An $H\alpha$ Imaging Survey of the all (Ultra-)Luminous Infrared Galaxies at $Dec. \ge -30^{\circ}$ in the GOALS Sample

This paper presents the result of $H\alpha$ imaging for luminous infrared galaxies (LIRGs) and ultra-luminous infrared galaxies (ULIRGs). \textbf{It is } a \textbf{complete subsample of Great Observatories All-sky LIRG Survery (GOALS) with $Dec. \ge -30^{\circ}$}, \textbf{and} consists 148 galaxies with $log(L_{IR}/L_{\odot}) \ge 11.0$. All the $H\alpha$ images were carried out using the 2.16-m telescope \textbf{at the Xinglong Station of the} National Astronomy Observatories, Chinese Academy of Sciences\textbf{ (NAOC),} during the year from 2006 to 2009. We obtained pure $H\alpha$ luminosity for each galaxy and corrected the luminosity for $[NII]$ emission, filter transmission and extinction. We also classified these galaxies based on their morphology and interaction. We found that the distribution of star-forming \textbf{regions} in these galaxies is related to this classification. As the merging process advanced, these galaxies tend to have a more compact distribution of star-forming region, higher \textbf{$L_{IR}$} and warmer IR-color ($f_{60}/f_{100}$). \textbf{These} results imply that the degree of dynamical disturbance plays an important role in determining the distribution of star-forming region.

An important question regarding the nature of (U-)LIRGs is what phase do they play in the general evolution of galaxies? Many statistical studies showed that the interaction between galaxies can enhance star formation activity (Kennicutt et al. 1987). And the ULIRGs are associated with the interaction (Zou et al. 1991;Sanders et al. 1988;Wu et al. 1998a,b) as well as active galaxy nuclear (AGN) activities (Wu et al. 1998a,b). Theoretical and observational works support an evolution scenario where two gas-rich spiral galaxies merge first and drive material from galaxies disk toward the merger center, triggering star-formation activity before dust-enshrouded AGN in the circumnuclear region (Sanders et al. 1988; Barnes & Hernquist 1992;Hopkins et al. 2008;Jin et al. 2018).
Former works showed that the degree of interaction has a great influence on the star formation process in ULIRGs. Hattori et al. (2004) performed an Hα imaging observation for 22 LIRGs and found that the distribution of star-formation region is strongly related to the property of galaxy interactions. Many works also showed that the interaction/merger rate increases with IR luminosity, and nearly all (U-)LIRGs show signs of interaction/merger events (Wu et al. 1998b;Kim et al. 2002;Veilleux et al. 2002;Zou et al. 1991). Larson et al. (2016) presented an analysis of morphologies and molecular fraction (MGFs) for 65 LIRGs in GOALS sample. They found that the mean MGF for non-interacting LIRGs is much less than that of intermediate stage in major merger LIRGs.
However, the diversities of star-forming and morphological properties in (U-)LIRGs are not fully understood. For example, Sanders & Mirabel (1996) suggested that 20 ∼ 30% of LIRGs with 10 11 L < L IR < 10 12 L are apparently single galaxies. Rigopoulou et al. (1999) presented a mid-IR spectroscopic survey for 62 ULIRGs and found there is no correlation between merging process and infrared luminosity (L IR ) which is contrary to conventional ULIRGs evolutionary scenario. The numerical simulation showed that the minor merger between gas-rich disks and less massive dwarf galaxies can also produce nuclear starbursts (Hernquist & Mihos 1995).
By using the luminosity (Hα, UV, IR) related to young massive stars, we can get the information of starformation. Among them, the Hα emission is proportional to the number of ionizing photos which are produced by young stars with age less than ∼10Myr and mass higher than 17M (Watson et al. 2016). Therefore, the Hα emission directly traces the presence of recently formed massive stars. Although former studies yield many interesting results, their samples are not enough (Hattori et al. 2004) or biased on one type of galaxies (Theios et al. 2016;Young et al. 1996;Kim et al. 2013). Great Observatories All-sky LIRG Survey (GOALS) 1 is a subset of the IRAS Revised Bright Galaxy Sample (RBGS; Sanders et al. 2003). In order to better understand the property of (U-)LIRGs, we have undertaken an Hα imaging survey of 148 (U-)LIRGs selected from GOALS which will help us to study the star-formation in (U-)LIRGs.
In this paper, we present initial result for our Hα imaging survey on 148 GOALS sample galaxies. The layout of this paper is as follow: In Section 2, the sample selection and observation are summarized. In Section 3, we describe the data reduction. And in Section 4, we present the main results of this survey which include the Hα catalog and reduced Hα images. The results based on morphological classification are presented in Section 5 and the discussion is provided in Section 6. At last, a summary of the paper is presented in Section 7. Throughout this paper, we adopt the cosmology H 0 = 75 km s −1 M pc −1 and a flat universe where Ω M = 0.3 and Ω Λ = 0.7.

