Star formation and dust extinction properties of local galaxies as seen from AKARI and GALEX

Abstract

An accurate estimation of the star formation-related properties of galaxies is crucial for understanding the evolution of galaxies. In galaxies, ultraviolet (UV) light emitted by recently formed massive stars is attenuated by dust, which is also produced by star formation (SF) activity, and is re-emitted at mid- and far- infrared (IR) wavelengths. In this study, we investigate the star formation rate (SFR) and dust extinction using UV and IR data. We selected local galaxies which are detected at AKARI FIS 90 μm and matched the IRAS IIFSCz 60 μm select catalog. We measured FUV and NUV flux densities from GALEX images. We examined the SF and extinction of Local galaxies using four bands of AKARI. Then, we calculated FUV and total IR luminosities, and obtained the SF luminosity, L_SF, the total luminosity related to star formation activity, and the SFR. We find that in most galaxies, L_SF is dominated by L_dust. We also find that galaxies with higher SF activity have a higher fraction of their SF hidden by dust. In fact, the SF of galaxies with SFRs > 20 M_⊙ yr⁻¹ is almost completely hidden by dust. Our results boast a significantly higher precision with respect to previously published works, due to the use of much larger object samples from the AKARI and GALEX all sky surveys.

1. Introduction

The evolution of galaxies is one of the most fundamental problems in modern observational cosmology. Since all heavy elements (elements whose atomic number is larger than Boron) have not been produced by Big Bang Nucleosynthesis but rather by stars, the investigation of star formation is related to the quest for understanding the origin of the Earth, planets, and ourselves.

An accurate estimation of the star formation-related properties of galaxies is crucial for an understanding of the evolution of galaxies. The total mass of newly-formed stars in a galaxy per year is referred to as the star formation rate (SFR) (Takeuchi et al., 2010). Massive stars are known to be good indicators of star formation (SF) activity, since they –have a much shorter lifetime (~10^6–8 yr) than the age of galaxies and the Universe (~10¹⁰ yr) and therefore are regarded as an “instantaneous” indicator of the SFR in galaxies.

Massive stars (OB stars) are hot and emit ultraviolet light. The UV spectra is dominated by emissions from massive stars. Here, we can in principle obtain the SFR of galaxies directly by measuring their UV luminosity. However, as mentioned above, stars produce heavy elements (or metals ^{1 1}) and release them through explosive phenomena during the final phases of stellar evolution such as planetary nebulae, supernova explosions, and some other mass ejection processes (e.g., Asano et al., 2013).

A significant fraction of metals are in the form of tiny solid grains (the typical size is smaller than 1 μm) in the ISM, referred to as dust (Mathis, 1990). This means that SF activity in galaxies is always accompanied by dust formation, except for the formation of the very first stars in the Universe, and the dust grains are gradually mixed with the ISM. The UV photons from massive stars are easily absorbed and/or scattered by dust grains and re-emitted as mid- and far-infrared photons. This is referred to as dust extinction. We stress here that if we only measure the UV photons from massive stars in galaxies, SF activity can be seriously underestimated because a significant amount of the energy is obscured by dust (e.g., Kennicutt, 1998). Indeed, Takeuchi et al. (2005a) have shown that a significant amount of the cosmic SFR density is obscured by dust and can only be observed through far-IR radiation. They found that the fraction of hidden SF increases from 50–60% at z = 0 to >90% at z = 1. This finding has been confirmed and further explored by recent studies (e.g., Murphy et al., 2011; Cucciati et al., 2012).

Many attempts have been made to explore star formation in UV and IR to obtain an unbiased view of star formation in the Universe (e.g., Martin et al., 2005; Seibert et al., 2005; Cortese et al., 2006; Buat et al., 2007; Lee et al., 2009; Noll et al., 2009; Bothwell et al., 2011; Haines et al., 2011).

In order to estimate the SFR of galaxies accurately, it is ideal to combine UV-related observables and dust-related ones. Recently, various methods on this matter have been proposed (e.g. Iglesias-Páramo et al., 2006; Kennicutt et al., 2009; Calzetti et al., 2010; Hao et al., 2011, among others). In this study, we adopt the simplest SFR estimator: a combination of SFRs estimated from UV and FIR continuum radiation (Iglesias-Páramo et al., 2006). To this aim, we constructed a new dataset from GALEX (UV) and AKARI (IR) observations. Thanks to these two All-Sky surveyor satellites, we have one database which contains an unprecedented amount of UV-IR data of Local galaxies. Making use of this new database, we explore the SF- and extinction-related properties of galaxies in the Local Universe in a similar manner to a previous work of Takeuchi et al. (2010) which used the beta version of AKARI data.

This paper is organized as follows: in Section 2, we introduce the AKARI and GALEX data and explain the construction of the infrared-selected IR-UV dataset. In Section 3, we describe the basic results of this study, and in Section 4 we interpret the UV and IR properties of our sample galaxies. Section 5 is dedicated to a summary and conclusions.

Throughout this paper we will assume Ω_M0 = 0.3, Ω_Λ0 = 0.7 and H₀ = 70 km s⁻¹ Mpc⁻¹. The luminosities are defined as νL _ν and expressed in solar units assuming L_⊙ = 3.83 × 10³³ erg s⁻¹.

2. Data

2.1 AKARI

AKARI was launched by JAXA (Japan Aerospace eXploration Agency) in February 2006 (Murakami et al., 2007) ^{2 2}. AKARI was equipped with two imaging instruments, the Far-infrared Surveyor (FIS: Kawada et al., 2007) and the Infrared Camera (IRC: Onaka et al., 2007), together with a Fourier spectrograph (FTS: Kawada etal., 2008).

