NE2025: An Updated Electron Density Model for the Galactic Interstellar Medium

Free electrons in the Galactic interstellar medium (ISM) disperse and scatter coherent radio waves, by amounts that depend on the distance to the radio source. Models of the Galactic electron density are thus widely used to predict distances and scattering of compact radio sources (including pulsars, fast radio bursts (FRBs), and long-period transients), in addition to mitigating ISM foregrounds in Galactic and extragalactic studies. We use a sample of 171 precise pulsar distances, based entirely on parallaxes and globular cluster associations, as well as scattering measurements of 568 pulsars, active galactic nuclei, and masers, to update the NE2001 Galactic electron density model. We refit the thick and thin disks and three of the spiral arms. The new parameters for these large-scale components significantly repartition free electrons between the thick disk and spiral arms, thereby correcting NE2001’s systematic underestimation of pulsar distance and scattering. Sightlines with excessive dispersion and scattering are used to identify new clumps that are added to the model, in addition to refining clumps that were already included (e.g., Cygnus, Vela, and Gum). The Galactic Center component is revised, yielding scattering time predictions that are $10^3$ times smaller than the Galactic Center in NE2001. The updated model, NE2025, provides a factor of $20\times$ improvement in median distance prediction accuracy and $100%$ median improvement in scattering predictions based on DM, relative to NE2001. There is a $15\times$ improvement in median distance prediction accuracy relative to YMW16. NE2025 is available on Github and the Python Package Interface.

💡 Research Summary

The paper presents NE2025, an updated Galactic electron‑density model that supersedes the widely used NE2001. The authors assemble a high‑quality calibration set consisting of 171 independent pulsar distances—126 very precise parallaxes (VLBI, timing, and Gaia optical) with fractional uncertainties below 25 % and 45 globular‑cluster distances derived from Gaia‑based measurements with < 7 % uncertainties. This sample, especially rich at high Galactic latitudes (|b| > 20°), provides a far more complete probe of the thick‑disk component than was available for NE2001, which relied heavily on H I kinematic distances and only 14 parallax measurements. In addition, the authors compile 568 scattering measurements (pulse broadening times, angular broadening, scintillation bandwidths) from pulsars, active galactic nuclei, and masers, giving a robust handle on the turbulent component of the ionised medium.

Using this dataset, the authors first demonstrate that NE2001 systematically over‑estimates the free‑electron column density along many sightlines, leading to under‑estimated distances and over‑predicted scattering. The discrepancy is most pronounced for high‑latitude lines of sight, implicating the thick‑disk as the primary culprit. Consequently, the thick‑disk scale height and mid‑plane density are refitted, shifting a substantial fraction of the electron budget from the thick disk into the thin disk and spiral arms. Three major spiral arms (Perseus, Carina–Sagittarius, and Orion) are also re‑parameterised (radial location, width, and density contrast) to better match the new distance and DM data.

The scattering component is updated by fitting the Kolmogorov turbulence parameters to the 568 scattering observables. A dramatic revision concerns the Galactic‑Center (GC) component: its electron density and turbulence strength are reduced by roughly three orders of magnitude, bringing predicted scattering times from the previously unrealistic tens of seconds down to the millisecond regime observed for sources near Sgr A*.

To address outliers where the model severely under‑predicts DM or scattering, the authors identify twelve pulsars whose lines of sight intersect known H II regions. Five of these are added as new high‑density “clumps” (e.g., in Cygnus and a newly defined Vela extension), while several previously included clumps and voids are refined or removed. This targeted addition of discrete structures eliminates the most egregious residuals without over‑complicating the global model.

Performance metrics show a 20‑fold improvement in median distance prediction accuracy relative to NE2001 and a 15‑fold improvement relative to the more recent YMW16 model. Scattering predictions based on DM improve by 100 % (i.e., the median error is halved). The authors provide both a Fortran implementation and a Python package (NE2025p) released on GitHub, ensuring immediate accessibility for the community.

The paper acknowledges remaining limitations: the southern sky remains under‑sampled in parallaxes, leading to larger uncertainties for longitudes ≲ −110°, and the clump/void parameters are still tuned manually rather than derived from a fully Bayesian hierarchical model. The authors suggest that forthcoming large‑scale surveys (e.g., SKA, CHIME, FAST) and continued VLBI parallax campaigns will enable a next‑generation, probabilistic electron‑density model.

In summary, NE2025 leverages a substantially expanded, high‑precision distance and scattering dataset to recalibrate the large‑scale components of the Galactic electron‑density distribution, introduces refined discrete structures, and dramatically improves distance and scattering predictions for pulsars, FRBs, and other compact radio sources. This advancement will benefit a broad range of astrophysical applications, from pulsar timing arrays and FRB localisation to Galactic foreground mitigation in cosmic microwave background studies and searches for extraterrestrial intelligence.