SAMPLE AND OBSERVATIONS
2.1. Sample 1 http://goals.ipac.caltech.edu Sanders et al. (2003) provided a complete flux-limited extragalactic sample (RBGS: the IRAS Revised Bright Galaxy Sample) with 60µm flux densities greater than 5.24 Jy and |b|> 5 • . Armus et al. (2009) constructed the GOALS sample from RBGS, which includes 181 LIRGs and 21 ULIRGs in the local universe (z < 0.088). This sample combines data from NASA's Spizer Space Telescope, Chandra X-Ray Observatory, Hubble Space Telescope (HST), and Galaxy Evolution Explorer (GALEX) observatories, together with ground-base data (Armus et al. 2009). Chu et al. (2017) have presented broadband Herschel imaging for entire GOALS sample for all six Herschel bands (PACS bands: 70,100,160 µm;SPIRE bands: 250,350,500 µm). This sample span a wide range of nuclear spectral types and interaction stages which provide an unbiased picture of the (U-)LIRGs in local universe. Out of the original list of 202 sources in the GOALS sample, two objects are omitted. One is NGC 5010, whose L IR drops significantly below the LIRG threshold (log(L IR /L ) = 11.0) due to a revised redshift. The other is IRAS 05223-1908, which is proved as a young stellar object (Chu et al. 2017).
Our (U-)LIRGs sample is a subset of the GOALS sample. Considering the observatory site (around 40 degrees north latitude), our sample include 148 objects with Dec. ≥ −30 • The full sample along with their basic properties is listed in Table 1 . The detailed measurements can be found in Armus et al. (2009). Given the poor IRAS resolution, sometimes there may be a pair or multiple galaxies within IRAS 3 σ uncertainty ellipse. In such case, we visually examined the region covered by the IRAS 3 σ uncertainty ellipse in the continuum-subtracted Hα images (achieved after data reduction ) and include all obvious objects within this region. The different counterparts in the system are represented by N (north), S (south), E (east) or W (west) in Table 1. There are 10 such galaxy pairs in our sample and each contains two counterparts. So, in the end, there are total 158 sources in our sample. Figure 1 shows the sample distribution of L IR and recession velocity (cz). The black solid line represents the GOALS sample and the red dashed line represents our sample. It is clear that two sample have the similar distribution and there is no significant differences between them. The cz concentrates in the range of 4000 km s −1 to 7000 km s −1 corresponding to the redshift from 0.013 to 0.023. Left panel shows the L IR distribution. The number of galaxies decrease rapidly as the L IR increases. Most galaxies in our sample are LIRGs with log(L IR /L ) lower than 12.0.

Observations
The observation were carried out on the 2.16-m telescope at the Xinglong Station of the National Astronomical Observatories, CAS. All the galaxies in our sample were taken in dark night between 2006 February and 2009 June. We used BAO Faint Object Spectrograph and Camera (BFOSC), which had 2048 × 2048 pixel 2 with the pixel scale of 0.30 arcsec and has a field of view (FOV) of 11. × 11. arcmin 2 . We adopted a readout noise of 8.6 e − pixel − with an average gain of 1.65 e − ADU − during the observation. The latest description of updated parameters for BFOSC can be seen in Fan et al. (2016).
To obtain the distribution of star-formation region, we observed with both the narrow Hα filter which cover the shifted Hα emission at the velocity of the target galaxy and the broad R-band filter which used to determine the nearby continuum level. There are a series of narrow band Hα filters, whose center wavelength ranges from 6563Å to 7052Å with the FWHM about 55Å. The detail description of Hα filter can be seen in Lei et al. (2018). The effective wavelength λ ef f of the broad Rband filter is 6407Å.
For each object, the typically integral time is about 600s in R-band and 3600s in Hα. Table 1 lists the Hα filters for each sources used in observation together with all the other observation information.

