component, and the seasonal component, respectively, at the current time point [2]. The parameters alpha(α), beta(β) and gamma(γ), all have values between 0 and 1, and values that are close to 0 means that relatively little weight is aligned on the most recent observations when making forecasts of future values [8]. The Exponential smoothing is relatively a simple method of short term forecasting for a time series. Time series of TEC data yt, is defned as a collection of data in equal interval of time over a period of time given by [9], t, Where, t=0,1,2,...t (1) Consider is the future TEC data value to be predicted from the original TEC data. . As the ionosphere TEC is having the anomaly characteristics like repeated trends and seasonal patterns, the Holt-Winters method is most suitable for forecasting the TEC time series [2]. It is a very popular method to forecast the time series of TEC of ionosphere which are dependent on diurnal and seasonal changes, solar cycle, geographical location and geomagnetic feld [9]. The time series patterns of TEC will be decomposed in Holt-Winters exponential smoothing method to estimate the level, slope and seasonal component at the current time point. The Holt-Winter method is correct forecasting method for the data infuenced by seasonal and trend changes phenomenon hence it has two types of models they are (i) Additive model and (ii) Multiplicative model [2]. When seasonal variations are roughly constant through the series the additive method is preferred, while the multiplicative method is preferred when the seasonal variations are changing proportional to the level of the series. A. Additive Model Time series that can be described using an additive model with increasing or decreasing trend and seasonality. The time series