We examine systematic differences between topside electron density measurements and different topside model formulations including ground-based α-Chapman extrapolated topside electron density profiles from auto scaled ionograms, International Reference Ionosphere Model (IRI-2016) NeQuick topside estimations and a recently improved NeQuick (Corrected NeQuick) topside formulation. The selected topside electron density measurements considered were taken, from radio occultation electron density profiles on board low Earth orbit (LEO) satellites from the COSMIC/FORMOSAT-3 mission, in the vicinity of digisonde stations on a global scale. A subset of these radio occultation profiles, with matched (within 5%) peak NmF2 and hmF2 characteristics is also exploited to focus the comparison to a high quality validation dataset. The comparison shows that α-Chapman and Corrected NeQuick underestimate, whereas IRI-NeQuick overestimates COSMIC topside electron density observations. The key parameter g which controls the change of scale height w.r.t. altitude near the F region peak is optimised to a value of 0.15 (compared to a currently adopted value of 0.125). The Corrected NeQuick topside formulation using the optimised g value of 0.15 (represented as Newg) outperforms all other topside formulations.