## Abstract

The Northern HIPASS catalogue (NHICAT) is the northern extension of the HIPASS catalogue, HICAT. This extension adds the sky area between the declination (Dec.) range of +2° < δ < +25°30′ to HICAT's Dec. range of −90° < δ < +2°. HIPASS is a blind H i survey using the Parkes Radio Telescope covering 71 per cent of the sky (including this northern extension) and a heliocentric velocity range of −1280 to 12 700 km s−1. The entire Virgo Cluster region has been observed in the Northern HIPASS. The galaxy catalogue, NHICAT, contains 1002 sources with vhel > 300 km s−1. Sources with −300 < vhel < 300 km s−1 were excluded to avoid contamination by Galactic emission. In total, the entire HIPASS survey has found 5317 galaxies identified purely by their HI content. The full galaxy catalogue is publicly available at http://hipass.aus-vo.org.

## 1 INTRODUCTION

The H i Parkes All-Sky Survey (HIPASS) survey is a blind H i survey using the Parkes Radio Telescope,1 and the Northern extension increases this survey by a further 37 per cent in sky area. The primary objective of extending Southern HIPASS to the north is to complement the southern census of gas-rich galaxies in the local Universe.

The H i mass function (HIMF; Zwaan et al. 2003) and galaxy two-point correlation function (Meyer et al. in preparation) based on Southern HIPASS showed that the statistical measures of the galaxy population from HIPASS are limited by cosmic variance. Recently, Zwaan et al. (2005) used HICAT to investigate the effects of the local galaxy density on the HIMF. Using the nth nearest neighbour statistic, they found tentative evidence that the low-mass end of the HIMF becomes steeper in higher density regions. These authors were able to examine the trend in the slope of the HIMF for different values of n in the statistic. Larger values of n correspond to sampling the density on larger scales. For each value up to n= 10, the slope became systematically steeper as the density increased. Thus it appears that the H i properties of galaxies might be influenced by environmental effects on quite large scales (where a typical separation of the n= 10 nearest neighbours is ∼5 Mpc), in addition to the well-known local effects, such as tidal interactions between neighbouring galaxies. Previously, Rosenberg & Schneider (2002) found α≈−1.2 and −1.5 for the slope of the HIMF in the immediate and field regions of the Virgo Cluster, respectively. Moreover, Springob, Haynes & Giovanelli (2005) found flatter slopes to the low-mass end of the HIMF in higher density regions. However, their galaxy sample was selected optically. Since Northern HIPASS covers the entire Virgo Cluster region, the Northern HIPASS catalogue (NHICAT) can be used in conjunction with the Southern HIPASS catalogue (HICAT) to explore these trends and investigate the effects of cosmic variance on HIPASS galaxy catalogue statistics.

It is worth noting that the Northern HIPASS also provides the first blind H i survey of the entire region in and around the Virgo Cluster. Assuming a Virgo distance of 16 Mpc and an integrated flux limit of 15 Jy km s−1 this corresponds to a mass sensitivity of 9 × 108 M. Thus the survey will detect any H i clouds above this mass limit in the vicinity of the Virgo Cluster, regardless of stellar content.

The Virgo Cluster provides a nearby example of processes that are more common at higher redshifts, such as galaxy–galaxy and galaxy–intracluster medium interactions. Northern HIPASS will be used to investigate the role of H i in a cluster environment in individual galaxies as well as statistically across the whole cluster. Understanding the role of H i is vital for galaxy evolution models. Kenney, van Gorkom & Vollmer (2004) found six galaxies in the Virgo Cluster showing distorted H i morphology. Using N-body simulations, Vollmer et al. (2001) investigated the effect of ram pressure stripping in the Virgo Cluster and found that H i deficiency is dependent on galaxy orbits within the cluster. They concluded that all the galaxies showing some form of distorted H i distribution have already passed through the centre of the cluster and are not infalling for the first time.

The catalogue of extragalactic H i sources from HIPASS was named HICAT and was presented in (Meyer et al. 2004, hereafter MZ04), while the completeness and reliability of HICAT was assessed by Zwaan et al. (2004). Here we present a catalogue of extragalactic H i sources from Northern HIPASS, named NHICAT. The basic parameters of HIPASS, Northern HIPASS, HICAT and NHICAT are given in Table 1. Apart from the declination coverage, the main difference between the two surveys and catalogues is the higher noise level in Northern HIPASS. For a full summary of parameters of existing blind H i surveys, including subsets of HIPASS (HIPASS Bright Galaxy Catalogue, Koribalski et al. 2004; the South Celestial Cap catalogue, Kilborn et al. 2002) see table 1 of MZ04. The Northern HIPASS Zone of Avoidance (NHIZOA) survey by Donley et al. (2005) covers northern declinations of the Galaxy – a subset of the Northern HIPASS area – at a higher sensitivity (rms = 6 mJy beam−1).

Table 1

Survey and catalogue parameters.

Survey name Survey range (°, km s−1rms (mJy beam−1Catalogue name Catalogue range (°, km s−1Sources
HIPASS δ < +2°, 13 HICAT δ < +2°, 4315
(MZ04) −1280 < v < 12 700  300 < v < 12 700
Northern extension +2° < δ < +26°, 14 NHICAT +2° < δ < +25°30′, 1002a
HIPASS, this work −1280 < v < 12 700  300 < v < 12 700
Survey name Survey range (°, km s−1rms (mJy beam−1Catalogue name Catalogue range (°, km s−1Sources
HIPASS δ < +2°, 13 HICAT δ < +2°, 4315
(MZ04) −1280 < v < 12 700  300 < v < 12 700
Northern extension +2° < δ < +26°, 14 NHICAT +2° < δ < +25°30′, 1002a
HIPASS, this work −1280 < v < 12 700  300 < v < 12 700
a

Note that one source was found slightly below Dec. +2°.

