The ionosphere layer of atmosphere environment is a highly variable media that performs significant weather variations with altitude, latitude, longitude, universal time, solar cycle, season, and geomagnetic activity. Therefore, ionosphere modeling and determining the total electron content (TEC) play an important role to know this layer of atmosphere specifications and control its effects on human activities. Different kinds of ionospheric models are widely used to monitor the changes in ionosphere in which single layer model (SLM) of ionosphere or TEC model has been always interesting for researchers. Complications of physical models for major of users, low accuracy of numerical (empirical) models for precise applications, the 24 hour delay in IGS daily global ionospheric map (GIM) propagation and the precise ionospheric information necessity in real-time and near real-time applications have been the reasons of development of new ionospheric models which is known as data assimilation models. These models combine measurements from observing system with the information obtained from background model trough the data assimilation technique. Assimilation algorithm involves a forecast step, in which a previous estimate of the state is evolved forward to the time of the observation, and an update or analysis step, where the evolved estimate of the state is updated using information from the observations. The outputs of assimilated models have parameters closer to the observations. The accuracy of the reconstructed ionosphere depends on the amount of assimilated data, the diversity of the data types and the quality of the data. Assimilated data may have different sources such as GPS slant TEC, in situ electron densities, electron density profiles (EDPs) from ground-based radars and ionosonde data in ionosphere data assimilation.

In this study, precise TEC derived from dual frequency GPS observations are assimilated in to an international reference ionosphere (IRI) numerical model in analysis and forecast steps of assimilation. Kalman filter is used to increase the accuracy of IRI extracted TEC in analysis step and Gauss-Marcov Kalman filter (GM-KF) is used to predict TEC in forecast step for real-time and near real-time applications. Observations of 40 stations of Iranian permanent GPS Network (IPGN) in May 03, 2016 are used to extract precise VTEC for assimilation in IRI model. The GPS observed VTEC are compared with TEC form IRI model, TEC from IGS GIM and assimilated IRI TEC in analysis step of assimilation. The rote mean square (RMS) of discrepancy between GPS VTEC and IRI TEC are reduced from 9.8 TECU to 1.47 TECU at t=10UT, from 3.16 TECU to 0.98 TECU at t=14 UT and from 4.59 TECU to 1.39 TECU at t=18UT, after assimilation. Comparing the IGS GIM and assimilated IRI TEC with GPS observed VTEC indicate that the assimilated model is more accurate than GIM in Iran region. The GPS observed VTEC are also compared with TEC form IRI model, TEC from IGS GIM and assimilated IRI TEC in prediction step of assimilation. This comparison shows 90% improvement in assimilated TEC respect to IRI TEC at t=10 UT for Dt=0.5, 1 hour prediction time intervals. This improvement at t=14, 18 UT is more than 50% for Dt=0.5, 1, 2 hour prediction time intervals. By increasing the prediction time interval to Dt=5 hour, the assimilation accuracy tends to IRI model. Therefore the assimilated model has a good accuracy for real-time and near real-time applications.