Image Preprocessing
After the observation, we checked the quality of the images by naked eyes. The subsequent data reduction were performed using IRAF, including overscan subtraction, bias subtraction, flat-field correction. The cosmicrays were identified and removed using L. A.Cosmic (van Dokkum 2001). Then, the celestial coordinate was added to each image using Astrometry.net 2 and the bad columns were replaced with a linear fit of surrounding pixels .
The next step is sky subtraction. The most critical step is the sky background construction. Firstly, Sextractor was employed to detect objects. Before detecting, we produced a gaussian smoothed image by convolving original images with a Gaussian function of FWHM=3 pixels to make the area of objects more extended. If the original image (Figure 2-a) is directly used for Sextractor to detect objects, it may be hard to derive a good object-masked image because the wings of bright objects can not be completely masked (Du et al. 2015). Secondly, we got the object-masked image by subtracting all of the detected objects according to their masked areas from original image. Thirdly, we used method provided by Lei et al. (2018) to get a reliable large-scale structures of the background by using this object-masked image. We applied a median filter of 70 × 70 pixel 2 to convolve the object-masked image in order to reduce the random noise and to fill in the mask regions by surrounding sky region. Unfortunately, due to the fact that our objects are too large, the backgrounds in the center of sources are not filled ( Figure  2-b). We adopted two methods to deal with this problem.
We found that the background of our images had similar pattern in the same filter. So we conducted image combination with a 3 sigma clip to get an average background for each filter, which removed most signatures for the regions masked incompletely, such as the region unfilled as well as the wings of the bright stars. (Figure  2-c). The average-background was then scaled and subtracted from the original image. Figure 2-d shows an example of the sky-subtracted image.
There were still some images with strange pattern that we can not get the background by this method and we made the background as did by ; Wu et al. (2002) and Du et al. (2015). We performed a least-squares polynomial fit of low order to the sky pixels of each row and column and then averaged the line-fitted and column-fitted images to get the averaged background. At last this averaged image (Figure 2-f) was smoothed with a Gaussian function of FWHM=31 pixels (Figure 2-g) and used as sky background. Figure  2-h shows the sky-subtracted example for the second method.
The sky background derived from these two methods both show the vignetting and non-uniformity distribution. Figure 3 (a) (b)shows the fluctuation of two example images for method-1 and method-2. It is clear that the mean values of the Gaussian distribution for images with our sky background subtraction are close to zero and have much less fluctuation than those of the original images.
Then the Hα image were scaled relatively to the continuum R-band images using field stars, and the continuum R-band images were subtracted from the scaled Hα image to yield continuum-free images. In this process, we assume the absence of feature lines on their continua of the field stars. The scaling factors are defined by the ratio between counts of field stars in the wide R-band and narrow Hα band. We adopt the median value first and then adjust the value around until the residual fluxes of foreground stars reached the minimum. Figure 4 shows the Hα band, R-band and continuum-subtracted Hα images of N GC5394/5 as an example.