Optical identification of NHICAT sources will use similar techniques to HICAT (Doyle et al. 2005) and will be presented in a later paper. With a total spatial coverage of 29 343 deg2 and 5317 sources, the combined HICAT and NHICAT catalogue is the largest purely H i-selected galaxy catalogue to date. The Arecibo L-Band Feed Array (ALFALFA) surveys will eventually cover the same region of sky as Northern HIPASS and will extend up to a Dec. of +36°. More information about the progress of ALFALFA can be found online at http://egg.astro.cornell.edu/alfalfa (Giovanelli et al. 2005).

In this paper we present NHICAT, together with the completeness and reliability analysis of the catalogue. Section 2 reviews Northern HIPASS and its properties. The source identification and the generation of the catalogue is described in Section 2.1. Section 2.2 discusses the noise statistics of Northern HIPASS and the completeness of NHICAT is analysed in Section 3.1. The narrow-band follow-up observations and the reliability of NHICAT will be described in Section 3.2.

## 2 NORTHERN HIPASS DATA

Northern HIPASS was designed to cover all RAs in the region of Dec. +2° < δ < +25°. Observations were undertaken using the 21-cm multibeam receiver (Staveley-Smith et al. 1996) on the Parkes radio telescope during the period from 2000 to 2002. Observations were made by scanning in 8° strips of sky with 7° in right ascension (RA) of separation between scans. A 1024 channel configuration covering a 64-MHz bandwidth was used in the multibeam correlator to give a channel separation of Δv= 13.2 km s−1 across the heliocentric velocity range of −1280 to 12 700 km s−1. The observation and reduction methods are exactly the same as Southern HIPASS and can be found in detail in Barnes et al. (2001). The northern data set consists of 102 8 × 8 deg2 cubes and 48 8 × 7 deg2 cubes in the northernmost declination band.

The catalogue includes sources in the range +25° < δ < +25°30′. At this declination, the telescope field of view is increased, though the sensitivity is significantly decreased.

### 2.1 Northern HIPASS catalogue (NHICAT)

NHICAT was generated using essentially the same method as HICAT, with some improvements in efficiency. An updated version of the tophat finder algorithm (see MZ04 for details of the original tophat) was used to identify sources. The updated tophat finder is very effective at filtering false detections which have narrow velocity widths. Velocity widths with a FWHM of less than 30 km s−1 were considered to be too narrow to be extragalactic sources. Such narrow velocity width detections are usually associated with hydrogen recombination frequencies or known interference lines. The consistency of the updated finder was tested by comparing the output of the original and updated finder for the southern cubes. The updated finder returned exactly the same sources as the original version without the narrow velocity width detections. Two galaxy finders were used to generate HICAT: ‘multifind’ and tophat. The ‘multifind’ finder uses a peak flux threshold method, whereas the tophat algorithm involves the cross-correlation of spectra with tophat profiles at various scales (MZ04). Although the original version of tophat was reported to find ∼90 per cent of final catalogue sources in southern HICAT (MZ04), further tests showed that all sources with Speak > 100 mJy were recovered by the updated tophat finder. Since ‘multifind’ generated many more false detections in the northern data due to the different level and character of noise (see Section 2.2), we decided to use the updated tophat finder only. The tophat finder was found to be quite robust against the increased level of noise and baseline ripple. In addition, since the updated tophat finder is much more effective, separate radio-frequency interference (RFI) and recombination line removal (as described in MZ04) was not necessary.

Although the known narrow-band RFI have been filtered out in the process described above, not all the known RFI have been removed from the data. Not only is the Sun (and the reflections of the Sun) the strongest source of interference but it also provides broad-band interference which could not be easily filtered out. This solar interference produces a standing wave effect (known as ‘solar ripple’) in our data which in turn will affect the effectiveness of galaxy detection. These standing waves are likely to be worse at lower elevations due to the ground reflection of the Sun.

NHICAT was constructed by first running the tophat galaxy finder on all cubes, excluding the region −300 < vhel < 300 km s−1 to minimize confusion with Galactic emission. 14 879 galaxy candidates were found. Each candidate source was then checked manually by simultaneously displaying the source in four ways: a spectrum, an integrated intensity map, a declination–velocity projection and a RA–velocity projection. A candidate source was accepted if it had a spectral profile which was easily distinguishable from the noisy baseline and a position–velocity profile which was wider than 2 pixel across the position axes. On the other hand, a candidate source was rejected if either its spectral profile was not distinguishable from the noise or if its position–velocity projections showed a distinct signal which was exactly 2 pixel wide, as these sources are usually the result of interference in the data. All the accepted sources were then passed to a semi-interactive parameter finder. Sources were flagged during parameter finding as either 1, 2, 3, 4 or a high-velocity cloud (HVC) detection: Flag = 1 represented a definite source detection, Flag = 2 represented source detection with less certainty, Flag = 3 represented a source detected on the edge of a cube and Flag = 4 represented a non-detection. The Flag = 4 option is provided to the parameter finder in the case of misclassification of a source during the checking stage. It should be noted that a significant number of HVCs were detected in Northern HIPASS at RA ∼ 23 h due to the Magellanic stream. The Magellanic stream from the Northern HIPASS data has been presented by Putman et al. (2003). Source lists from all cubes were then merged and duplicates were removed. The process of merging matched every Flag = 3 source with the detection of the same source in the neighbouring cube with the same overlap region. The procedures for the manual checking and parametrization used are exactly the same as MZ04. More detailed descriptions of these procedures can be found in Sections 3.2 and 3.3 of MZ04. Bivariate distributions of the basic properties (heliocentric velocity, velocity width, peak flux and integrated flux) of the sources detected in Northern HIPASS are shown in Fig. 1. Single-parameter histograms are shown on the diagonal of this figure.

