Ionosphere electron density models - Present trend and validation issues

Terminology regarding "ionosphere electron density models" will be reviewed. Differences between "model", meaning a theoretical formulation of the electron density distribution essentially in 3D, and "mathematical representation of experimental data" (often called model) like the total electron content (TEC) global maps obtained from experimental data and mapping techniques, will be clarified. Possible conceptual errors linked to misuse of terms will be indicated. Both theoretical electron density models based on physical first principles and empirical models and their utilization from different perspectives will be mentioned and briefly discussed. These perspectives range from that of the GNSS community to the one of the geospace research community. Among the models to be mentioned, the last versions of the empirical models IRI and NeQuick and their use will be described in more detail. Their limitations and potential unsuitable conclusions that can be derived from the improper use of model drivers will be analyzed. The evolution from models designed to give a "climatelike" description of the ionosphere - like IRI and NeQuick - to the present modeling trend that is centered on an ionosphere "weather-like" specification is analyzed. The use of data ingestion or assimilation techniques into background climatic models to obtain real-time or near-real-time specifications of the electron density in the ionosphere will be described. The impact of uncertainties in the data used for ingestion or assimilation will be mentioned and exemplified. Climatic models used for assessment studies, forecasting purposes or background models for assimilation or data ingestion need to be based mostly on statistical representations of ionosphere characteristics like maximum electron density or TEC given by median or mean values. These values are considered "representative" or "nominal" values of such characteristics. The need of a better definition of "representative" or "nominal" ionosp- ere conditions will be discussed making use of vertical TEC, represented by global maps of this ionosphere characteristic particularly in critical geographic regions like the low or high latitudes. In particular, it is found that median values cannot be often considered " representative" values. This is particularly important in the geographical areas dominated by a highly variable behavior due to the complex dynamics of the Equatorial Anomaly of the ionosphere. More specifically this is due to the large dayto-day variability of the position and strength of the low geomagnetic latitudes electron density peaks and corresponding TEC at a given time of the day. In addition, because the variability cannot be associated only to specific geo-heliophysic conditions, it is not appropriate to assume particular days behavior as representative behavior for a given month or period. Possible solutions will be indicated to revisit the concepts of "representative" or "nominal" conditions of ionosphere characteristics by a more appropriate way of handling the data. Another problem related to ionosphere modeling is the validation of the models with experimental data. The results obtained by comparing modeled with experimentally obtained data are affected by two different causes. The first being the limited ability of the model itself to reproduce the real conditions found experimentally, and the second is the presence of uncertainties in the data themselves. This latter aspect of the modeling efforts will be discussed. The impact on validation of the uncertainties found in the scaling, calibration and interpretation of the experimental data will be outlined. These uncertainties are exemplified in the case of TEC calibration but also to some extent in ionosonde data scaling.