``The Hidden Realm of Cold Interstellar Hydrogen''

S. J. Gibson, A. R. Taylor (University of Calgary);
C. M. Brunt, P. E. Dewdney, & L. A. Higgs (HIA)

2001, 32nd Annual Meeting of the Canadian Astronomical Society, 73 [#P-066]

Electronic Poster Contents


NOTE: This electronic poster contains more material than the paper poster given at the CASCA meeting. I have added a few extra images from my talk at the Green Bank IGPS meeting, with explanatory captions. I have also changed the figure order from the paper version.


Though a major constituent of the interstellar medium, cold atomic gas, with T < ~100 K, is elusive. Maps of 21cm emission are dominated by warm H I, and most observations of H I absorption against continuum sources are limited to discrete points. However, H I self-absorption (HISA) against warmer background H I can give a better view of the structure and distribution of cold H I clouds in the Galaxy.

A systematic HISA study of cold Galactic H I requires broad angular coverage to remain unbiased, as well as high angular resolution to detect small-scale features which might otherwise be washed out. Our investigation is the first to employ wide-field synthesis imaging to these ends. We are using Canadian Galactic Plane Survey (Taylor et al. 2001) maps taken with the DRAO Synthesis Telescope. Our CGPS images have ~ 1' resolution with 0.8 km/s velocity channels over the region [147.3° > l > 74.2°, -3.6° < b < +5.6°].

This poster gives preliminary results for a full-fledged analysis of the gas properties and distribution of HISA features in the CGPS. We employ several automated algorithms to identify self-absorption within the H I data, estimate the background brightness being absorbed, and compute its physical properties from a limited set of assumptions. Figures 1-3 show some of the HISA mapped to date, while Figures 4-6 show measured and derived HISA gas properties and an analysis of the relation between HISA and CO emission.


A simple radiative transfer model representing a HISA cloud with foreground and background H I emission and background continuum emission is described by the expression

where TON and TOFF are observed brightnesses on and off the HISA feature, TS is the spin or excitation temperature of the HISA gas, tau is its optical depth, TC is the continuum intensity, and p is the fraction of H I emission lying behind the HISA feature. We measure TON, TOFF, and TC, and assume a likely value for p, but TS and tau remain unknown. To constrain these two variables, we make use of line integral and ideal gas relations to derive a second equation

which gives TS in terms of the line center opacity tau0, linewidth delta-v, the physical thickness of the HISA feature along the sightline delta-s, and the partial pressure of the atomic gas, P fn. With reasonable values applied to these new parameters, TS and tau can be obtained by solving the two equations together (see Gibson et al. 2000 ApJ 540, 851 for details).

For this poster, we assume p = 1, the most favorable value for seeing HISA, and we determine delta-s from estimated characteristic angular scales of features and likely distances based on velocity and a simple spiral arm model (e.g., gas near L=140, V=-40 is in the Perseus arm, with a distance of ~ 2 kpc, for which 1 arcminute corresponds to 0.6 pc). We use a canonical ISM pressure of 4000 cm-3 K and consider two extreme values for fn of 1.0 and 0.01.



Larger versions of each image below are available via links.

Figure 1: HISA Survey Map (l,v)

Longitude-velocity projection of the HISA currently mapped over the full CGPS, from l = 147 to 74 degrees; data between l = 114 and 87 degrees have yet to be processed. Contours of latitude-integrated HISA TON - TOFF contrast are shown on top of H I emission from the Leiden-Dwingeloo survey. Red lines mark approximate velocity boundaries between gas in the Local, Perseus, and Outer spiral arms.


Figure 2a: HISA Survey Map - East (l,b)

Top: Sky projection of detected HISA over a 28-degree section of the CGPS, showing velocity-integrated TON - TOFF values. Darker features have stronger absorption.
Bottom: 12CO emission in the Perseus spiral arm, where most of the HISA is found for this longitude range. Note: In the CASCA poster, this CO map was a transparency overlaid on the HISA map.


BONUS FEATURE: This figure was not in the CASCA poster!

Figure 2b: HISA Survey Map - West (l,b)

HISA found in a 12-degree section on the west end of the CGPS, looking down the tangent of the Local spiral arm in Cygnus. Unfortunately, no CO data is available here for comparison, but the long sightline column gives a very rich HISA field, with a covering factor approaching unity in places.


