A blind H i survey in the Ursa Major region★

Abstract

We have conducted the first blind H i survey covering 480 deg² and a heliocentric velocity range from 300 to 1900 km s⁻¹ to investigate the H i content of the nearby spiral-rich Ursa Major region and to look for previously uncatalogued gas-rich objects. Here we present the catalogue of H i sources. The H i data were obtained with the four-beam receiver mounted on the 76.2-m Lovell telescope [full width at half-maximum (FWHM) 12 arcmin] at the Jodrell Bank Observatory (UK) as part of the H i Jodrell All Sky Survey (HIJASS). We use the automated source finder duchamp and identify 166 H i sources in the data cubes with H i masses in the range of 10⁷–10^10.5 M_⊙. Our Ursa Major H i catalogue includes 10 first-time detections in the 21-cm emission line.

We identify optical counterparts for 165 H i sources (99 per cent). For 54 H i sources (∼33 per cent) we find numerous optical counterparts in the HIJASS beam, indicating a high density of galaxies and likely tidal interactions. Four of these H i systems are discussed in detail.

We find only one H i source (1 per cent) without a visible optical counterpart out of the 166 H i detections. Green Bank Telescope (FWHM 9 arcmin) follow-up observations confirmed this H i source and its H i properties. The nature of this detection is discussed and compared to similar sources in other H i surveys.

1 INTRODUCTION

The distribution of galaxies in the Universe is not uniform – galaxies are found in filamentary structures, groups and clusters (e.g. Colless et al. 2001). The observed ‘cosmic web’ structure is in good agreement with numerical simulations according to cold dark matter cosmology (e.g. Springel et al. 2005; Simon & Geha 2007), whereby gravity drives the formation of structure from small, primordial, Gaussian fluctuations. Dark matter haloes grow through merging and accretion of smaller ones. Within the dark matter haloes, galaxies form by cooling of baryons (e.g. White & Rees 1978). Hence, small dark matter haloes are the primordial building blocks of larger structures in the hierarchical model of structure formation.

Dwarf irregular galaxies with low stellar luminosity and low surface brightness galaxies are typically discovered in neutral hydrogen (H i) surveys (Schombert et al. 1992, Minchin et al. 2003, Giovanelli et al. 2005), as they are rich in neutral gas but difficult to detect in optical or infrared surveys. Whilst optical surveys are most sensitive to high surface brightness early-type galaxies (Disney 1976), H i surveys are unbiased by stellar populations and are most likely to detect gas-rich late-type spirals and irregular galaxies. Hence optical and H i surveys complement each other as there is a rough correlation between increasing H i content from early- to late-type galaxies and decreasing optical surface brightness with morphological type (Toribio et al. 2011). Discoveries in H i surveys also include intriguing objects, e.g. the detection of an H i massive cloud without stellar counterpart, within the NGC 2442 group (Ryder et al. 2001), extended H i arms or tidal tails (Koribalski & López-Sánchez 2009) and H i between galaxies in groups (Koribalski & Manthey 2005, Kilborn et al. 2006, English et al. 2010). Large-area blind H i surveys have been conducted such as the Arecibo H i Strip Survey (AHISS; Zwaan et al. 1997), the Arecibo Dual Beam Survey (ADBS; Rosenberg & Schneider 2000), the H i Jodrell All Sky Survey (HIJASS; Lang et al. 2003), the H i Parkes All Sky Survey (HIPASS; Staveley-Smith et al. 1996, Barnes et al. 2001; Koribalski et al. 2004, Meyer et al. 2004), the Arecibo Legacy Fast ALFA survey (ALFALFA; Giovanelli et al. 2005) and the Arecibo Galaxy Environment Survey (AGES; Auld et al. 2006). Upcoming H i surveys such as the ASKAP H i All-Sky Survey (known as WALLABY; Koribalski & Staveley-Smith 2009¹) will further contribute to our knowledge of gas-rich galaxies in the local Universe.

Isolated galaxies tend to have noticeably different properties from galaxies residing in dense regions. Observations have shown that early-type galaxies are common in clusters whereas late-type galaxies dominate the less dense environments (morphology–density relation; e.g. Dressler 1980). The importance of environment and the idea of transformation of galaxies in dense environments arose when Butcher & Oemler (1978) discovered that galaxy clusters at higher redshift contain a population of blue galaxies which nearby galaxy cluster counterparts are missing. However, it is still a matter of debate as to which properties correlate with the decrease of blue galaxies from high- to low-redshift galaxy clusters, e.g. X-ray luminosity (Andreon & Ettori 1999; Fairley et al. 2002), richness (Margoniner et al. 2001; De Propris et al. 2004), cluster concentration (Butcher & Oemler 1984; De Propris et al. 2004), presence of substructure (Metevier, Romer & Ulmer 2000) and cluster velocity dispersion (De Propris et al. 2004). Lewis et al. (2002) and Gómez et al. (2003) discuss the correlation between star formation rate and local projected density. The authors find low star formation rates at 2–3 virial cluster radii – which is equivalent to poor galaxy group densities. Therefore, galaxies may be preprocessed in groups, as physical processes in the studied clusters cannot solely account for the observed variations in galaxy properties. Galaxy transformation as a smooth function of environment is also suggested by Cappellari et al. (2011) in terms of a kinematic morphology–density relation, i.e. separation into fast and slow rotators.

There have been a number of studies on the effect of environment on the H i content of galaxies in clusters (e.g. Giovanelli & Haynes 1985, Solanes et al. 2001, Gavazzi et al. 2005, Chung et al. 2009), and loose groups (e.g. Kilborn et al. 2005; Omar & Dwarakanath 2005). Through observations of isolated galaxies, we can define the expected H i mass in normal spiral galaxies of a particular linear optical diameter and morphological type. In dense regions, such as clusters, there is often an H i deficiency observed, indicating that H i gas is removed from galaxies in these environments (e.g. Solanes et al. 2001). Recent observations have found a similar H i deficiency in the less dense group environments (Omar & Dwarakanath 2005, Kilborn et al. 2009) indicating that the spirals may begin to lose their H i reservoir well before reaching the dense centre of a cluster.

An ideal target to study the effect of environment on the evolution of gas-rich galaxies is the nearby Ursa Major cluster and its surroundings as most of the known member galaxies are late-type galaxies. The Ursa Major cluster is quite different in its nature compared to the well-known and similarly distant Virgo and Fornax clusters. Generally clusters are composed of hundreds of gas-poor early-type galaxies clustered around a central larger elliptical (e.g. Dressler 1980). The Ursa Major cluster is smaller – Tully et al. (1996) identified 79 high-probability cluster members. The cluster is essentially composed of late-type galaxies, in particular spiral galaxies with normal gas content, that are not centrally concentrated (Tully et al. 1996). This lack of a concentration towards any core and the overlap with many groups nearby the cluster might explain why the Ursa Major cluster is less studied compared to the Virgo and Fornax clusters. The several groups nearby the Ursa Major cluster make studying its dynamics a complex problem and the cluster difficult to define (Appleton & Davies 1982, Tully et al. 1996). However, Tully et al. (1996) define the cluster members to lie within a projected 7| ${.\!\!\!\!\!\!^{\circ}}$|5 circle centred at α = 11^h56| ${.\!\!\!\!\!\!^{m}}$|9, δ = +49°22′ and with heliocentric velocities between 700 and 1210 km s⁻¹.