Figure 1

Log–log bivariate distributions of velocity width (Wmax50), peak flux (Sp) and integrated flux (Sint). Plotted along the diagonal are single-parameter histograms.

Figure 1

Log–log bivariate distributions of velocity width (Wmax50), peak flux (Sp) and integrated flux (Sint). Plotted along the diagonal are single-parameter histograms.

The cubes were then rechecked for missed sources using a semi-interactive process. Sources detected in each cube were marked with a cross on an integrated intensity map. These maps were then manually checked for unmarked sources. From these manual checks, 15 additional sources were detected. The galaxy finder missed sources which were close to several other sources.

Two sources may also have been identified as the same source. Such sources can be differentiated by inspecting the spectra and the spatial moment maps since separate sources will not overlap both spatially and spectrally. These particular sources were flagged as ‘confused’. There are two ‘confused’ sources in NHICAT.

Extended sources were identified and fitted in the same fashion as in MZ04 (see Section 3.5 of MZ04 for more details). All sources greater than 7 arcmin in size were identified as potentially extended sources. The integrated flux limit corresponding to a fixed source size can be determined using the relationship between integrated flux, Sint, and source diameter,

1

An explanation and derivation of equation (1) can be found in MZ04. The Sint limit corresponding to a source size greater than 7 arcmin is 57 Jy km s−1. In total, 41 candidates were found to have a measured flux greater than this limit. The moment maps of each of these sources were then examined and two sources were found to be extended.

All the cube velocities are heliocentric and use the optical definition where velocity, v=c(fo/f)/f where c, fo and f represent the speed of light, rest and observed frequency, respectively. The sources were then assigned names according to the convention of HICAT (MZ04).

The final stage of the processing involved checking the 19 sources located at δ < +2°. Of these, 18 sources have already been catalogued in HICAT showing that the datasets and analysis techniques are consistent between the two surveys at the 95 per cent level. The extra source is located only slightly south of δ=+2° and has been retained in NHICAT so that the final combined NHICAT and HICAT catalogue will be a complete catalogue of the southern skies up to δ=+ 25°30′.

Northern HIPASS covers 7997 deg2 of sky and 1002 galaxies were found from their H i content in this region. The parameters provided with the catalogue are the same as the parameters given in HICAT. Detailed descriptions of each parameter can be found in table 4 of MZ04. Fig. 2 shows the spatial distribution of all the sources found in NHICAT. Note that the cluster of sources at RA α∼ 12 h is the Virgo cluster. Galaxies are also found at low Galactic latitudes – a region often avoided by optical galaxy surveys. As mentioned in the introduction, NHIZOA (Donley et al. 2005) covers a subset of the Northern HIPASS survey region. We re-identify 23 galaxies from the NHIZOA catalogue in the overlapping regions. The rms of NHIZOA is 2.33 times less than that of the observations in Northern HIPASS. Assuming that the detected number of galaxies is based solely on the sensitivity of the observations, one would expect Northern HIPASS to find only 33 of the 77 NHIZOA sources. There are fewer matches between Northern HIPASS and NHIZOA because the source detection rate is not fully described by the sensitivity (as explained in the following section).

Figure 2

Skymap of detections found in Northern HIPASS. The lines mark increasing declinations inwards where the centre is the north pole. The radial divisions show increasing RA in an anticlockwise direction starting with 0 h at the top of the diagram. The dotted lines mark lines of Galactic latitude, b.

Figure 2

Skymap of detections found in Northern HIPASS. The lines mark increasing declinations inwards where the centre is the north pole. The radial divisions show increasing RA in an anticlockwise direction starting with 0 h at the top of the diagram. The dotted lines mark lines of Galactic latitude, b.

In accordance with HICAT conventions, NHICAT only included detections with vhel > 300 km s−1. One galaxy, HIPASSJ1213+14a, was found to have a mean H ivhel of −222.7 km s−1, Speak= 0.181 Jy and Sint= 42.9 Jy km s−1. This galaxy was detected because its velocity width extended into the heliocentric velocity range where vhel < −300 km s−1. In summary, no galaxies with vhel < −300 km s−1 were found in NHICAT. The number density of the sources found in Northern HIPASS was approximately 0.13 sources deg−2 of sky. In comparison, 0.20 sources deg−2 of sky were found in Southern HIPASS. The cause of this difference will be discussed later in this paper.