Photometry
The continuum-subtracted Hα images were flux calibrated using photometry from Panoramic Survey Tele-scope and Rapid Response System (Pan-STARRS). The Pan-STARRS survey is designed for collecting wide-field astronomical imaging and operated by Institute for Astronomy at the University of Hawaii. This survey used a 1.8-m telescope with a 1.4 Gigapixel camera to image the sky in five broadband filters (g, r, i, z, y). The systematic errors in Pan-STARRS photometric system is about 0.02 mag (Tonry et al. 2012). In our work, the Pan-STARRS's PSF magnitude of g-band (m g ) and r-band (m r ) were used to get the Johnson/Cousins Rband magnitude (m R ) by the formula given by Tonry et al. (2012): Then the m R is transformed to flux density with following equation (Oke & Gunn 1983;Frei & Gunn 1994): Then by comparing the field star in our observation and Pan-STARRS, we derived the flux calibration in this observation.
Before the photometry, field stars must be masked in R-band images. The Sextractor was used to find stars across the image and replaced them with the median of background value. The counterparts of galaxy pair were masked in a similar way. When we measure one object, the other is masked, as an example be seen in Figure  5. The continuum-subtracted Hα images also require masking in the case of galaxy pair or the residuals from star were not subtracted clearly. The L IR was also assigned into two counterpart of galaxies pair according to their Hα fluxes ratio. Although the L IR of some separated galaxies does not meet the requirement of LIRGs (10 11 L ), we still include these sources in our sample.
Then, we performed the photometry with IRAF ellipse package. Firstly, we fitted ellipse isophotes to Rband images. The center of the galaxies were determined by the contour map of R-band images. Many objects in our sample show the features of bars, rings, and interaction disturbance. For this reason, the ellipse isophotes were derived by allowing the position angle (P A), ellipticity (e = a−b a ) and galaxy center to vary along the radius during the fitting process. Starting values of ellipticity and position angle were determined by eye from the contour map of the galaxies in R-band. We derived a set of concentric elliptical isophotes which are extend from nuclear region to outskirt of galaxies. When the variation of enclosed flux was close to zero among at last five isophotes, we used this radius as the boundary of the galaxy (radii R). We also get the half-light radius (R e ) at which the enclosed R-band fluxes reach the half of total.
The photometry of Hα was measured using the ellipse isophotes obtained from R-band images. Figure 6 shows the two ellipses for N GC 6926, which enclosed the total flux (red one) and half flux (blue one).
The R-band and continuum-subtracted Hα images are presented for each object in appendix A ( Figure 16).
Because some of the redshifted Hα lines may locate at lower transmission part of filter band, we applied transmission correction to the objects as suggested by . A normalized transmission T (Hα) was used for the correction: where T (λ) presents the transmission curve, T (H α ) is the directly transmission of redshifted Hα emission, the F W HM represent the width of Hα filter curve at half of it's peak value, λ1 and λ2 represent the begin and end wavelength of transmission curve, respectively. The transmission-corrected Hα flux is obtained by dividing the T (Hα). In addition, the flux in R-band also contains the Hα emission, which will result in underestimate for Hα flux in the process of continuum subtraction. Such loss (4%) was estimated by Lei et al. (2018) and corrected for our objects. Wu et al. (1998b) performed a spectroscopic observation of 73 LIRGs and provided ratio of [N II]/Hα of this sample. We took the mean value (0.55) of these LIRGs and then used it to correct the [N II] emission in our sample.  The Galactic extinction was corrected by using the Schlegel et al. (1998) map and the extinction curve from Fitzpatrick (1999). There are several methods for estimating the intrinsic extinction of Hα. Young et al. (1996) derived this correction from [SIII]/Hα ratio. Theios et al. (2016) assumed a correction of 1 magnitude based on L Hα -SF R relation. The internal extinction in work of Gavazzi et al. (2012) were performed by using Balmer decrement. As there were no spectral emission lines which we can used for internal extinction correction, we adopted the mid-IR luminosity to estimate the intrinsic extinction of Hα flux. Zhu et al. (2008) presented a correlation between spitzer 24 µm mid-infrared and extinction-corrected Hα luminosities for star-forming galaxies. By combing the Hα emission line and 24 µm measurements for nearby galaxies, Kennicutt et al. (2009) got the similar relation and derived the formula. We chose their formula as: where L(24) is spitzer mid-infrared luminosity (MIR) at 24 µm (here we adopted as IRAS 25 µm instead) and the L(Hα) obs is observed Hα luminosity without internal extinction correction.  The main errors of Hα fluxes include photometry and continuum-subtraction. The photometric errors due to the Hα photon counting noise and background noise are typically smaller than 4%. The scaling factor of continuum-subtraction is the dominant source of uncertainty. Even small uncertainties in scaling factor can result in large uncertainties in Hα flux with relatively weak Hα emission. We produced continuum-subtracted Hα images with a range of scaling factor. And then the accuracy of scaling factor was estimated by the value at which the continuum-subtracted Hα image are clearly oversubtracted and undersubtracted. The typical errors in continuum subtraction is around 25% and in few exceptional case, this error reaches 70%. By the way, the errors of the internal extinction correction is mainly composed of two parts. One is the uncertain of extinction correction formula (typically 15%) and another is the errors of IRAS 25µm fluxes (typically 5%).

Hα IMAGING RESULTS
In this section, we present the primary results of Hα imaging observation. The Table 2, together with Figure  16, constitutes the main results of our observation.