The cluster lies at a distance of 17.1 Mpc (Tully et al. 2008). The few previous studies of the Ursa Major cluster have found that it has a low velocity dispersion of only 148 km s⁻¹, a virial radius of 880 kpc (Tully et al. 1996) and no X-ray-emitting intracluster gas has yet been detected (Verheijen & Sancisi 2001). The Ursa Major cluster has a total mass of 4.4 × 10¹³ M_⊙ (Tully 1987) and a B-band absolute luminosity of 5.5 × 10¹¹ L_⊙ (Tully et al. 2008). Compared to the Virgo cluster this is about 30 per cent of its light but only 5 per cent of its mass (Tully 1987; Tully et al. 2008). The number of H i-rich galaxies in Ursa Major is comparable to those in Virgo (Tully et al. 1996) and the Ursa Major cluster almost meets the Abell richness 0 standard. To qualify as an Abell richness class 0 the criteria are (i) 30 galaxies in B or (ii) 26 in R within 2 mag of the third brightest galaxy. The Ursa Major cluster is slightly short – 29 galaxies in B and 25 in R (Tully et al. 1996). Besides Ursa Major, only the Virgo cluster has such a high abundance of H i-rich galaxies on a 2 Mpc scale in the nearby Universe (cz < 3000 km s⁻¹) (Tully et al. 1996).

Zwaan, Verheijen & Briggs (1999) note that the mostly late-type galaxies in the Ursa Major cluster do not seem to be seriously affected by tidal interactions indicating that the system is in an early evolutionary state. However, Verheijen & Sancisi (2001) studied 43 spiral galaxies in detail and found 10 interacting systems that show distorted morphologies and tidal debris. There is a deficiency of known dwarf galaxies in the Ursa Major cluster – there are only two dozen dwarfs, whereas up to 10³ could be expected to be found if the same steep luminosity function claimed for Virgo was valid for the Ursa Major cluster (Trentham, Tully & Verheijen 2001). To date, (i) the small number of dwarf galaxies, (ii) the low velocity dispersion, (iii) the non-detection of X-ray-emitting intracluster gas, (iv) the low abundance of elliptical galaxies in the cluster and (v) the total mass <10¹⁴ M_⊙ challenge the question ‘why not call it a group?’. Clusters usually have a broader velocity distribution, a hot intracluster gas which can strip the galaxies off their H i gas and an enhanced population of both the early-type (S0 and E) galaxies and the H i-depleted spiral galaxies. As the majority of the Ursa Major members are H i-rich spiral galaxies such as those one finds in the field, the distinction between the ‘Ursa Major cluster’ as opposed to an ‘Ursa Major group’ is not clear. One hypothesis is that Ursa Major is a newly forming cluster given that the estimated crossing time is approximately half a Hubble time, in contrast to Virgo and Fornax where the crossing times are 0.08 and 0.07 H^{− 1}₀, respectively (Tully et al. 1996).

In this paper we present the catalogue of H i sources as detected in the first blind H i survey of the Ursa Major region (approximately 480 deg²). The analysed region overlaps with the Canes Venatici region for which Kovač, Oosterloo & van der Hulst (2009) have conducted a deep blind H i survey (H i mass sensitivity ∼10⁶ M_⊙) using the Westerbork Synthesis Radio Telescope (WSRT) in an area of 86 deg² (v_sys ≤ 1330 km s⁻¹). Our H i data were obtained as part of the HIJASS. The data acquisition and determination of the H i sources are described in Section 2 including a discussion of the catalogue completeness and reliability. The H i properties of the sources are presented in a catalogue in Section 3. The H i mass limit of the presented catalogue is 2.5× 10⁸ M_⊙ assuming a velocity width of 100 km s⁻¹ and a distance of 17.1 Mpc. First, results and basic property distributions of the HIJASS sources are presented in Section 4. Furthermore, we identify galaxy pairs/systems that are high-probability candidates for galaxy–galaxy interactions and discuss four systems in detail, two of which show strikingly extended H i content.

In a forthcoming paper we will investigate the nature of the Ursa Major cluster/group and probe the hypothesis that it is a newly forming cluster. We will study substructures and their dynamics, and present dynamical masses, total mass-to-light ratios, etc. for both individual galaxies and the subregions. Furthermore, to investigate the H i properties of galaxies residing in different density regions, we will analyse the H i sources in the Ursa Major cluster and compare them to galaxies residing in foreground groups and the less dense filament connecting the Ursa Major with the Virgo cluster, which lies south of Canes Venatici.

2 OBSERVATIONS AND DATA ANALYSIS

2.1 Data acquisition and processing

The H i data cubes analysed in this paper were obtained in 2001–2002 as part of the HIJASS (Lang et al. 2003). The blind H i survey was conducted using the four-beam receiver mounted on the 76.2-m Lovell telescope at Jodrell Bank Observatory, UK. HIJASS covered more than 1000 deg² in the northern sky (δ ≥ +22°) complementary to (i) the HIPASS which surveyed the entire southern sky (δ ≤ +2°) from 1997 to 2000 (Staveley-Smith et al. 1996; Barnes et al. 2001; Koribalski et al. 2004) and (ii) the northern extension to HIPASS (+2° ≤ δ ≤ +25°30^′) carried out in 2000 to 2002 (Wong et al. 2006). HIJASS provides the first blind H i survey of the nearby Ursa Major cluster and its surroundings. To investigate the Ursa Major region we only consider velocities in the range 300–1900 km s⁻¹ (all velocities are given in the optical convention cz). Sources with systemic velocities below 300 km s⁻¹ are excluded to minimize confusion with high-velocity clouds and Galactic emission – an exception are 15 obviously real galaxies with systemic velocities between 150 and 300 km s⁻¹ that are included in the catalogue. We note that HIJASS data suffer from a broad-band of radio-frequency interference between 4500 and 7500 km s⁻¹ (Lang et al. 2003), which lies outside our studied velocity range.

The Ursa Major region is covered by seven HIJASS data cubes measuring 8° × 8° and one HIJASS cube measuring 5° × 8° in size (the sky coverage is approximately 480 deg²). The cubes span a range in right ascension (RA) between 10^h45^m and 13^h04^m and a range in declination (Dec.) between 22° and 54°. We note that the analysed region overlaps with the Canes Venatici region for which Kovač et al. (2009) have conducted a deep blind H i survey in an area of 86 deg² (v_sys ≤ 1330 km s⁻¹). The HIJASS cubes have a spatial pixel size of 4 × 4 arcmin². In the third dimension, the channel spacing is 62.5 kHz which corresponds to an increment of 13.2 km s⁻¹ at the rest frequency of H i. The velocity resolution is 18 km s⁻¹ after smoothing to reduce ‘ringing’ caused by Galactic emission (applying 25 per cent Tukey filter). The root mean square (rms) noise level in the analysed data is 12–14 mJy beam⁻¹. To minimize edge effects the overlapping HIJASS data cubes are mosaicked to one megacube. For further information on data reduction, the Lovell telescope and HIJASS, see Lang et al. (2003).