### 2.2 Noise characteristics

Fewer sources were detected in NHICAT at higher declinations than in HICAT, as can be seen in Table 2 (which shows the number density in NHICAT and HICAT). Although some deviation is expected due to cosmic variance, the most likely cause of this density difference is the higher level of noise in Northern HIPASS. The variation in gain and system temperature (Tsys) of the telescope with respect to elevation are insufficient to account for the higher noise observed in the northern survey. The level of noise in Northern HIPASS is greater than in Southern HIPASS because the Parkes radio telescope observes northern sources at lower sky elevations. This results in the telescope gathering a greater amount of interference from ground reflections and the Sun. As the most northerly areas of the survey were only observable during a short local sidereal time (LST) window, sidelobe solar interference was often unavoidable.

Table 2

Comparison of source number density in NHICAT and HICAT. Note that one source in NHICAT was detected below 2° in Dec.

Catalogue Declination range Area of sky (deg2No. of sources Number density (deg−2
HICAT −90° < δ < +2° 21 346 4315 0.20
NHICAT +2°≤δ < +10° 2862 413 0.14
+10°≤δ < +18° 2792 364 0.13
+18°≤δ < +25° 2343 224 0.10
Catalogue Declination range Area of sky (deg2No. of sources Number density (deg−2
HICAT −90° < δ < +2° 21 346 4315 0.20
NHICAT +2°≤δ < +10° 2862 413 0.14
+10°≤δ < +18° 2792 364 0.13
+18°≤δ < +25° 2343 224 0.10

As shown in Table 1, the rms of both Northern and Southern HIPASS are very similar. However, the Northern HIPASS cubes appeared much ‘noisier’ in the flux density maps than that of the cubes in Southern HIPASS. Hence, the rms method is not an effective way of illustrating why Northern HIPASS appeared ‘noisier’.

The aim of this section is to characterize the distribution of noise observed in Northern HIPASS cubes. One method is to examine the distribution of all the pixel flux values in a given cube and measure the 99-percentile value of this distribution. This measure illustrates the noise characteristics in a given cube by measuring the extent of the outlying pixel flux values in the pixel distribution. Instead of characterizing the noise in terms of the width of the flux distribution (as in the rms method), we now examine the extent to which the outlier population is extended.

Fig. 3 shows the peak-normalized pixel flux distributions of two average cubes, one from Northern HIPASS (cube number 538) and the other from Southern HIPASS (cube number 194). Since H i detections comprise very few pixels compared with the entire cube, the H i sources have not been removed from the plot. The excess of negative flux in Northern cubes (as seen in Fig. 3) is due to the bandpass removal and calibration method. Negative bandpass sidelobes occur at declinations either side of bright sources (Barnes et al. 2001). This means that bright interference is surrounded by negative artefacts, leading to an excess of both positive and negative pixels in the data with stronger interference. An ideal cube with only Gaussian noise would have a parabolic distribution (as shown by the solid line) since the natural log of a Gaussian distribution, exp(−x2), is −x2. As can be seen in Fig. 3, the offset from the parabola is greater in the Northern cube than in the Southern cube, suggesting broader distributions of pixel noise values in the north.

Figure 3

Peak-normalized distributions of pixel flux of entire cubes in HIPASS. The distribution of the pixel flux from a typical northern cube (cube number 538) is represented by crosses and the distribution marked by circles represent the pixel flux from a typical southern cube (cube number 194). A parabola (solid line) is shown to compare these distributions with Gaussian noise statistics.

Figure 3

Peak-normalized distributions of pixel flux of entire cubes in HIPASS. The distribution of the pixel flux from a typical northern cube (cube number 538) is represented by crosses and the distribution marked by circles represent the pixel flux from a typical southern cube (cube number 194). A parabola (solid line) is shown to compare these distributions with Gaussian noise statistics.

Thus the ‘noise’ level can now be characterized by measuring the extent of outliers using the 99-percentile rank of the pixel flux distribution. In actuality, the 1-percentile rank has been used to avoid bias caused by actual H i sources in this measure. Fig. 4 shows the skymap of the 1-percentile measures for both Northern HIPASS and Southern HIPASS. It should be noted that the three outer bands of declinations (cube numbers 389–538) form the regions observed in Northern HIPASS, and the inner declination bands (cube numbers 1–388) represent the observed regions in Southern HIPASS.

Figure 4

Skymap of the 1-percentile pixel flux map of both Northern and Southern HIPASS. The south pole (δ=−90°) is in the centre and RA increases in an anticlockwise direction starting with 0 h at the top of the diagram. Observations through the Galaxy (where b= 0°) correspond to the darker horizontal band of cubes. The southern cube identification numbers range from 1 to 388 and the northern cubes are from 389 to 538.

Figure 4

Skymap of the 1-percentile pixel flux map of both Northern and Southern HIPASS. The south pole (δ=−90°) is in the centre and RA increases in an anticlockwise direction starting with 0 h at the top of the diagram. Observations through the Galaxy (where b= 0°) correspond to the darker horizontal band of cubes. The southern cube identification numbers range from 1 to 388 and the northern cubes are from 389 to 538.

It is qualitatively obvious from Fig. 4 that observations through the Galaxy appear much ‘noisier’ than observations away from the Galaxy. It can also be seen that the northernmost declination band in Northern HIPASS is much noisier than the rest of the HIPASS observations.