Hα catalog
The Hα photometry result of 158 galaxies are listed in Table 2. Both Hα Luminosity before and after internal extinction correction are given, as well as the ratio between Hα flux enclosed in R e and that of total galaxy.
Column (1): Source name; Column (2): Type. -The morphology and interactions type of (U-)LIRGs (the detail description will be given in next section); Column ( Column (5): F rac(Hα). -The ratio between Hα flux enclosed inside R e and the total; Column (6): P A. -The adopted position angles at galaxies boundary (radii R).
Column (7): e. -The adopted ellipticity at galaxies boundary (radii R). Figure 7 shows a comparison of Hα emission flux measured by us with those of Young et al. (1996). The objects of Young et al. (1996) were measured without the correction of [N II] emission, internal galaxies extinction and Galactic extinction. In the comparison, we do the same steps as their work and give the comparison result. All objects show a good agreement around 0.24 dex.  Comparison of Hα fluxes for galaxies in our work with Young et al. (1996).

Hα images
These images were reduced by using standard IRAF task. The WCS parameters in the FITS header were added using Astrometry.net. We calibrate the images by adding the flux calibration scale value to the imaging header as "scale". The counts value can then scaled to flux (erg cm −2 s −1 ) by multiplying this value. This "scale" value also include the correction for the Hα filter transmission and 4% underestimate for Hα flux. We didn't make [N II] emission, Galactic and internal galaxies extinction corrections for the "scale" value. These images are listed in order of object name and the solid line on the R-band images represent 10 . All these images in FITS format can be download via the ApJS website.

Morphology Classification
We divided the sample into several morphological classes in order to understand their role in the evolution of galaxies. We made our own morphological classification based on R-band as follows: S (Spiral) -spiral galaxies with symmetrical disk and shows no signs of tidal interaction; P M (Pre-Merger) -two galaxies can be separated with asymmetrical disks or tidal tails, which could be the phase before merging; M (Merger) -galaxies contain two nuclei with tidal tail or the galaxies are disturbed severely which is associated with most violent dynamical events; LM (Later Stage of Merger) -single nucleus with short faint tidal tail which may in the late stage of merger; E (Elliptical Galaxy) -elliptical galaxies with an approximate ellipsoidal shape and without tidal interaction, which could be in final phase of merger; UN (Uknown) -objects can not classified by their morphology clearly.
Examples of different morphological classes are given in Figure 8. The classification was done independently by different people and a consistent classification was adopted after deliberated discussions. In our classification, the E-type occupies the smallest fraction (2.5%) among morphological types. The M -type occupy the largest fraction (39.2%). (U-)LIRGs in S-type, P Mtype and LM -type occupy the percentage of 10.8, 20.9 and 11.4, respectively. The U N occupies 15.2%. The Figure 9 shows the distribution of morphological class in histograms.

Infrared Luminosity
The Figure 10 shows the distributions of L IR . The black lines show the distribution for the whole sample. The distributions for other morphology types are also given in this figure (S: orange, P M : green, M : blue, LM : purple and E: red). As can be seen, the S-type appears to be skewed toward smaller value and none of them have L IR larger than 10 11.65 L , which is consistent with previous works Lam et al. 2015;Larson et al. 2016).
The median L IR in P M -type, M -type, LM -type and E-type is showed in each panel of Figure 10. And as the merging process advanced the objects have a tendency to have relatively extended tail toward larger L IR . Ishida et al. (2004) showed the same result in their study of 56 LIRGs that the separation between merging galaxies decrease as IR luminosity increases.

Hα Luminosity and Concentration
In Figure 11, the histogram of F rac(Hα) is showed for each morphological type. It is clear that for most (U-)LIRGs, the star-forming is dominated by the central region with F rac(Hα) > 0.5. The S-type have a moderate concentration among all types with a median value of 0.77.
The F rac(Hα) is also expected to be higher following the advancing of merging process (Bryant & Scoville 1999;Hattori et al. 2004). The F rac(Hα) of P Mtype is the minimum, and increase along the merging sequence from M-, LM-to E-type. The median value of F rac(Hα) is showed in each panel of Figure 11. Figure 12 shows the profile for various morphology types. The 10 edge-on galaxies (S:2; I:6, M:1, LM:1) are not involved. In the direction of intensity, we normalize the profile with center intensity. In the direction along the galaxy's radii, we normalize the profile with galaxy boundary radii R. Then we combine the profiles according to their morphology types.