The angular resolution of the Lovell telescope at 1.4 GHz is 12.0 arcmin (Lang et al. 2003). The gridding process (using similar parameters as for HIPASS) slightly degrades the angular resolution (see Barnes et al. 2001). According to Barnes et al. (2001) the resultant beam size is not well defined and depends on smoothing radius, pixel size, source position, shape and strength (signal-to-noise ratio) as the gridding is a non-linear process. One can only assess the gridded beam size by analysing simulations and synthetic source injections. To determine the gridded beam size, we inject synthetic point sources into the raw data and measure their properties after the gridding process. We determine the full width at half-maximum (FWHM) beamwidth of a two-dimensional Gaussian fit to the synthetic sources, analogous to Barnes et al. (2001). The measured beamwidths vary as a function of source peak brightness (in Jy beam⁻¹, which corresponds to Jy for point sources). Given a wide range of peak flux densities in the presented catalogue – the majority of sources have 33 ≤ S_p ≤ 300 mJy and a few strong sources range to 3.5 Jy in peak flux density – we determine two appropriate beamwidths: (i) 13.1 ± 0.9 arcmin for the majority of sources with S_p ≤ 300 mJy and (ii) 12.4 ± 0.6 arcmin for sources with S_p > 300 mJy.

The peak and integrated fluxes of point sources are preserved during the gridding process but for extended sources, the peak and integrated fluxes require a correction: we calculated the correction factors by similarly injecting synthetic extended sources into the raw data and measure their properties after the gridding process (analogous to Barnes et al. 2001). The correction factors for extended sources as a function of FWHM are given in Table 1. We obtain this function by fitting a third-order polynomial to the median data points (with R² of 99.9 per cent). In our catalogue (Section 3.2), we quote both the measured and corrected fluxes for extended sources.

2.2 Data analysis using duchamp

We use the automated source finder duchamp-1.1.8 (Whiting 2012)² on the H i data cube to compile our initial candidate galaxy list. duchamp is based on a threshold method and searches for groups of adjacent pixels in position and velocity that are above the given threshold, which is estimated from the cube statistics (in Jy beam⁻¹) or alternatively specified from the user as a peak flux density value. Furthermore, the user can (i) reduce the noise prior to the searching via wavelet reconstruction and/or smoothing (spatial or spectral) and (ii) control various parameters such as the minimum extent of the detections in pixels and/or channels. duchamp provides a list of the H i detections and their properties, the H i spectra and integrated H i intensity maps (moment zero). The duchamp outputs need to be inspected carefully as H i sources are occasionally merged. We can identify such sources by eye and separate them when they do not overlap both spatially and spectrally.

We made a comprehensive test of various input parameters and methods to facilitate duchamp detections of low peak flux and extended H i sources without disproportionally increasing noise detections (see Sections 2.3 and 2.5). We first apply a spectral Hanning smoothing to the HIJASS data and then reconstruct them with the ‘à trous’ wavelet method. Hanning smoothing correlates neighbouring pixels in the spectral domain and determines the weighted average flux over a certain number of channels – the channel spacing remains 13.2 km s⁻¹. We determine that the optimal Hanning smoothing width for the data is seven channels (see Section 2.3). The wavelet method reconstructs each spectrum separately and we chose the following criteria when compiling the duchamp detection list: (i) the detection must be contiguous in two channels and (ii) the peak flux density of the candidate galaxies must be above 16 mJy. This is an approximate 2.5σ peak flux density threshold as the mean noise level in the smoothed and reconstructed cube is ∼6.6 mJy beam⁻¹.

In this paper, we only consider 5σ detections in peak flux in the edge-masked HIJASS megacube to present a reliable and complete catalogue of H i sources in the Ursa Major region within 300–1900 km s⁻¹. This corresponds to a minimum in peak flux density of 33 mJy and an H i mass limit of 2.5 × 10⁸ M_⊙ (assuming a velocity width of 100 km s⁻¹ and a distance of 17.1 Mpc). These selection criteria lead to a duchamp detection list with 140 candidate galaxies including merged detections (doubles to quads) that are labelled in Section 3.2.

2.3 Catalogue reliability – negative troughs

To test the reliability of the 5σ peak flux duchamp detection list, we investigate the negative troughs in the edge-masked HIJASS cube. When using the selection criteria as discussed in Section 2.2, we identify 25 negative troughs with S_p ≤ −33 mJy and w₅₀ ≥ 26.4 km s⁻¹. All 25 negative troughs can be associated with negative bandpass sidelobes, i.e. an artefact of bandpass removal which is apparent north and south of strong H i sources (see e.g. Barnes et al. 2001). This lack of negative detections (noise peaks) indicates that the positive duchamp candidate galaxy list with S_p ≥ 33 mJy and w₅₀ ≥ 26.4 km s⁻¹contains high-probability real H i sources.

Hanning smoothing is known to enhance the brightness of sources – on the scale of the smoothing width – with respect to the noise in each channel map, but reduces the peak strength (signal-to-noise ratio) of sources on other scales. Whiting (2012) shows that an appropriately chosen smoothing width can still increase the recovery rate of sources on different scales than the smoothing width as the noise level is suppressed. However, to prevent loss of information on different scales, we also use the wavelet reconstruction to further suppress the noise and to enhance structures on various scales (narrow and broad profiles). Using wavelet reconstruction also suppresses noise peaks, baseline ripples or radio-frequency interferences as discussed in Westmeier, Popping & Serra (2012). For further information on duchamp and the noise suppression methods see Whiting (2012).

Some restrictions on spatial and spectral extent of duchamp detections greatly improve the reliability, such as the applied minimum number of channels (≥2) in which the detection must be contiguous (Whiting 2012). The selection criteria as listed in Section 2.2 are chosen to limit the number of false detections and to present a reliable catalogue of H i sources in the Ursa Major region in this paper.

2.4 Catalogue reliability – follow-up observations

The reliability of the catalogue is increased by follow-up observations and rejection of unconfirmed sources from the catalogue. We re-observed 17 apparent first-time detections in HIJASS with the Green Bank Telescope (GBT; FWHM 9 arcmin) – nine of which have an associate optical counterpart that match in position and velocity to the H i detection and eight candidate galaxies/H i clouds without a previously catalogued optical counterpart. The rms noise level is between 4 and 7 mJy for H i sources with an associate optical counterpart and approximately 3 mJy for the candidate galaxies/H i clouds. Given that the lowest peak flux density of the candidate galaxies/H i clouds is 33 mJy, this will lead to signal-to-noise ratio detections >10. The velocity resolution of the GBT data is 3.3 km s⁻¹ after boxcar smoothing.

The GBT follow-up observations confirmed all nine H i sources with an associate optical counterpart. However, only one out of the eight candidate galaxies/H i clouds was confirmed, namely HIJASS J1219+46. Therefore, seven initial duchamp detections are excluded from the presented catalogue in this paper. Inspecting the original as well as the smoothed and reconstructed data cubes by eye reveal that the unconfirmed detections are noise peaks on top of large-scale baseline ripples. duchamp provides an option ‘flag baseline’ which removes the baseline from each spectrum prior to compiling the detection list. However, we did not utilize this option as not only does it remove all candidate galaxies/H i clouds from the detection list, some of the faint H i detections with an associate optical counterpart were not included as well.

