A reassessment of the PRIMO recommendations for adjustments to mid-latitude ionospheric models

In the late 1990s, in response to the realization that ionospheric physical models tended to underestimate the dayside peak F-region electron density (NmF2) by about a factor of 2, a group of modelers convened to find out why. The project was dubbed PRIMO, standing for Problems Relating to Ionospheric Models and Observations. Five ionospheric models were employed in the original study, including the Utah State University Time Dependent Ionospheric Model (TDIM), which is the focus of the present study. No physics-based explanation was put forward for the models' shortcomings, but there was a recommendation that three adjustments be made within the models: 1) The inclusion of a Burnside factor of 1.7 for the diffusion coefficients; 2) that the branching ratio of O+ be changed from 0.38 to 0.25; and 3) that the dayside ion production rates be scaled upward to account for ionization by secondary photons. The PRIMO recommendations were dutifully included in our TDIM model at Utah State University, though as time went on, and particularly while modeling the ionosphere during the International Polar Year (2007), it became clear that the PRIMO adjustments sometimes caused the model to produce excessively high dayside electron densities. As the original PRIMO study [Anderson et al, 1998] was based upon model/observation comparison over a very limited set of observations from just one station (Millstone Hill, Massachusetts), we have expanded the range of the study, taking advantage of resources that were not available 12 years ago, most notably the NGDC SPIDR Internet data base, and faster computers for running large numbers of simulations with the TDIM model. We look at ionosonde measurements of the peak dayside electron densities at mid-latitudes around the world, across the full range of seasons and solar cycles, as well as levels of geomagnetic activity, in order to determine at which times the PRIMO adjustments should be included in the model, and when it is best not to include them. It remains to be seen whether these same criteria for including or not including the PRIMO adjustments will produce the same improvements in the other ionospheric models that were used in the original study in 1998. This study highlights the great difficulty in modeling, of identifying which aspect of a model (ie, theory, boundary conditions, input, numerics, etc) is problematic, especially when there is ambiguity in the ways in which the model can be adjusted to provide improved agreement with observations. ------------------------- Reference: Anderson, D. N., et al. (1998), Intercomparison of physical models and observations of the ionosphere, J. Geophys. Res., 103(A2), 2179-2192, doi:10.1029/97JA02872.;