Hα Profile
The P M -type are characterized by exceptionally extend profile. The S-type also shows some extensions in the outer region. The P M -type galaxies may be dynamically young system which are predecessors for advanced merger stage. On the other hand, the E-type galaxies which are relaxed from interaction without sign of interaction are the most compact one. The M -type and LM -type are similar and have the intermediate profile among others.
By using the mid-infrared emission of LIRGs which can show the structures for different merge stage, Hwang et al. (1999) found out that the peak-to-total flux ratios of LIRGs increase as projected separation of interacting galaxies become smaller. The profile as well as F rac(Hα) in our study are consistent with previous studies that (U-)LIRGs tend to have the more concentrated star-formation distribution as the merging process advance (Bryant & Scoville 1999;Hattori et al. 2004).

Infrared Color
The infrared colors have been interpreted by various models (Helou 1986;Sekiguchi 1987). A cool component temperature (20K) is used to represent the emission from dust in infrared cirrus heated by older stellar population and peaks at λ 100−200µm. A warmer component temperature (30 ∼ 60K) represents the starburst in galaxies and peaks near 60µm. And a even warmer component peaking around 25µm, represents the dust emission heated by AGN.
The distributions of IR color (log(f 25 /f 60 ) and log(f 60 /f 100 ) ) is showed in Figure 13. The E-type have a warmer f 25 /f 60 which may indicates they host an AGN. The N GC1068, a spiral galaxy, which also has warmer f 25 /f 60 (-0.35) is a well-known Seyfert2 galaxy. The rest objects don't have clearly tendency in f 25 /f 60 . In addition, (U-)LIRGs tend to have warmer log(f 60 /f 100 ) as the merging process advanced. The median values of both log(f 60 /f 100 ) and log(f 25 /f 60 ) are also showed in each panel of Figure 13. The GOALS sample has become a "reference sample" for studying the properties of (U-)LIRGs in the local universe. Extensive multi-wavelength (radio to X-ray) imaging and spectroscopic data have been obtained for  In this work, we performed an Hα survey for a complete GOALS subsample with Dec. ≥ −30 • . After continuum-subtracted, we obtain 148 pure Hα emission images, which can provide the star-formation distribution for this GOALS subsample. We also provide a relatively complete Hα photometric data for GOALS subsample for the first time. In sum, this survey provides the imaging and photometry component which is a useful data addition to the GOALS data archive, and is helpful in revealing the formation and evolution of (U-)LIRGs.

Comparison with Other Morphological Classifications
Though there are many previous works focus on morphological classifications (eg. Haan et al. (2011); Kim et al. (2013) and Larson et al. (2016) ), their data are not completely covered objects in this work and the classification results cannot be directly used. Here we don't use the same classification criterion as they do. Previous studies of LIRGs morphology either relied on HST higher resolution (∼ 0.1 ) imaging (Larson et al. (2016)) or mainly focus on merger stages (Kim et al. (2013) and Larson et al. (2016) ). Such as there are no distinction between spirals and ellipticals in the classification of single galaxies in Larson et al. (2016) and Kim et al. (2013)'s work. Considering these factors, as well as intending to distinguish the merge stage clearly, we adopt a simple approach focusing on the classification for most important merge stages.
To ensure the reliability of morphology classification in this work, we compare our results (J19) to that of Larson et al. (2016) (L16) and Kim et al. (2013) (K13) in Figure ?? and Figure 15 with method provided by Larson et al. (2016). It should be noted that some merge stages, like minor merge (m) in L16, have no suitable analogs in our works. Moreover, because of different criteria for classification, as well as difference in division of stages, there will be cases where one merge stages corresponds to multiple adjacent stages. For example, the merge stage M in our work corresponds to M3 and M4 in L16, and merge stage M4 in L16 correspond M and LM in our work.
In Figure 14 and Figure 15, we mark the cells as green when the corresponding merger stages are agree between two works. The classifications that are shifted by only one stage to earlier or later stage are marked as yellow. Considering the morphology classification focusing on different characteristics, we treat this case as a consistent result. When the classifications differ more than one stages, they are marked as red. There are total 63 objects are both included in our work and L16. But   Figure 14 that our classification agrees fairly well with those of L16 as 79.5% objects have very consistent classification (green) and 12.2% objects have a slight change (yellow). And there are still 4 objects require a change more than a single stage. Overall, our classification are very consistent with that of L16. The 4 objects which require a change more than a single stage are described in the appendix B. 61 objects in our sample are also previously classified by K13 (Figure 15). 7 of them don't have very certain classification in our work, and at last there are 54 objects in comparison. The result of Figure 15 also shows that our classification are consistent with that of L13 ( 92.6 % are roughly the same: 46.3% objects have very consistent classification. 46.3% objects have a slight change between our work and K13). The 4 objects which require a change more than a single stage are described in the appendix B. The reason why a few objects differ in our classification with that of L16 or K13 is due to different resolution of image and subjective factors. Finally, in order to maintain the consistency of our sample, we don't change our classification during the fellow analysis.