2.5 Catalogue completeness – synthetic source injection

To measure the catalogue completeness we insert numerous synthetic point sources into the eight HIJASS cubes, prior to running duchamp. The rate at which the synthetic sources are recovered reflects the completeness, as discussed explicitly in Zwaan et al. (2004) (also see Wong et al. 2006). The criteria for a recovered synthetic source are a duchamp detection within 10 arcmin and 200 km s⁻¹ of the original position and velocity of the synthetic source (the average position and velocity accuracies for H i sources with peak flux densities below 100 mJy are ∼1.2 arcmin and ∼14 km s⁻¹, respectively).

The synthetic point sources are given a Gaussian spectral shape and have random positions, 50 per cent velocity widths and peak flux densities (Kilborn et al. 2009). The 50 per cent velocity widths are w₅₀ ≤ 500 km s⁻¹ and the peak flux densities are S_p ≤ 100 mJy. The synthetic sources are placed away from the noisy edges of the data cube and are not placed on top of other sources either synthetic or real (the positions of the real sources are according to an initial duchamp output list).

A total of 2000 synthetic sources are inserted into each HIJASS data cube to obtain the duchamp completeness. For a representative HIJASS cube the completeness is shown in Fig. 1. We use natural neighbour interpolation to achieve the grey-scale representation of the completeness depending on S_p and w₅₀ or S_p and integrated flux S_int (which is a combination of both S_p and w₅₀).

To measure the catalogue completeness as a function of one parameter only – S_p, w₅₀ or S_int – we integrate along the other variable of the completeness as discussed in Zwaan et al. (2004). We integrate above the duchamp selection parameters which are w₅₀ = 26.4 km s⁻¹ and S_p = 16 mJy. These values mark the duchamp detection limit in the smoothed and reconstructed cube – below these values the completeness drops to zero. Fig. 2 shows this differential completeness (circles) as well as the cumulative (crosses) completeness as defined by Zwaan et al. (2004). Fig. 2 also shows the error function fit to the data to determine the 95 per cent confidence levels which lie at S_p = 33 mJy and S_int = 2.63 Jy km s⁻¹. The 99 per cent completeness levels are achieved at S_p = 45 mJy and S_int = 3.55 Jy km s⁻¹. Therefore, the peak flux density limited catalogue (at 33 mJy) presented in Section 3 is 95 per cent complete.

2.6 Catalogue completeness – literature comparison

As an additional measure of completeness we compare our results (Section 3.2) to an overlapping deep blind H i survey in the Canes Venatici region using WSRT (area of 86 deg², v_sys ≤ 1330 km s⁻¹ with an H i mass sensitivity of ∼10⁶ M_⊙; Kovač et al. 2009). We find that our presented catalogue contains all but one H i sources in the velocity range 300–1330 km s⁻¹, with S_p ≥ 33 mJy and S_int ≥ 2.63 Jy km s⁻¹ (95 per cent completeness level) as listed in Kovač et al. (2009) – the exception is NGC 4460 (v_LG = 542 km s⁻¹) with S_p = 41 mJy and S_int = 4.30 Jy km s⁻¹ (Kovač et al. 2009). We investigated the HIJASS data cube (by eye) and identify NGC 4460 at the end of negative bandpass sidelobes emerging from the strong double source NGC 4490/NGC 4485. NGC 4460 has reduced peak brightness (S_p ≃ 31 mJy beam⁻¹) due to the negative bandpass sidelobes, which prevents this H i source to be listed in our catalogue. We conclude that negative bandpass sidelobes may cause a deficit in completeness for sources with 16 ≤ S_p ≤ 45 mJy and w₅₀ ≤ 100 km s⁻¹ (see Fig. 1), as their peak flux densities may lie below the detection threshold. However, this is a small effect, given the little area affected by negative bandpass sidelobes.

3 THE H i CATALOGUE

3.1 Parametrization

The sources found using duchamp are parametrized with an interactive script utilizing the miriad software package (Sault, Teuben & Wright 1995). The parametrization is necessary as duchamp gives initial values – dealing with double detections and obtaining reliable H i parameters require user input. Our method is similar to Lang et al. (2003) and Koribalski et al. (2004) and this will be briefly described in the following section.

A spatially integrated spectrum is generated from the original HIJASS cube, using the duchamp output parameters for each source (RA, Dec., velocity, velocity widths). The spectral extent of the H i source (velocity range) can be adjusted as necessary. An integrated H i intensity map over this velocity range is generated using the miriad task MOMENT. Given a specified fitting area (in relation to the source size), imfit is used to determine the central position of the H i source by fitting a two-dimensional Gaussian to the integrated H i intensity map. The obtained central coordinates are used to generate a new spatially integrated spectrum weighted according to the extent of the H i source (see below). In general, we fit and subtract a first-order baseline fit to the spectrum (higher order baseline fits are labelled in the catalogue). The H i parameters such as peak and integrated flux, the velocity widths, the clipped rms noise and the systemic velocity are measured using mbspect. In the catalogue we quote the integrated flux measured with the robust moments, the 20 and 50 per cent velocity widths calculated from the width maximizer and the systemic velocity as the centre of the 20 per cent velocity width (see Meyer et al. 2004 for definitions).

An H i source can be unresolved or extended with respect to the gridded beam size. To identify point and extended sources, we compare the integrated flux measurements obtained from (i) spectra where every pixel is weighted according to the expected beam response (beam-weighted as for point sources) to (ii) spectra where the flux is directly summed (as for extended sources). The considered area is given by a box surrounding the central position, e.g. of size 5 × 5 and 7 × 7 pixel² that correspond to 20 × 20 and 28 × 28 arcmin², respectively. A potentially extended source is identified due to an increase in the integrated flux from a beam-weighted spectrum to a summed spectrum in a bigger box. This increase can also occur due to confusion with a nearby H i source which might be included in the bigger box – such sources are flagged as double sources (d) in the catalogue and are parametrized by generating a beam-weighted spectrum as for point sources. Truly extended sources (e), which are indicated by an increase in the integrated flux, are confirmed by investigating the residual map of the Gaussian fit. An extended source will show flux in the residual map whereas a point source is well described by the Gaussian fit.

3.2 The H i catalogue

The duchamp detection list in the smoothed and reconstructed HIJASS megacube contains 140 candidate galaxies with systemic velocities between 300 and 1900 km s⁻¹. After the parametrization process, four galaxies have systemic velocities below 300 km s⁻¹. In addition, we identified 11 H i sources with systemic velocities between 150 and 300 km s⁻¹ that are included in the sample as they are obviously real galaxies as opposed to Galactic emission or high-velocity clouds. However, H i sources with systemic velocities below 300 km s⁻¹ are not included in Section 4. Furthermore, we identified several double, triple and quad detections that are separated during the parametrization process leading to a total of 173 H i sources that are parametrized from the original HIJASS megacube. We exclude seven detections from the presented catalogue that are unconfirmed in GBT follow-up observations (see Section 2.4). Therefore, a total of 166 H i sources are listed in our catalogue in Section 3.2.

The parametrized H i sources are cross-correlated with galaxies listed in the NASA/IPAC Extragalactic Database (NED; as of 2012 September) and the Sloan Digital Sky Survey Data Release 8 (SDSS DR8). We only consider galaxies with cross-identifications in NED and galaxies with spectroscopic redshifts with confidence levels z_conf > 0.95 in SDSS as possible counterparts. We associate an H i source with a catalogued galaxy, if this galaxy is within the gridded beam size (or within the extent of the H i intensity contours for large sources) and within the velocity range of the H i source. We find 54 H i sources that have multiple candidate counterparts. These are potentially confused H i sources and may be interacting systems. To determine the most likely optical counterpart (listed in the first row of the catalogue), the H i position uncertainty, the heliocentric velocity, the optical galaxy size (in relation to the H i mass and size) and the morphology are considered.