A quantitative version of Fig. 4 is as shown in Fig. 5 with three normalized distributions of cube ‘noise’ levels (characterized by 1-percentile rank). The cube ‘noise’ levels of Southern HIPASS are represented by the solid line distribution; the dashed and dotted lines represent cube ‘noise’ level distributions of Northern HIPASS. The cube ‘noise’ level of the southernmost declination band in Northern HIPASS is shown by the dashed-line distribution and the cube ‘noise’ level of the rest of Northern HIPASS is shown by the dotted-line distribution. The median ‘noise’ level for the solid line and dashed-line distributions are at 42 and 43 mJy, respectively and the median ‘noise’ level for the dotted-line distribution (northernmost HIPASS) is 55 mJy. By this measure, the noise level has increased by 31 per cent.

Figure 5

Normalized histograms of ‘noise’ levels (1-percentile measures). The distribution of ‘noise’ levels in Southern HIPASS is represented by the solid line distribution. The dashed-line distribution represents the ‘noise’ levels in the declination band between +2° < δ < +10° and the dotted-line distribution represents the ‘noise’ levels in the declination bands between +10° < δ < +26°.

Figure 5

Normalized histograms of ‘noise’ levels (1-percentile measures). The distribution of ‘noise’ levels in Southern HIPASS is represented by the solid line distribution. The dashed-line distribution represents the ‘noise’ levels in the declination band between +2° < δ < +10° and the dotted-line distribution represents the ‘noise’ levels in the declination bands between +10° < δ < +26°.

In conclusion, the ‘noisiness’ observed in the Northern HIPASS cube can be described quite accurately using the 1-percentile rank which measures the values of the pixel flux outliers in each cube. As mentioned before, the increase in the 1-percentile rank values for the northernmost declination band in Northern HIPASS is most likely a result of the increase in solar interference from increasingly lower elevation observations.

## 3 COMPLETENESS AND RELIABILITY OF NHICAT

The techniques used to calculate the completeness and reliability in NHICAT are the same as the methods used for HICAT. Detailed descriptions of the methods used to analyse the completeness and reliability of HICAT can be found in Zwaan et al. (2004).

### 3.1 Completeness of NHICAT

Synthetic sources were inserted into all the Northern HIPASS cubes before NHICAT was constructed in order to measure the completeness of the resulting catalogue. These synthetic sources were then extracted after the parameter finding process. In total, 774 non-extended synthetic sources were inserted into the Northern HIPASS cubes. These sources represent a random sample of sources ranging in 50 per cent velocity width (W50) from 20 to 650 km s−1, from 0.02 to 0.13 Jy in peak flux (Sp) and from 300 to 10 000 km s−1 in heliocentric velocity (vhel).

The completeness of recovery can be easily estimated by measuring the fraction of fake sources (recovered in each parameter bin), D:

2

However, the completeness as a function of one parameter cannot be effectively measured solely using D. The completeness, C, of NHICAT can be measured via the ratio of the number of detected real sources, N, over the true number of sources in each bin. The true number of sources in each bin can be estimated by N/D. Since the number of sources in each bin differs and the parameter distribution of the fake sources may be different from the distribution of the real galaxies, this method cannot give a very good estimate of the completeness as a function of one parameter. A correction can be made by integrating over another parameter and applying a weighting to account for the different number of sources in each bin. As an example, C(Sp) can be estimated by integrating over W:

3

Likewise, C(W) and C(Sint) have been calculated by integrating over Sp. Fig. 6 shows the completeness of NHICAT as functions of Sp, W and Sint, respectively, where the solid lines are error function fits to the data point. The error bars on the data points were determined using bootstrap resampling and show 68 per cent confidence levels. The error function fits and the completeness limits at 95 per cent confidence levels are given in Table 3. Using the same method as Zwaan et al. (2004), different fitting functions were tested in order to fit the completeness as a function of two parameters. The completeness of NHICAT as a function of Sp(Jy) and Sint (Jy km s−1) is also shown in Table 3.

Figure 6

Completeness of NHICAT as a function of Sp, W50 and Sint.

Figure 6

Completeness of NHICAT as a function of Sp, W50 and Sint.

Table 3

Completeness of NHICAT.

Parameter Completeness fit C= 0.95
Sp (Jy) erf[18.5(Sp− 0.02)] 0.095
Sint (Jy km s−1erf[0.1(Sint− 1.0)] 15.0
Sp(Jy), Sint (Jy km s−1erf[20.0(Sp− 0.005)]erf[0.14(Sint− 1.0)]
Parameter Completeness fit C= 0.95
Sp (Jy) erf[18.5(Sp− 0.02)] 0.095
Sint (Jy km s−1erf[0.1(Sint− 1.0)] 15.0
Sp(Jy), Sint (Jy km s−1erf[20.0(Sp− 0.005)]erf[0.14(Sint− 1.0)]

The completeness [C(Sp)] limits at 95 per cent confidence level is at 68 mJy for HICAT, while NHICAT's completeness at the same confidence level is 91 mJy. It appears that to first order, the completeness of Sp scales with the noise level. However, it would be too simplistic to assume that the noise levels and source detection scale linearly. It should also be noted that cosmic variance has not been taken into account.

### 3.2 Reliability of NHICAT

The reliability of NHICAT was measured by re-observing a subsample of NHICAT sources in the narrow-band mode at the Parkes Telescope. As with the reliability estimation of HICAT (Zwaan et al. 2004), a random sample representing the full range of NHICAT parameters was chosen, while giving preference to NHICAT detections with low Sint and Sp (generally with Sint < 8 Jy km s−1 and Sp < 0.07 Jy).