Isolated spiral LIRGs
In our study, the local (U-)LIRGs are dominated by merging system and the S-type only occupies 10.8%   As can be seen in Figure 10, the L IR of spirals galaxies in our sample are no more than 10 11.65 L which is consistent with previous works. Wang et al. (2006) also found that none of their spiral LIRGs have L IR higher than 10 11.6 L . Both our and Wang et al. (2006) results suggest that the infrared luminosty of all the local S-type LIRGs are much lower than the boundary of ULIRGs (L IR = 10 12 L ). And in Lam et al. (2015)'s work, the spiral galaxies also tend to have lower L IR . Larson et al. (2016) presented an analysis of morphologies for 65 LIRGs in GOALS sample. They found that the sources with log(L IR /L ) ≥ 10 11.5 are dominated by major mergers between gas-rich spirals, and all ULIRGs are late-stage merger. All these results mean that the spiral LIRGs tend to have lower L IR than those in merging systems in local universe. The objects in Stype also have moderate concentration of star-forming distribution in our sources. It seems that in the local universe, the galaxies have less gas than their counterpart at intermediate redshift. And without interaction, such kind of galaxies can not enhance the extreme starburst. At intermediate redshift, the L IR of normal disk galaxies could be higher, because of their more gas and corresponding more extensive disk star formation (Reddy et al. 2006). The decrease of disk gas could be one of the keys to explain the decrease in fraction of spiral galaxies in LIRGs from intermediate redshift to local universe.

Merging sequence
As normal spiral galaxies can not reach a higher L IR , a merger/interacting process is needed to induce a extreme nuclear starburst. Here we explore the possible merging sequence according to our classification.
When two gas-rich spiral galaxies start their interaction, the tidal torques begin to lead a inflow of gas from outer region to central region. At this stage (P Mtype ), star-formation occurs in the both nuclear and outer region. The objects also tend to show a relatively low concentration of Hα ( Figure 11) and a more extended Hα profile ( Figure 12) with lower L IR and colder IR-color. It is consistent with evolutionary sequence described by Hattori et al. (2004). In their study, the objects in the early stage of interaction have a significant star formation contribution from outer region. Scoville (2001) showed that the interaction/merging increases the colud-cloud collision which lead a transport of molecular from ISM to nuclear region and triggers the star burst in the overlap region of galaxies disks.
As the merging process advanced (M -type and LMtype), two nuclei of galaxies are closer. The gas continues to fall into the galaxy center and fuel the intense star-formation activity. As a result, the starformation activities become more active and begin to concentrate toward galaxy center. The results of previous section indicate that these objects in this stage have more concentrated star-formation region, higher L IR and warmer dust temperature (larger f 60 /f 100 ) which is caused by more intensive star-formation activity. Wu et al. (1998b) showed the same result that both infrared luminosity and Hα equivalent width (EW) increase as galaxy-galaxy nuclear separation decreases. In the work of Lutz (1992) and Hattori et al. (2004), they both showed a warmer IR-color in this later merging stage. Many works also showed that in later merging stage, the distribution of star-formation gradually towards the center (Mihos & Bothun 1998;Hwang et al. 1999;Xu et al. 2000;Hattori et al. 2004). All these evidences indicate a intense nuclear starburst in this later merging stage.
In the last stage (E-type), the objects need to take some time to relax and toward to elliptical morphology. Such they not only shows the most concentrated distribution of star-formation region, highest L IR and warmest f 60 /f 100 color but also have statistically warmer f 25 /f 60 color. This means E-type may have strong and concentrated star-formation activities as well as an AGN in their galaxies center . It is a stage of coexistence of both star formation and AGN. This result is consistent with classical evolution from (U-)LIRGs to QSOs.
In a word, this work shows that as the merging process advanced from PM-, M-, LM-to E-type, the objects tend to present higher L IR , more concentrated starformation and warmer IR-color. All these properties support the evolutionary sequence in the (U-)LIRGs of many former works (Sanders et al. 1988; Barnes & Hernquist 1992;Bryant & Scoville 1999;Hopkins et al. 2008;Jin et al. 2018).