We find only one out of the 166 catalogued H i sources that does not have an associate optical counterpart listed in NED/SDSS or visible on Palomar Digitized Sky Survey (DSS) images. This candidate galaxy/H i cloud – HIJASS J1219+46 – is discussed in Section 3.4.

Table 2 lists the H i sources and their optical counterparts as follows: column (A1) gives the HIJASS name; columns (A2)–(A4) list the central H i position [RA (J2000), Dec. (J2000)] and its uncertainty σ_pos (all uncertainties are discussed in Section 3.3); column (A5) contains the systemic velocity (km s⁻¹) with its uncertainty σ_v; column (A6) states ‘ID’ for H i detections that have an optical identification and ‘PUO’ for the previously uncatalogued gas-rich object; columns (A7)–(A10) list the details of the optical counterparts, i.e. name, RA (J2000), Dec. (J2000) and heliocentric velocity (km s⁻¹); columns (A11) and (A12) give the angular separation (arcmin) and velocity offset (km s⁻¹) from the H i parameters. The most likely optical counterpart is given in the first row and galaxies that lie within the gridded beam or within the extent of the H i intensity contours (for large sources) and might confuse the H i content are quoted in the subsequent rows.

The measured H i properties of the HIJASS detections are given in Table 3. The columns are as follows: column (B1) gives the HIJASS name; columns (B2)–(B4) gives the measured and corrected peak flux densities (S_p) with its uncertainty (in Jy); columns (B5)–(B7) list the measured and corrected integrated flux (S_int) with its uncertainty (in Jy km s⁻¹); columns (B8) and (B9) contain the velocity widths at 50 and 20 per cent of the peak flux density with their uncertainties (in km s⁻¹); column (B10) gives the clipped rms measured in mbspect; column (B11) lists the distance to the galaxy (in Mpc) as obtained from the Extragalactic Distance Database (EDD; Tully et al. 2009 – only distances obtained from colour–magnitude diagrams, tip of the red giant branch and other high-quality methods are considered) or calculated from the Local Group velocity (see Koribalski et al. 2004 for HIPASS) – we adopt H₀ = 70 km s⁻¹ Mpc⁻¹; for galaxies within the cluster definition by Tully et al. (1996) we adopt a distance of 17.1 Mpc; column (B11) gives the logarithm of the H i mass and column (B12) contains flags for 78 extended (e) and 54 confused (c) H i sources, 38 H i sources which are part of duchamp double detections (d), 10 H i sources without a central position from a position fit (f) and therefore without major/minor axis measurements, 10 H i sources lying nearby negative bandpass sidelobes (b), 10 newly detected H i sources in the 21-cm emission line (n), 15 H i sources with systemic velocities below 300 km S⁻¹ (v), 34 sources where independent distance measurements are available (i) and numbers to indicate higher order baseline fits. The H i mass in column (B11) is calculated from the integrated flux [S_int and S_int(corr) for extended sources, respectively] according to |$M\mathrm{_{H\,{\scriptscriptstyle {I}}}}=2.356 \times 10^5 \times D^2 \times S\mathrm{_{int}}$| (M_⊙) with distances as given in column (10). We note that Kovač et al. (2009) also determine adequate distances in the Local Group frame, neglecting peculiar velocities (using Yahil, Tammann & Sandage 1977).

We show a velocity integrated H i map with symbols marking the projected positions of all parametrized H i sources in this catalogue in Fig. 3. H i sources with an associated optical counterpart are marked with open ellipses whereas the previously uncatalogued H i source is marked with a filled ellipse (black). Furthermore, previously catalogued sources that are newly detected in H i are marked with crosses. The size of the ellipses indicate the measured major axis – ellipses surrounded by boxes mark H i detections for which no central position and no extent (major and minor axis) is measurable (usually weak H i sources or double detections for which we retain the optical counterpart position). The projected position of the Ursa Major cluster as defined by (Tully et al. 1996) is overlaid with a large circle.

3.3 H i parameter uncertainties

In Table 2 and 3 we give several uncertainties of measured H i parameters which are calculated as follows.

3.4 Candidate galaxy/H i cloud

The presented catalogue of 166 H i sources in the Ursa Major region contains 162 H i detections, which are clearly associated with previously catalogued galaxies as listed in the NED (with cross-identifications) or the Sloan Digital Sky Survey (SDSS DR8 – redshift confidence level z_conf > 0.95). Another three HIJASS detections – namely HIJASS J1226+27, HIJASS J1243+24 and HIJASS J1233+24 – can be matched up with H i sources listed in the ALFALFA 40 per cent catalogue (Haynes et al. 2011) and have been assigned to IC 3341, KUG 1240+247 and MAPS-NGP O|$\_$|378|$\_$|0206985, respectively. Note that HIJASS J1233+24 was first catalogued in Lang et al. (2003) using the same HIJASS data set and is therefore listed in NED (without cross-identifications). Therefore, only one H i detection – HIJASS J1219+46 – does not have a previously catalogued optical counterpart with known velocity listed in NED, SDSS or ALFALFA (HIJASS J1219+46 lies outside the ALFALFA declination range).

To identify a potential optical counterpart of the candidate galaxy/H i cloud we investigate SDSS and DSS images. We compare the images centred on the central position of HIJASS J1219+46 to the ones of the lowest H i mass and previously catalogued galaxies in the sample (see Fig. 4). One might expect that the candidate dwarf galaxy has a similar optical counterpart as the known galaxies within the same mass bin – however, as it has not been previously catalogued one might also expect it to have lower surface brightness and thus be less easily visible. All of the lowest H i mass galaxies show faint patches of nebulosity in the multicolour SDSS images (Fig. 4). Furthermore, we show HIJASS sources that are newly detected in the 21-cm emission line in Fig. 5. These SDSS images are centred on the H i position and therefore show typical offsets from the optical counterparts. For clarification we invert the colours of the multicolour SDSS images in Fig. 5 (as opposed to Fig. 4). We do not find a similar galaxy in the images for HIJASS J1219+46 within the search radius (the position uncertainty is 1.2 arcmin as given in Table 2). There are multiple galaxies listed in SDSS with no redshift measurements, which match in position to the candidate galaxy/H i cloud. To determine the optical counterpart(s) and/or confirm the H i cloud accurate position is required from follow-up observations. It is also quite possible that deeper optical imaging is required to determine any optical counterparts.

In Fig. 6 we show a 14 × 14 DSS B-band image of the candidate galaxy/H i cloud and its surroundings with overlaid H i intensity contours integrated over the velocity range from 360 to 430 km s⁻¹. HIJASS J1219+46 is a confused source in the duchamp output with HIJASS J1220+46/NGC 4288. The angular separation to NGC 4288 is 03, which corresponds to a distance of ∼40 kpc assuming that the region lies at a distance of 7.6 Mpc. To the north of HIJASS J1219+46 lies HIJASS J1218+46 (SDSSJ121811.04+465501.2) and HIJASS J1218+47/M106 (including NGC 4248, UGC 07356 and two SDSS galaxies). The angular separation to M106 is 07, which corresponds to ∼93 kpc. Although HIJASS J1219+46 lies nearby negative bandpass sidelobes emerging from HIJASS J1218+47/M106 the GBT follow-up observations confirm the H i properties measured in HIJASS as given in Table 4 (the GBT and HIJASS spectra are shown in Fig. 5 – the negative bandpass sidelobes are apparent at 220–360 km s⁻¹ in the HIJASS spectrum). The nature of HIJASS J1219+46 could be tidally stripped gas or a dwarf companion.