#### 3.2.1 Narrow-band observations

In addition to calculating the reliability of NHICAT, the narrow-band observations were also used to remove spurious detections from the catalogue. As such, the less certain source detections (Flag = 2) were observed as higher priority. In addition, definite source detections (Flag = 1) were chosen randomly by the observer for observation. These narrow-band observations took place over four observing sessions from 2003 July to 2005 February.

A spectral resolution of 1.65 km s−1 at z= 0 was used by observing with the narrow-band mode which consisted of 1024 channels over 8 MHz. Integration times were approximately 15 min for each source. The specific details of the observing mode used can be found in Zwaan et al. (2004). The narrow-band observations were reduced using the aips++ packages livedata and gridzilla (Barnes et al. 2001) as with the narrow-band observations of HICAT.

The percentage of rejected sources in the NHICAT narrow-band observations was greater than for HICAT. A likely explanation is that the signal-to-noise-ratio level is less in the north and thus source detection algorithm is less effective than in the south.

A total of 857 sources were observed, of which 172 (20 per cent) were rejected, compared with the narrow-band observations of HICAT where less than 10 per cent of the observed sources were rejected. Fig. 7 shows the peak flux distributions of the follow-up observations in both Northern and Southern HIPASS. As can be seen from this figure, more sources were rejected at Speak > 0.05 Jy in the Northern follow-up observations than in the Southern follow-up observations. Moreover, the percentage of sources with Speak > 0.05 observed in the Southern follow-up observations are higher than in the Northern observations which may also explain the difference in the percentage rejected.

Figure 7

Peak flux, Speak (Jy), distributions of the follow-up observations in Northern HIPASS (left-hand panel) and in Southern HIPASS (right-hand panel). The distributions marked with lines on a 45° angle represent the population of observed and confirmed detections, while, the distributions marked with horizontal lines represent the population of non-detected sources.

Figure 7

Peak flux, Speak (Jy), distributions of the follow-up observations in Northern HIPASS (left-hand panel) and in Southern HIPASS (right-hand panel). The distributions marked with lines on a 45° angle represent the population of observed and confirmed detections, while, the distributions marked with horizontal lines represent the population of non-detected sources.

#### 3.2.2 Reliability measure

As explained in Zwaan et al. (2004), the reliability will improve when more uncertain sources are removed after the narrow-band follow-up observations. Therefore, we start by examining the original catalogue before unconfirmed sources were removed. The ratio of the number of confirmed sources (Nrelconf) to the number of observed sources (Nrelobs) is defined to be

4

The reliability R as a function of a single parameter is the mean of T, weighted by the number of sources in each bin. For example, the reliability as a function of peak flux (Sp) is

5

Similarly, R(W) and R(Sint) can be measured by integrating over Sp.

Since 20 per cent of the observed sources in the initial NHICAT have been rejected and removed from the catalogue, the reliability of the catalogue has been improved by re-observing a subsample of the sources.

An estimate of the expected number of real sources has to be made in order to calculate the reliability of NHICAT after the removal of unconfirmed sources. The expected number of sources is given by

6

where Nunobserved is the number of sources that have not been re-observed.

T can now be redefined as

7

where N(Sp, W) is the total number of sources in NHICAT excluding the unconfirmed sources. Equation (5) can now be used to calculate the reliability of the final NHICAT.

Fig. 8 shows the Sp, W50 and Sint distributions of NHICAT sources as well as the reliability of NHICAT as functions of Sp, W50 and Sint where the dashed lines are error function fits to the data points. It should be noted that 871, 1002 and 764 sources are plotted in the Sp, W and Sint distributions, respectively. This is due to the fact that the rest of the sources have Sp and Sint greater than the parameter ranges given in Fig. 4. The error bars on the data points were determined using bootstrap resampling and indicate 68 per cent confidence levels. The error function fits and the completeness limits at 95 per cent confidence levels are given in Table 4.

Figure 8

Reliability of NHICAT as a function of Sp, W50 and Sint. The histograms show the Sp, W50 and Sint distributions of NHICAT sources.

Figure 8

Reliability of NHICAT as a function of Sp, W50 and Sint. The histograms show the Sp, W50 and Sint distributions of NHICAT sources.

Table 4

Reliability of NHICAT.

Parameter Reliability fit C= 0.95
Sp (Jy) erf[58(Sp− 0.012)] 0.036
Parameter Reliability fit C= 0.95
Sp (Jy) erf[58(Sp− 0.012)] 0.036

It is interesting though to note that the 95 per cent level of reliability in NHICAT is lower (in Sp and Sint) than the 95 per cent level of reliability in HICAT. This result can be attributed to the fact that a larger proportion of the original NHICAT have been re-observed in the narrow-band follow-up observations.

## 4 SUMMARY

In the northern extension of HIPASS, which covers the region of Dec. +2° < δ < +25°30′, 1001 extragalactic sources were found and catalogued in NHICAT. In addition, an extra source found with δ slightly less than +2° (which was not detected in HICAT) has also been included into NHICAT.