SUMMARY
In this paper, we have presented Hα imaging observation for a complete subsample for GOALS with Dec. ≥ −30 • . The observation was carried out using 2.16-m telescope at the Xinglong Station of the National Astronomical Observatories, CAS, during the year from 2006 to 2009. The data present here are that so far most complete Hα imaging survey of the GOALS sample. For many of these objects, this paper presents the first imaging data and photometry of Hα emission.
1) There are total 148 (U-)LIRGs were observed during the Hα imaging survey. Given there are 10 galaxies systems, our sample contain 158 galaxies at last. The subsequent data reduction mainly contains sky-background, continuum-subtracted, flux calibration, photometry and the correction of [N II] emission, filter transmission, galactic extinction and internal extinction. Finally, we obtained the Hα images ( Figure  16) and luminosity catalog (Table 2) for this sample.
2) We have visually classified our sample using a simplified classification which includes: S(spiral), PM(Pre-Merger), M(Merger), LM(Later Stage of Merger) and E(Elliptical). After compare our classification with previous works, we find that our classification is consistent with those of others.
3) The fraction of spiral galaxies is lower in LIRGs compare to their counterparts in higher redshift. The lower L IR in local spiral galaxies also indicate that interaction between galaxies is need to induce a extremely L IR in local universe. 4) We also found that the advanced merging objects tend to have concentrated star-formation distribution , higher L IR and warmer far-IR color. All these results are consistent with the model that merger drive gas inward toward the nucleus and the star formation activity will be concentrated and enhanced as the merging process advances.
We are grateful to the referee for their careful reading of the paper and useful comments and suggestions.  By the way, we also show the R-band and continuum-subtracted Hα images for each object in appendix A, in order of object name. The solid line on the R-band images represent 10 and the object name is noted in the top center of images.  In this section, we provide further details and discussion for the classification of 8 objects which are not exactly matched with L16 or K13. We will state why our classification differs from L16 or K13, and the reasons for our classification. The images for these objects are listed in Figure 17 IRAS F01364-1042 classified as E in our work based on the detection of a single compact nucleus and there are no signs of tides can be distinguished in our image. This object is classified as M3 in L16 based on disturbed disk with small projected nuclear separation. The difference between this two classification is due to difference in resolution, and ultimately, we retain our original decision for consistency among our sample.
CGCG 052-037 classified as LM in our work based on it seems to have some tidal structure in this galaxy. This object is classified as s in L16 based on it's appearance of a single object with no clear sign of interaction. The difference in classification for this objects is more due to subjective.
ESO 602-G025 classified as M in our work based on it seems have some interaction structure which may be due to a minor merge with a small galaxy. This object is classified as s in L16 based on they think there is no clear sign of an interaction. The difference in classification for this objects is more due to subjective.
UGC2982 classified as LM in our work based on it's tidal structure. This objects is classified as s in L16. Combined with the Hα image, we believe that this object is more likely to be in an interaction.
IC5298 classified as LM in our work based on single nucleus with faint tidal tail. This objects is classified as 0 (single undisturbed galaxy, shows no signs of tidal interaction) in K13. In L16, this object is classified as m (minor merge). They think there is a small companion in SW and they are connected together with a tidal tail. In a way, our result is consistent with L16.
IRAS 20351+2521 classified as M in our work based on it's clearly disturbed disk. L16 don't contain this object and K13 classify this object as 0.

IRAS F17138-1017
This object is not contain in the work of L16. We classify this object as M based on it's disturbed disk. But K13 classify this object as 0. The difference in classification for this objects is more due to subjective.
NGC695 classified as M in our work based on it's disturbed disk. This objects is classified as 0 in K13. In L16, this object is classified as m (minor merge). They think there is a minor companion NW of the main galaxy along with appearance of a tidal perturbation. In a way, our result is consistent with L16. Figure 17. The details and justifications for the classification of objects which require a change more than a single stage when compared with L16 and K13. The panel (a) lists the objects compared to L16, and the panel (b) lists the objects compared to K13. The left image in each row is derived from L16 (IRAS 20351+2521 and IRAS F17138-1017 don't have the images from L16), and the classification type made by L16 or K13 are marked. The middle image is the R-band image in this work, and the classification type made by this work are marked. The right image is the Hα image in this work.