Wilcots, Lehman & Miller (1996) observed NGC 4288 using the Very Large Array D-configuration ( hereafter VLA D-array; channel width of 5.2 km s⁻¹ and a rms noise of 1.0 mJy beam⁻¹) finding a gas cloud to the south of NGC 4288 (also see Kovač et al. 2009). Furthermore, the gas cloud shows a tidally disrupted H i stream that extends well beyond the optical disc of NGC 4288 (Wilcots et al. 1996) towards HIJASS J1219+46. However, HIJASS J1219+46 lies just outside the area covered by Wilcots et al. (1996) and Kovač et al. (2009).

3.5 Comparison of measured H i parameters with values in the literature and follow-up observations

We find 165 H i sources with one or more associate optical counterparts as listed in NED and SDSS in the HIJASS data. For H i detections with one, clear associate counterpart, i.e. H i sources that are not confused (c), not double detections (d) and not in the vicinity of negative bandpass (b) sidelobes (as labelled in this catalogue in Section 3.2), we show the position and velocity offsets from our measured H i parameters to previously obtained H i parameters in Fig. 7. We compare our results to recent literature values: (i) Verheijen & Sancisi (2001) extensively studied 43 spiral galaxies in the Ursa Major cluster using WSRT, (ii) Springob, Haynes & Giovanelli (2005) have compiled a homogenous sample of H i properties of more than 9000 galaxies from single-dish observations, (iii) Kovač et al. (2009) have conducted a blind H i survey in the Canes Venatici region using the WSRT and (iv) the ongoing ALFALFA survey using the Arecibo covers a strip in the southern part of the region (+24° ≤ δ ≤ +28°) (Haynes et al. 2011). The congruent sample of clear H i sources in this study and the literature includes 78 galaxies. The central positions and velocities presented in this paper are generally consistent with the literature values within the errors (see Fig. 7).

Some integrated fluxes differ by more than 20 per cent, in particular in comparison to Kovač et al. (2009). This may be explained as they use an interferometric data set (obtained with WSRT) which is not sensitive to diffuse H i emission compared to a single-dish observation.

There are some data points that lie outside the displayed range in Fig. 8 (top): HIJASS J1254+27/NGC 4789A with Diff = 219 per cent in comparison with Springob et al. (2005). We find a wide range of literature values for this dwarf irregular galaxy. NGC 4789A has a large H i to optical radius ratio (Hoffman et al. 1996), therefore some telescopes may resolve out significant flux. There are two outliers in comparison with the GBT follow-up observations – HIJASS J1218+46/SDSSJ121811.04+465501.2 (Diff = 152 per cent) and HIJASS J1145+50/SDSSJ114506.26+501802.4 (Diff = 202 per cent). See Fig. 5 for the comparison of the GBT and HIJASS spectra. HIJASS J1218+46 is a double detection with HIJASS J1218+47 including M106, UGC 07356, NGC 4248 and two SDSS galaxies (also see Fig. 6). Given that the angular separation between SDSSJ121811.04+465501.2 and M106 is 24.5 arcmin and the angular separation between SDSSJ121811.04+465501.2 and UGC 07356 is 14.4 arcmin, these galaxies may contribute a significant amount of flux to the HIJASS detection (FWHM 13.1 arcmin) in comparison to the GBT observation (FWHM 9 arcmin). Regarding SDSSJ114506.26+501802.4, the mean signal-to-noise ratio in the HIJASS data is only 1.5 – therefore the rms noise contributes significantly to the measured integrated H i flux in comparison to the GBT data. In general, this literature comparison shows that H i properties of galaxies measured with different telescopes and obtained with different parametrization procedures lead to consistent results within the errors. However, for H i detections near the detection limit, the noise can cause fluctuations in the H i measurements.

4 DISCUSSION AND RESULTS

4.1 Velocity distribution

The velocity distribution of galaxies residing in the Ursa Major region is shown in Fig. 9. The top panel shows the velocity distribution of galaxies as listed in SDSS (white) with spectroscopic redshifts with confidence levels z_conf > 0.95 and additional galaxies listed in NED (grey) with cross-identifications. To avoid duplicates we cross-correlate the galaxies listed in NED and SDSS within 1 arcmin and 50 km s⁻¹ of the central position and velocity. The bottom panel shows the velocity distribution of the H i-selected galaxies as catalogued in this paper with systemic velocities between 300 and 1900 km s⁻¹. The H i detections are colour-coded according to H i sources with associate optical counterparts (in grey) and the previously uncatalogued candidate galaxy/H i cloud (in black).

One can see the large-scale structure of the region – the overdensity appearing at ∼950 km s⁻¹ in the optical and (less clearly) in the H i-selected sample indicates the presence of the Ursa Major cluster. There are 212 galaxies listed in SDSS and NED within the Ursa Major cluster definition by Tully et al. (1996). In this paper we present 51 H i detections within the cluster definition by Tully et al. (1996) that contain 96 previously catalogued galaxies. The high number of confused detections in our data is due to a high density of galaxies in the cluster. We will discuss cluster/group membership, substructures and HIJASS non-detections in a forthcoming paper but the number of high-probability cluster members is likely to increase significantly from the 79 cluster members identified in Tully et al. (1996).

There are many groups nearby the cluster, which overlap and make the cluster difficult to define. In particular foreground galaxy groups with velocities v < 700 km s⁻¹ lead to an asymmetric galaxy distribution in the optical and H i-selected sample. The number of galaxies with velocities v > 1300 km s⁻¹ is lower than the number of galaxies residing in the foreground. This may be a selection effect or could indicate the filamentary structure: the Ursa Major cluster (at 17.1 Mpc) is connected with the Virgo cluster (at ∼16 Mpc; Mei et al. 2007) via the galaxies and galaxy groups lying at the low velocity end.

Fig. 9 also shows one nearby candidate galaxy/H i cloud. We find no candidate galaxies/H i clouds in the Ursa Major cluster, where one might expect to find some e.g. due to previously uncatalogued gas-rich dwarf irregular galaxies or the presence of stripped H i gas. Given that our H i mass limit at the distance of the Ursa Major cluster is 2.5× 10⁸ M_⊙ (assuming a velocity width of 100 km s⁻¹) we cannot rule out candidate galaxies/H i clouds below the H i mass limit.

4.2 H i mass distribution

Fig. 10 shows the H i mass distribution of 151 H i sources as listed in this catalogue (excluded are 15 galaxies with central velocities below 300 km s⁻¹). H i sources that have been previously detected and are listed in NED/SDSS are shown in grey whereas the candidate galaxy/H i cloud is presented in black. The latter inhabits the low H i mass end and is likely to be a dwarf galaxy, i.e. of the same galaxy population as the previously catalogued galaxies within the same mass bin. Taylor & Webster (2005) note that ‘dark’ (optically invisible) H i galaxies are unlikely to be detected in our studied mass range between 10⁷ and 10^10.5 M_⊙ as the H i gas is normally associated with star formation or is destroyed. Meurer et al. (2006) also find that dormant (non-star-forming) galaxies with H i masses above 3 × 10⁷ M_⊙ are very rare. However, high-resolution follow-up observations are required to determine the nature of HIJASS J1219+46. We discussed the candidate galaxy/H i cloud in more detail in Section 3.4. We note that we did not find a numerous low-mass population of gas-rich dwarf galaxies to solve the dwarf deficiency problem as discussed in Trentham et al. (2001).