NHICAT has been found to be 95 per cent complete at peak flux 95 mJy and at an integrated flux 15 Jy km s−1. This catalogue is also reliable at a 95 per cent level at peak flux 36 mJy. The entire catalogue and source spectra will be made publicly available online at http://hipass.aus-vo.org. The catalogue parameters presented in the online archive are the same 33 parameters detailed in table 4 of MZ04. An excerpt of the online catalogue is shown in Table 5. The online archive also contains the detection spectra and integrated intensity maps in various projections. Examples of these data products are shown in Fig. 9. In addition to the Aus-VO archive, the catalogue and spectra will also be submitted to the NASA/IPAC Extragalactic Database.

Table 5

Excerpt from NHICAT. Note that all the velocities are cz and heliocentric.

Name RA v$S p$ (km s−1) vspechi (km s−1) Sint(Jy km s−1) Comment Dec. vgsr (km s−1) vmask (km s−1) rms (Jy) Follow-up vmax50 (km s−1) vlg (km s−1) Wmax50 (km s−1) rmsClip (Jy) Confused vmin50 (km s−1) vcmb (km s−1) Wmin50 (km s−1) rmsCube (Jy) Extended vmax20 (km s−1) vlo (km s−1) Wmax20 (km s−1) Cube vmin20 (km s−1) vhi (km s−1) Wmin20 (km s−1) Sigma (km s−1vmom (km s−1) vspeclo (km s−1) Sp(Jy) Box size (arcmin)
HIPASSJ0030+02 00:30:00.6 02:05:46 762.0 5509.0 5347.5 765.3 5362.3
5514.5 5370.3 5344.9 5336.2 5111.8 5590.2 4055.5
6973.1 5112, 5590 361.8 73.5 434.8 434.8 0.060
13.5 0.0065 0.0049 0.012 09 389 12

HIPASSJ0033+02 00:33:44.3 02:40:37 4340.8 4454.8 4390.4 4336.6 4402.7
4478.3 4411.8 4386.3 4377.6 4242.8 4542.0 2989.5
5941.5 4243, 4542 225.9 94.6 248.0 248.0 0.069
9.5 0.0080 0.0073 0.012 22 390 12

HIPASSJ0142+02 01:42:28.4 02:56:20 2988.9 1763.9 1764.0 2989.1 1762.8
1744.6 1772.0 1746.6 1737.9 1685.4 1843.3 321.3
3221.2 1685, 1843 80.6 80.6 109.9 109.9 0.112
9.0 0.0070 0.0058 0.013 62 392 12

HIPASSJ0150+02 01:50:15.2 02:18:58 1503.8 1695.8 1696.6 1505.4 1696.0
1677.9 1703.0 1677.6 1668.9 1618.9 1750.0 255.4
3127.1 1619, 1750 73.0 73.0 94.9 94.9 0.053
3.5 0.0065 0.0060 0.013 62 392 12

HIPASSJ0249+02 02:49:06.4 02:08:27 1024.6 1104.7 1103.6 1023.3 1104.5
1105.4 1113.8 1088.4 1079.7 1047.8 1167.3 −324.4
2536.0 1048, 1167 58.6 58.6 81.0 81.0 0.945
55.6 0.2363 0.0543 0.013 80 394 12

HIPASSJ0253+02 02:53:48.6 02:20:42 3645.8 6787.8 −9999.0 3634.9 6730.6
6833.6 6739.1 6713.7 6705.0 6532.9 6924.2 5129.9
8319.0 6533, 6924 349.4 235.8 0.000 0.000 0.037
9.2 0.0072 0.0066 0.014 40 395 12

HIPASSJ0254+02 02:54:05.6 02:57:22 6731.0 2952.5 3039.7 −9999.0 3036.2
2937.1 3043.9 3018.5 3009.8 2864.7 3212.5 1490.0
4603.0 2865, 3212 264.4 93.4 293.6 293.6 0.090
17.1 0.0075 0.0063 0.014 40 395 12

HIPASSJ0259+02 02:59:48.2 02:45:38 376.0 2793.6 2848.4 −9999.0 2844.4
2762.3 2850.4 2825.0 2816.4 2668.9 3025.0 1516.0
4384.8 2669, 3025 242.6 137.5 307.0 307.0 0.088
16.9 0.0075 0.0069 0.014 40 395 12

Name RA v$S p$ (km s−1) vspechi (km s−1) Sint(Jy km s−1) Comment Dec. vgsr (km s−1) vmask (km s−1) rms (Jy) Follow-up vmax50 (km s−1) vlg (km s−1) Wmax50 (km s−1) rmsClip (Jy) Confused vmin50 (km s−1) vcmb (km s−1) Wmin50 (km s−1) rmsCube (Jy) Extended vmax20 (km s−1) vlo (km s−1) Wmax20 (km s−1) Cube vmin20 (km s−1) vhi (km s−1) Wmin20 (km s−1) Sigma (km s−1vmom (km s−1) vspeclo (km s−1) Sp(Jy) Box size (arcmin)
HIPASSJ0030+02 00:30:00.6 02:05:46 762.0 5509.0 5347.5 765.3 5362.3
5514.5 5370.3 5344.9 5336.2 5111.8 5590.2 4055.5
6973.1 5112, 5590 361.8 73.5 434.8 434.8 0.060
13.5 0.0065 0.0049 0.012 09 389 12

HIPASSJ0033+02 00:33:44.3 02:40:37 4340.8 4454.8 4390.4 4336.6 4402.7
4478.3 4411.8 4386.3 4377.6 4242.8 4542.0 2989.5
5941.5 4243, 4542 225.9 94.6 248.0 248.0 0.069
9.5 0.0080 0.0073 0.012 22 390 12

