You are here

Kevin Jardine's blog

The structure of the solar neighbourhood

Submitted by Kevin Jardine on 21 July, 2017 - 00:45
Enclosure diagram
An enclosure diagram showing the 419 OB star density peaks with 5 or more stars in the solar neighbourhood (within 650 pc or 2100 ly).

Sometimes less is more. The above image is an enclosure diagram showing the star density peaks in the solar neighbourhood and how they are contained within each other.

It was constructed by computing the OB star density isosurfaces for each integer value from 10% to 99% and maintaining a list of stars contained by each connected subregion.

An enclosure diagram lacks position or shape data, but reveals the star distribution and structure in a clear way. The circle size represents the number of stars in the region and the colour intensity the density.

Using the large version of the enclosure diagram here you can hover over each component to see its region label, name, and the number of stars it contains.

The names are based on clusters, associations or the brightest star contained by the region.

You can see that the solar neighbourhood contains four major dense OB star concentrations: Scorpius OB2, Vela OB2, the Orion Belt (Orion OB1) and the Perseus / Taurus dark cloud concentration that includes the Pleiades and the Perseus OB3 association. Less dense but still large concentrations include the three northern regions (ASCC 123, Cepheus OB6, and the Sulafat highway) as well as the Wishing Well region named after its core Wishing Well cluster (NGC 3532).

In April 2018, Gaia DR2 will be released with distances to more than a billion stars. Density isosurfaces and enclosure diagrams will have key roles to play in mapping this dataset and identifying the major regions within it.

Version 3 of the solar neighbourhood map

Submitted by Kevin Jardine on 20 July, 2017 - 01:34
Orion region
A detail of the 18% isosurface around Orion taken from the third version of the solar neighbourhood map. Oriented so that the direction to the galactic centre is at the top.

As I mentioned in my previous blog post, I have created a new version (v. 3) of the solar neighbourhood map. This uses colours taken from the 2MASS catalog and has simpler controls (dust is always turned on and the views always show a 70% bright star isosurface). You can select several different hot star isosurface densities as well as three different label schemes.

The brightest stars now have labels if you select the isosurface or stars label options.

The new dust overlay is taken from figure 3 (top) in this preprint:

Three-dimensional mapping of the local interstellar medium with composite data,
Capitanio, Letizia; Lallement, Rosine; Vergely, Jean Luc; Elyajouri, Meriem; Monreal-Ibero, Ana
eprint arXiv:1706.07711

You can read detailed documentation by clicking on the Help link at the upper right of the map system, which can be found here.

TGAS, 2MASS and the "Fingers of God"

Submitted by Kevin Jardine on 4 July, 2017 - 00:38
Side view
Density isosurfaces of major OB star concentrations in the solar neighbourhood . Oriented so that the direction to the galactic centre is at the top.

Gaia DR1, which includes the TGAS parallax data set, provides no colours for its stars. Colour data is essential in order to produce maps of the hot OB stars, which mark the young star formation regions in our galaxy.

Colours will be provided in the next Gaia release, but for now, the second version of my solar neighbourhood map takes its colour information from the Tycho-2 catalog, which makes sense because TGAS is itself based on stars from that catalog.

However, I noticed that many professional astronomers publishing on Gaia were using colours from the near-infrared 2MASS catalog instead. Eric Mamajek has put up an invaluable colour reference table, which converts colours from many catalogs into star temperatures and spectral types. Mamajek's table includes the 2MASS colours.

Using the Tycho-2 colours with my usual filters (err/plx < 0.2, absolute magnitude brighter than 1.5) extracts 3400 OB stars from TGAS. However, using the 2MASS colours extracts 5800 OB stars - more than a 70% increase!

I suspect 2MASS colours are more accurate because near infrared colours are less likely to be distorted by dust than visual light colours.

I am now working on a third version of my map that combines the 2MASS-derived hot star colours with the latest solar neighbourhood dust map. I expect the map will be done before the end of the month. However, I already have some interesting results.

Perhaps the most important result is a structure analysis of the density isosurfaces for the hot OB stars. What I've done is compute a density isosurface for each integer percentage from 10% to 99% as well as generate statistics and a list of contained stars for each connected region inside each isosurface.

There are many isosurfaces and thousands of connected regions within them, but not all are equally important.

If we start with the low density 10% isosurface and watch what happens to each connected region within it as the density increases, we see that each region fragments into smaller and denser regions. The lower density regions contain a nested sequence of higher density regions, much like Russian matryoshka dolls.

Most of the time, the fragments are small. But occasionally a large fragment breaks off. The enclosure diagram below shows each fragment with 20 or more stars and the less dense enclosing region that it fragments from. Hovering over a circle should show the name I have given to the region.

There are many interesting features visible in this enclosure chart. I'll look at some of these in a future post. For now, I want to show what happens when I map these large fragments. The map at the beginning of this article shows the fragments denser than 12% with at least 20 stars. Other than Orion X, which is from the

Bouy, H., and J. Alves. "Cosmography of OB stars in the solar neighbourhood." Astronomy & Astrophysics 584 (2015): A26.

article, the others are named after a contained bright star, cluster or association. The prominent Wishing Well region is named after the common name for its largest cluster, NGC 3532.

The map looks very promising, but it has one disappointing feature. Many of the elongated structures are oriented so that they point towards the Sun.

You can see this here:

Elongated structures like these are a common artefact in Milky Way maps and are sometimes jokingly called "fingers of God" because of the way the galaxy appears to point at our minor G2 class main sequence star. Often the cause is dust, which blocks our view in some places but has gaps that allow us to see in certain directions for a long distance. It is possible that dust is a cause here too, but it also seems likely that the problem is due to errors in parallax measurement.