Figure 3: HISA Survey Coverage and Completeness

A cumulative count of the number of image voxels containing HISA, normalized by the number of voxels with bright enough H I for HISA to be detected. About 8% of the voxels in which HISA might be found appear to have HISA at some level, but there may be more HISA which is too faint to be seen in our data.


Figure 4: HISA - CO Correlation

HISA TON - TOFF contrast and 12CO brightness temperature show no obvious correlation, in contradiction to the traditional expectation of HISA tracing a small, fixed fraction of atomic gas in molecular clouds. The intensity scale of the plot is logarithmic.


Figure 5: HISA - CO Association

Despite the lack of quantitative correlation, many HISA clouds do contain CO at some level, and vice versa. Here we plot the number of image voxels with HISA above a given contrast and CO above a given brightness, normalized by the number of voxels which would be associated in a purely random distribution. The intensity scale is linear. The red contours mark S/N levels of 3, 10, 30, and 100 for the normalized association statistic.


BONUS FEATURE: This figure was not in the CASCA poster!

Figure 6a: Observed HISA Properties

Before invoking the ideal gas and line integral assumptions to derive the HISA temperature and opacity, we consider what can be measured directly from the raw data. The 4-component radiative transfer equation

can be re-written in linear terms as

where the first term in square brackets [ ] is the slope and the second [ ] term is the intercept. TON - TOFF, TOFF, and TC are observed, while TS, tau, and p are unknowns. If all three of the latter were constant for all HISA features (TC is typically insignificant), then a plot of TON - TOFF vs. TOFF would show a single, linear correlation.

From left to right, the four plots below show the raw distribution of points from the CGPS MW1 mosaic, then the same distribution normalized by the number of points in each TOFF bin, then the same data with a set of comparison slopes (ranging from 0.1 to 1.0 in steps of 0.1) with an arbitrary zero intercept, and finally a similar plot for the MN1 mosaic. Intensity scales are logarithmic.


Clearly there is no single linear relation describing these data, which exhibit considerable scatter. HISA contrast does not correlate well against off-HISA brightness, and most of the HISA is quite low-contrast regardless of TOFF. If the faint HISA comprising the bulk of the scattered points has any coherent slope, it is no steeper than 0.1; if the points are distributed homogeneously, so that on average p ~ 0.5, then tau < 0.2 for most HISA.

But leaving aside the faint HISA points in the scatter core, many of the stronger HISA features comprising the upper envelopes do exhibit a limited degree of linearity, with slopes of 0.2 or steeper (occasionally ~ 1.0). Since p <= 1, these must have tau > 0.2 (occasionally > 1.0), as might be expected. But even here, the linearities are only partial, breaking down for TOFF > ~ 90-100 K. This suggests either tau or p (or TS via the intercept term), is changing with TOFF for the stronger HISA and brighter off-HISA emission -- in other words, while weak HISA may be homogeneously distributed temperature fluctuations in the ambient H I, the strong HISA cannot be, since its properties and/or sightline geometry vary with TOFF. One possibility is that the strong HISA is preferentially associated with spiral structure in some way.

Figure 6b: Derived HISA Properties

These histograms show the value distributions of a number of properties. The top row of TON - TOFF, delta-theta, and delta-v are all directly measured quantities. The two lower rows give derived values for TS, tau, and NHI, using the method outlined above with fn = 1 in the middle row and fn = 0.01 in the bottom row. The group of plots on the left is linearly scaled, while those on the right are log-scaled.


BONUS FEATURE: This figure was not in the CASCA poster!

Figure 6c: HISA Property Correlations and Limits

When one has properties for large numbers of objects, one might ask whether these correlate in any interesting way. The answer appears to be no. The figure on the left shows feature angular size vs. linewidth, while that on the right shows TS vs. tau.


While some property values are preferred over others, there are no obvious correlations (the linear features are sampling artifacts), and in fact there a general scatter over the full range of parameter values which are permissible by the limits of the telescope and the image processing software. No size vs. linewidth relation is seen in the HISA (perhaps indicating the clouds are not gravitationally bound), and the bimodality in the characteristic temperatures and optical depths results from Local arm gas being systematically closer than Perseus arm gas, which translates to smaller delta-s sightline dimensions in the property solution method (the HISA we detect has similar angular size regardless of distance). In effect, this survey is seeing the gas which it is able to see (which is colder in the local arm), but by no means the full range of what is actually there.


Return to the Neutral Hydrogen Self-Absorption page.