We can place an upper limit on the H i mass for the galaxies listed in NED/SDSS within the Ursa Major region, which we did not detect in HIJASS. The survey sensitivity is 2.5 × 10⁸ M_⊙ (assuming a velocity width of 100 km s⁻¹ and a distance of 17.1 Mpc). A lower H i mass limit is valid for galaxies lying in the foreground groups and filamentary structure (with v < 700 km s⁻¹ and distances D < 17.1 Mpc), whereas for background galaxies (v > 1200 km s⁻¹) with distances D > 17.1 Mpc a slightly higher H i mass limit is appropriate (as |$M\mathrm{_{H\,{\scriptscriptstyle {I}}}} = 8.38 \times 10^5 \times D^2$|⁠).

4.3 Candidates for galaxy–galaxy interactions

In this paper, there are 54 H i detections (33 per cent) which have two or more optical counterparts as listed in NED/SDSS within the gridded beam size and within the velocity range of the H i detection (see H i detections labelled with ‘c’ in the catalogue in Section 3.2). We cannot clearly associate the H i content with only one optical counterpart. However, these H i detections with multiple optical counterparts are intriguing objects as galaxy–galaxy interactions might take place due to the small separation between the galaxies (e.g. 10 arcmin corresponds to 50 kpc at a distance of 17.1 Mpc).

For example, HIJASS J1141+32 encompasses four previously known dwarf galaxies within 1700 and 1900 km s⁻¹ (see Fig. 11): Mrk 0746, KUG 1138+327, WAS 28 and SDSSJ114135.98+321654.0 (which has no cross-identification in NED and is therefore not listed in our catalogue) – there are no previous detections in the 21-cm emission line for any of these galaxies. High-resolution follow-up observations are required to study the H i distribution, kinematics and possible interaction of these galaxies.

HIJASS J1220+29 encompasses three previously catalogued galaxies (see Fig. 11) at 610–710 km s⁻¹, namely NGC 4278 (elliptical), SDSSJ122005.73+291650.0 (spiral) and NGC 4286 (spiral) (ordered with increasing angular separation from the central H i position). Morganti et al. (2006) have conducted a WSRT observation of NGC 4278, NGC 4286 and SDSSJ122005.73+291650.0 – the observations reveal two tail-like structures west and south-west of NGC 4278.

Out of the 54 confused HIJASS detections, 33 galaxies appear to have dwarf companions. This leaves us with 21 high-probability candidates for galaxy–galaxy interactions of which 11 HIJASS detections (52 per cent) are within the Ursa Major cluster as defined by Tully et al. (1996). It is remarkable that the Ursa Major cluster as defined by Tully et al. (1996) only comprises approximately 12 per cent of the surveyed volume discussed in this paper but contains about half of all high-probability candidates for galaxy–galaxy interactions. This is also the case (48 per cent) if we include two nearby duchamp double detections, i.e. we do not only consider galaxies within the gridded beam (as the physical size of the beam is smaller for nearby regions) but also merged H i detections as candidates for high-probability galaxy–galaxy interactions. Verheijen & Sancisi (2001) have observed 10 of the high-probability candidates for galaxy–galaxy interactions in the Ursa Major cluster. The sample includes the most massive galaxy (Messier109), the largest (NGC 3718) and some of the brightest systems (NGC 4026, NGC 4111 and NGC 4088).

According to Verheijen & Sancisi (2001) the interacting systems are as follows. (i) NGC 3769 is strongly interacting with its dwarf companion NGC 3769A. (ii) NGC 3893 is an interacting pair with NGC 3896. (iii) Some tidal debris can be seen in the interacting pair IC 0750 and IC 0749. (iv) The largest galaxy in the cluster – NGC 3718 – shows distorted morphology and may be interacting with NGC 3729. (v) NGC 4088 is one of the brightest systems in the Ursa Major cluster with ongoing vigorous star formation. The system is strongly disturbed and in close vicinity of NGC 4085. (vi) The NGC 4111 and (vii) NGC 4026 groups are also known to be interacting and discussed in more detail in Section 4.4.

Three of the high-probability candidates – namely NGC 3917, Messier109 and NGC 4157 – do not show any signs of interactions with their dwarf companions (Verheijen & Sancisi 2001).

Verheijen & Sancisi (2001) find further interacting galaxy pairs: NGC 3949 may be interacting with a small companion (SDSSJ115340.69+475156.4) and UGC 06818 may be tidally interacting with a faint companion – these systems are not included in our sample of high-probability candidates for galaxy–galaxy interactions as they include interactions with dwarf companions. Note that we quote UGC 06818 as a clear single detection in our catalogue – there are no galaxies listed in close vicinity of UGC 06818 in NED/SDSS with know redshifts. Verheijen & Sancisi (2001) also find NGC 3998 with disturbed H i content, which lies just outside the HIJASS data cubes at Dec. δ ≃ 55°27^′13^′′.

4.4 HIJASS detections with extended H i envelopes

We find two H i detections with extended H i envelopes that do not have previously catalogued galaxies within the H i extension.

Within the Ursa Major cluster lies the H i extension/plume in the NGC 4026 group (see Fig. 12), which has already been studied by Appleton (1983; Jodrell Bank), van Driel et al. (1988; WSRT) and Verheijen & Zwaan (2001; VLA D-array). Six galaxies are listed in NED with heliocentric velocities 700 ≤ v ≤ 1100 km s⁻¹, four of which are included in the H i envelope (boxes) and two of which are in vicinity of the H i extensions (triangles; non-detection in HIJASS). The WSRT observations reveal an H i filament at α ≃ 11^h58^m57^s and δ ≃ 50°46^′42^′′ with a systemic velocity of ∼960 km s⁻¹ (van Driel et al. 1988). The H i filament measures approximately 38 kpc in size (which corresponds to ∼7.5 arcmin at the distance of 17.1 Mpc) and contains 2.1 × 10⁸ M_⊙. Furthermore, van Driel et al. (1988) determine the H i mass of NGC 4026 as ≤0.71 × 10⁸ M_⊙, that of UGC 06956 as 5.1 × 10⁸ M_⊙, that of UGC 06922 as 4.9 × 10⁸ M_⊙ and that of UGC 06917 as 1.28 × 10⁹ M_⊙. We detect NGC 4026, UGC 06956 and the H i filament combined in HIJASS J1158+50 and we measure an H i mass of 7.94 × 10⁸ M_⊙ – this matches the total H i mass measured by van Driel et al. (1988) within 1 per cent. Regarding UGC 06922 and UGC 06917, we measure an H i mass of 6.17 × 10⁸ and 1.58 × 10⁹ M_⊙, respectively. One may expect to measure a larger H i flux using a single-dish telescope compared to van Driel et al. (1988) using WSRT. However, the H i content of UGC 06922 is likely confused as UGC 06956 lies in close vicinity.