A poster displayed last week at an European astronomy conference in Prague by Larreina, Alves, Bouy, observes that "Due to errors in parallax many [TGAS] structures appear elongated towards the Sun".

We can hope that with better calibration and more observations, these "fingers of God" will be eliminated or at least significantly reduced in Gaia DR2, due out in April 2018.

A box of stars

Submitted by Kevin Jardine on 22 June, 2017 - 09:23
Side view
Frame from animation showing higher density bright (green) and hot (blue) regions.

Here's a round up of some of my recent tweets about the solar neighbourhood map.

A link to downloadable 3D meshes:

A video showing me unbox a 3D print of the main region in the hot 17% isosurface:

Some still images of the 3D print:

Southern stars with Orion peninsula in the foreground, Per OB3 at left, Sco OB2 at top:

Southern stars. Sco OB2, M7 and LP Trianguli Australis region in foreground, NGC 3532 at left (notice the vertical stream of stars left of Orion!):

Northern stars, Cep OB6 and zeta Cephei complex in foreground, Sheliak highway in background:

(The Sheliak highway is a long stream of stars above the galactic plane that starts in the region surrounding Sheliak (beta Lyrae, at the left of this image) and then extends to the right.)

Here is a Blender animation combining higher density hot (blue) and bright (green) star density meshes. The sun is the dot at the rotation centre:

A topographic map of the solar neighbourhood (version 2)

Submitted by Kevin Jardine on 18 June, 2017 - 23:59
Bright and hot star clouds
This image shows dust (reddish brown), hot star clouds (blue) and bright star clouds (green) in the solar neighbourhood within 650 parsecs (2100 ly). The direction to the galactic centre is at the top.

My first map of the solar neighbourhood, released in February 2017, had a number of limitations and a couple of significant errors. This second attempt fills in some gaps and corrects the known errors.

This map version also includes 32 thousand "beacon" stars - bright stars contained in the dense star clouds within 650 parsecs or about 2100 light years. The region names in this version are based on star clusters or extremely bright stars contained within the region.

Filling in missing data

The TGAS data set released as part of Gaia DR1 is known to be missing many stars. Not only are stars missing in certain underscanned directions, but Gaia is missing high proper motion stars, many relatively bright stars, and stars that are very blue or very red. Unfortunately many of these stars are the ones that are needed to construct a map, especially within a few hundred parsecs.

In this map version I have used the older Hipparcos data set to add missing bright stars within about 300 parsecs. This is a useful stop gap until Gaia DR2 is released in April 2018.

Colour errors

In my previous map I used the Tycho-2 database to determine star colours as the Gaia-determined colours will not be available until DR2. However I did not consider the colour errors given in the Tycho-2 catalog and as a result, many of the stars I included in my "hot" star map are in fact of unknown temperature. In this version of the map, I have excluded all stars with a colour error greater than 0.1 and, further, at the suggestion of Gaia scientist Ronald Drimmel, I have excluded all stars with a relative magnitude dimmer than magnitude 11 as dimmer Tycho-2 stars are known to have dubious colour values.

Magnitude filtering

In my previous map I did not take into account the fact that the Tycho-2 catalog is itself very incomplete, especially for stars dimmer than relative magnitude 11. After I filtered out the stars dimmer than relative magnitude 11 to deal with the colour errors, I realised that this also imposed a limit on the absolute magnitudes of the stars I used to construct a map. Here is the graph showing the absolute magnitudes of stars with relative magnitude 11 out to various distances:

After some analysis I decided to make the second version of the map out to 650 parsecs. In order to ensure that my selection of stars was even reasonably complete out to that distance, a look at this chart showed that I needed to filter my list to stars brighter than absolute magnitude 1.5 (or so, making the limit slightly more restrictive than the chart allowed for some dust obscuration as well).


Adding missing stars from Hipparcos expanded my star list, but filtering out dim stars considerably restricted it. My new map is constructed using about 140 thousand bright stars (1.5 absolute magnitude or brighter) including about 3400 hot stars (colour index < -0.02, corresponding to O and B class stars).

This second map version looks quite different from the first version. The first version showed three separate star concentrations or stellar continents. The second version fills in the gaps between these continents. There are now two major star concentrations, which I have called the Northern and Southern stars. I will look at these star concentrations in more detail in a future blog post.

The map application

There is much else I want to show including 3D meshes, animations etc., but this blog post has gone on long enough so it is time for the link to the map application itself:

A topographic map of the solar neighbourhood (version 2)

I've added more configuration options to add different densities of bright and hot isosurfaces, turn on or off the dust clouds, and display different kinds of labels. You can read more about the map controls by clicking on the Help link at the bottom of the control area on the upper right.

Also note that at the highest zoom level you can hover over individual stars for names and distances above and below the galactic plane, and click on individual stars for more details. Click on the Help link below the map controls for more information. The JSON loading of the star data may slow down less powerful devices - I am looking into speeding that up as well as tweaking the display to work better with mobile devices.


I am confident that this map is more accurate than the first version because the hot star densities now correspond closely to the Hipparcos map shown in the paper:

Bouy, H., and J. Alves. "Cosmography of OB stars in the solar neighbourhood." Astronomy & Astrophysics 584 (2015): A26.

Compare the dense hot star clouds shown in this image from the second version of my map:

with the Bouy map here:

and the 3D display here:

Many of the structures line up after a 90 degree rotation.

Having said that, I can see some more improvement possibilities so there may be a third map version before Gaia DR2 in April 2018 allows for a much better map of a large part of the galaxy. Exciting times ahead!


Subscribe to RSS - Kevin Jardine's blog