HIPASSJ0142+02 01:42:28.4 02:56:20 2988.9 1763.9 1764.0 2989.1 1762.8
1744.6 1772.0 1746.6 1737.9 1685.4 1843.3 321.3
3221.2 1685, 1843 80.6 80.6 109.9 109.9 0.112
9.0 0.0070 0.0058 0.013 62 392 12

HIPASSJ0150+02 01:50:15.2 02:18:58 1503.8 1695.8 1696.6 1505.4 1696.0
1677.9 1703.0 1677.6 1668.9 1618.9 1750.0 255.4
3127.1 1619, 1750 73.0 73.0 94.9 94.9 0.053
3.5 0.0065 0.0060 0.013 62 392 12

HIPASSJ0249+02 02:49:06.4 02:08:27 1024.6 1104.7 1103.6 1023.3 1104.5
1105.4 1113.8 1088.4 1079.7 1047.8 1167.3 −324.4
2536.0 1048, 1167 58.6 58.6 81.0 81.0 0.945
55.6 0.2363 0.0543 0.013 80 394 12

HIPASSJ0253+02 02:53:48.6 02:20:42 3645.8 6787.8 −9999.0 3634.9 6730.6
6833.6 6739.1 6713.7 6705.0 6532.9 6924.2 5129.9
8319.0 6533, 6924 349.4 235.8 0.000 0.000 0.037
9.2 0.0072 0.0066 0.014 40 395 12

HIPASSJ0254+02 02:54:05.6 02:57:22 6731.0 2952.5 3039.7 −9999.0 3036.2
2937.1 3043.9 3018.5 3009.8 2864.7 3212.5 1490.0
4603.0 2865, 3212 264.4 93.4 293.6 293.6 0.090
17.1 0.0075 0.0063 0.014 40 395 12

HIPASSJ0259+02 02:59:48.2 02:45:38 376.0 2793.6 2848.4 −9999.0 2844.4
2762.3 2850.4 2825.0 2816.4 2668.9 3025.0 1516.0
4384.8 2669, 3025 242.6 137.5 307.0 307.0 0.088
16.9 0.0075 0.0069 0.014 40 395 12

Figure 9

Example of three data products available online at http://hipass.aus-vo.org for source HIPASSJ0030+02. Clockwise from top left-hand panel: integrated intensity map, H i spectra and a position–velocity projection intensity map.

Figure 9

Example of three data products available online at http://hipass.aus-vo.org for source HIPASSJ0030+02. Clockwise from top left-hand panel: integrated intensity map, H i spectra and a position–velocity projection intensity map.

1
The Parkes telescope is part of the Australia Telescope which is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO.

The multibeam system was funded by the Australia Telescope National Facility (ATNF) and an Australian Research Council LIEF grant. The HIPASS collaboration was partially supported by an ARC grant (DP0208618). The collaborating institutions are the Universities of Melbourne, Western Sydney, Sydney and Cardiff, Research School of Astronomy and Astrophysics at Australian National University, Jodrell Bank Observatory and the ATNF. The multibeam receiver and correlator was designed and built by the ATNF with assistance from the Australian Commonwealth Scientific and Industrial Research Organization Division of Telecommunications and Industrial Physics. The original low-noise amplifiers were provided by Jodrell Bank Observatory through a grant from the UK Particle Physics and Astronomy Research Council.

We thank N. Bate, A. Karick, E. MacDonald, M. J. Pierce, R. M. Price, N. Rughoonauth, S. Singh, D. Weldrake and M. Wolleben for their help with the Parkes narrow-band follow-up observations. Finally, we would like to thank the staff at the Parkes Observatory for all their support.

## REFERENCES

Barnes
D. G.
et al
,
2001
,
MNRAS
,
322
,
486
Donley
J. L.
et al
,
2005
,
AJ
,
129
,
220
Doyle
M. T.
et al
,
2005
,
MNRAS
,
361
,
34
Giovanelli
R.
et al
,
2005
,
AJ
,
130
,
2613
Kenney
J. D. P.
Van Gorkom
J. H.
Vollmer
B.
,
2004
,
AJ
,
127
,
3361
Kilborn
V. A.
et al
,
2002
,
AJ
,
124
,
690
Koribalski
B. S.
et al
,
2004
,
AJ
,
128
,
16
Meyer
M. J.
et al
,
2004
,
MNRAS
,
350
,
1195
(MZ04)
Putman
M. E.
Staveley-Smith
L.
Freeman
K. C.
Gibson
B. K.
Barnes
D. G.
,
2003
,
ApJ
,
586
,
170
Rosenberg
J. L.
Schneider
S. E.
,
2002
,
ApJ
,
567
,
247
Springob
C. M.
Haynes
M. P.
Giovanelli
R.
,
2005
,
ApJ
,
621
,
215
Staveley-Smith
L.
et al
,
1996
,
Proc. Astron. Soc. Aust.
,
13
,
243
Vollmer
B.
Cayatte
V.
Balkowski
C.
Duschl
W. J.
,
2001
,
ApJ
,
561
,
708
Zwaan
M. A.
et al
,
2003
,
AJ
,
125
,
2842
Zwaan
M. A.
et al
,
2004
,
MNRAS
,
350
,
1210
Zwaan
M. A.
Meyer
M. J.
Staveley-Smith
L.
Webster
R. L.
,
2005
,
MNRAS
,
359
,
L29