High-resolution observations of this system conducted with the VLA D-array (Verheijen & Zwaan 2001) reveal an H i filament of ∼97 kpc in size (which corresponds to ∼18 arcmin and indicated by a cross in Fig. 12). The origin of the H i filament is likely to be tidally stripped gas of NGC 4026. Verheijen & Zwaan (2001) note that all three bright lenticular galaxies in the Ursa Major cluster including NGC 4026 have disturbed H i.

Another intriguing H i extension in the Ursa Major cluster is seen in the NGC 4111 system, which has already been studied by Verheijen & Zwaan (2001; VLA D-array). There are six previously catalogued galaxies as listed in NED within 20 arcmin angular separation and within 150 km s⁻¹ of the central H i velocity (see Fig. 12). NGC 4111 is also one of the three brightest lenticular galaxies in the Ursa Major cluster with disturbed H i (Verheijen & Zwaan 2001). The H i extends approximately 122 kpc (which corresponds to ∼28 arcmin at the distance of 15.0 Mpc; Tully et al. 2009 and indicated by a cross in Fig. 12) south of the projected central position of NGC 4111 (Verheijen & Zwaan 2001). Furthermore, the disturbed H i connects NGC 4111 with its close neighbouring galaxies NGC 4117 and NGC 4118.

The third of the brightest lenticular galaxy in the Ursa Major cluster with disturbed H i content according to Verheijen & Zwaan (2001) – NGC 3998 – lies just outside the HIJASS data cubes at Dec. δ ≃ 55°27^′13^′′.

We do not find any H i detections with extended H i envelopes outside the Ursa Major cluster as defined by Tully et al. (1996).

4.5 Comparison to large area blind H i surveys

Blind H i surveys have the potential for detecting previously uncatalogued galaxies and H i clouds. The two largest blind H i surveys to date are HIPASS (Staveley-Smith et al. 1996; Barnes et al. 2001; Koribalski et al. 2004; Meyer et al. 2004) and ALFALFA (Giovanelli et al. 2005; Haynes et al. 2011). While the HIPASS catalogues contain numerous sources without optical or infrared counterpart (e.g. in regions of high Galactic extinction), there is no evidence for any optically dark galaxies (Ryan-Weber et al. 2002; Koribalski et al. 2004). One of the most massive starless H i clouds is HIPASS J0731−69 (⁠|$M\mathrm{_{H\,{\scriptscriptstyle {I}}}} \sim 10^9$| M_⊙; Ryder et al. 2001). It is located near the asymmetric spiral galaxy NGC 2442 in the NGC 2434 group. A possible scenario for its origin is discussed in Bekki et al. (2005). Australia Telescope Compact Array (ATCA) observations by Ryder & Koribalski (2004) reveal several H i clumps that are probably embedded in a common envelope of diffuse H i. A group of H i clouds found around 400 km s⁻¹, including HIPASS J1712−64 (Kilborn et al. 2000), HIPASS J1718−59 and HIPASS J1616−55 (Koribalski et al. 2004), are likely tidal debris related to the Magellanic Clouds and the Leading Arm. In areas of high stellar density and dust obscuration, it can be difficult to detect optical or infrared counterparts to H i-detected galaxies. Apart from the Galactic plane, other sources—like the Magellanic Clouds – may hide the stellar component of galaxies. An example is HIPASS J0546−68 (v_sys = 1306 km s⁻¹) behind the Large Magellanic Cloud (Ryan-Weber et al. 2002).

The ALFALFA 40 per cent survey has detected nearly 16 000 H i sources in the velocity range from −470 to ∼18 000 km s⁻¹ (Haynes et al. 2011); among these are 1013 sources without assigned optical counterparts. Of the latter, 814 are at velocities from −470 to ∼340 km s⁻¹, i.e. likely to be High-Velocity Clouds (HVCs). The remaining are H i cloud candidates, with ∼75 per cent located near galaxies of similar redshifts (e.g. Leo Ring; Stierwalt et al. 2009). About 50 candidates (0.3 per cent) for isolated H i clouds remain; these require follow-up optical and interferometric H i observations to identify their nature.

In our survey of the Ursa Major region we were able to confirm just one H i source without a definite optical counterpart (HIJASS J1219+46; v_sys ∼ 392 km s⁻¹). It is located near the peculiar galaxy NGC 4288 (HIJASS J1220+46; v_sys ∼ 520 km s⁻¹; also see Wilcots et al. 1996; Kovač et al. 2009) in the Canes Venatici region. This H i source is discussed in more detail in Section 3.4. The nature of this H i source could be a dwarf companion or tidally stripped gas. H i imaging and deep optical follow-up will be required to determine the true nature.

5 SUMMARY

The first blind H i survey of the Ursa Major region was conducted in 2001–02 as part of HIJASS. The nearby Ursa Major region is within a velocity range from 300 to 1900 km s⁻¹ and covers an area of ∼480 deg². Using the automated source finder duchamp we identify 166 H i sources in the data based on their peak flux density. The measured H i properties of the sources are presented in a comprehensive catalogue. We address the catalogue completeness and reliability – we find that the catalogue is 95 per cent complete above our peak flux density cut (33 mJy). The H i mass limit of the catalogue is 2.5 × 10⁸ M_⊙ (assuming a velocity width of 100 km s⁻¹ and a distance of 17.1 Mpc).

The catalogue includes 165 H i sources for which optical counterparts are identified using NED and SDSS. We find multiple counterparts within the gridded beam for 33 per cent of the H i detections, which are candidates for galaxy–galaxy interactions. There are 21 high-probability interacting systems – about half of these are within the Ursa Major cluster and have been studied by Verheijen & Sancisi (2001). It is remarkable that the Ursa Major cluster as defined by Tully et al. (1996) only comprises approximately 12 per cent of the surveyed volume discussed in this paper but contains about half of all high-probability candidates for galaxy–galaxy interactions. We discuss four examples of interacting systems, two of which show intriguing extended H i content. The latter have been already been studied with higher resolution and revealed H i filaments and H i distortions of the bright lenticular galaxies.

The HIJASS data contain 10 H i sources that are newly detected in the 21-cm emission line. One of these does not have a visible associate optical counterpart and is likely to be a candidate galaxy/H i cloud. This previously uncatalogued H i source requires follow-up observations to investigate its nature.

We thank the HIJASS team for providing the data. HIJASS was funded by the UK PPARC (Grant No. PPA/G/S/1998/00620 to Cardiff). We also acknowledge the UK PPARC (Grant No. GR/K28237 to Jodrell Bank) that funded the multibeam receiving system at Jodrell Bank. We thank David Barnes and Mark Calabretta for their help with the fake source injection into the raw data. We thank Felix James ‘Jay’ Lockman for his help with the GBT follow-up observations from setting up, execution to the data reduction. We thank Emma Ryan-Weber, Glenn Kacprzak and the anonymous referee for useful comments. This research has made use of the NASA/IPAC Extragalactic Database, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. Digitized Sky Survey material is acknowledged, which was produced at the Space Telescope Science Institute under US Government grant NAG W-2166. This research has also made use of the Sloan Digital Sky Survey (SDSS). Funding for the SDSS and SDSS-II was provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the US Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS was managed by the Astrophysical Research Consortium for the Participating Institutions.

REFERENCES