Before AKARI, the Infrared Astronomical Satellite, IRAS, performed the first All-Sky survey at mid- and far-IR (MIR and FIR) wavelengths. The IRAS All-Sky survey has yielded a vast amount of statistics for dusty galaxies in the Local Universe. The survey provided a point source catalog (IRAS PSC) which has long been used for extragalactic studies (see e.g., Soifer et al., 1987). IRAS covered mid-and far-IR wavelength bands: 12 μm, 25 μm, 60 μm, and 100 μm. Though the IRAS PSC has made a revolutionary impact on the studies of IR objects, the lack of wavebands at λ > 100 μm restricted the range of application to dusty object studies.

This is one of the reasons why AKARI was designed to perform an IR All-Sky survey, especially at wavelengths longer than 100 μm. The AKARI FIS All-Sky Survey was carried out with four photometric wavebands at FIR, centered at 65 μm (N60), 90 μm (WIDE-S), 140 μm (WIDE-L), and 160 μm (N160), with abetter sensitivity and angular resolution than that of IRAS. The point source catalogs and diffuse maps have been gradually made public to the astronomical community.

2.2 GALEX

The UV satellite GALEX (Galaxy Evolution Explorer) was launched by NASA (the National Aeronautics and Space Administration) as one of the SMEX (Small Explorer program) missions in April 2003 ^{3 3}. GALEX has two photometric bands, FUV (1350–1750 Å, λ_mean = 1530 Å) and NUV (1750–2750 Å, λ_mean = 2310 Å). GALEX data products include a series of All Sky surveys and deep sky surveys in the imaging mode, and partial surveys in the NUV and FUV spectroscopic modes. In this study, we use the data of GALEX releases GR4/GR5 All-Sky imaging survey (AIS) with detection limits of 19.9 mag and 20.8 mag in AB system (Morrissey et al., 2007).

2.3 Sample construction

2.3.1 Matching AKARI BSC with IRAS IIFSCz

As a basis for the construction of our galaxy sample, we used the AKARI FIS bright source catalog (BSC) v.1 from the AKARI All-Sky survey (Yamamura et al., 2010). The sources in the AKARI FIS BSC v. 1 have an S/N ratio three times better than the previous catalog, AKARI FIS BSC β1 (Yamamura et al., 2008), which was used by Takeuchi et al. (2010). This provides various advantages for the analysis of this work. Since AKARI BSC sources contain all kinds of IR objects, such as AGBs, Vega-like stars, Hii regions, planetary nebulae, etc. (Pollo et al., 2010), the first step of this study is to construct a catalog of galaxies. In order to have a secure sample of galaxies with redshift data, we made a cross-match of AKARI sources with the Imperial IRAS-FSC Redshift Catalogue (IIFSCz), a redshift catalog recently published by Wang and Rowan-Robinson (2009). The IIFSCz is based on the IRAS Faint Source Catalog (FSC). In the IRAS FSC, the flux density quality (FQUAL) is classified as high (=3), moderate (=2) or upper limit (=1). The IIFSCz has FQUAL = 3 and a signal-to-noise ratio > 5 at 60 μm. Wang and Rowan-Robinson (2009) selected galaxy candidates by using the following color (flux ratio) conditions: (i) log(S₁₀₀/S₆₀) < 0.8 (if FQUAL ≥ 2 at 100 μm), (ii) log(S₆₀/S₁₂) > 0 (if FQUAL ≥ 2 at 12 μm) ^{4 4}. This catalog contains 60, 303 galaxies, and 90% of them have spectroscopic or photometric redshifts at S₆₀ > 0.36 Jy. In this sample, we put a limit on the recession velocity of v > 1000 km s⁻¹, in order to avoid the effect of the peculiar velocities of galaxies. Then, we matched the AKARI BSC sources with these IIFSCz galaxies using a search radius of 36 arcsec, which corresponds to the position uncertainty of the IRAS catalog. If there are multiple counterpart candidates in the search radius, the closest one is chosen. After concluding all of the steps described above, we obtained 6674 IR galaxies and used them for the GALEX photometry.

2.3.2 GALEX photometry

We measured FUV and NUV flux densities for the 6674 galaxies from GALEX images by using a software package developed specially for this purpose. Details of this package are explained in Iglesias-Páramo et al. (2006).

We performed the UV photometry as follows:

• (1) Cut out a 20′ × 20′ subimage from GALEX GR4/GR5 images around each AKARI galaxy.

• (2) Select the subimage with the largest exposure time.

• (3) Measure the FUV and NUV flux densities from the imagelet.

  • (a) Define the center, major axis, and minor axis to fit the ellipse using the NUV subimages and apply the fitted regions to FUV subimages.

  • (b) Define the sky background level around the major axis.

  • (c) Expand the ellipsoidal measured range without changing the axis ratio from the center.

  • (d) Stop to measure when the flux density does not change more than the background.

The FUV and NUV flux densities are corrected for Galactic extinction by extracting the E(B –V) values from the Schlegel Map (Schlegel et al., 1998) and assuming the Galactic extinction curve calculated by Cardelli et al. (1989).

IR objects which did not have a counterpart in FUV and NUV were excluded from further analysis. The images which contain stars in the considered region or partial galaxies (due to the GALEX image limit), were also omitted from subsequent research. This process reduced the number of selected galaxies in the parent sample to 3981. Among them, there are galaxies which have flux densities below the detection limits of GALEX AIS at NUV and/or FUV. In such a case, we replaced these unreliable values with the flux density detection limits. Therefore, the effective number of galaxies which satisfied the completeness criteria was reduced to 3567, and these were used for our analysis.

3. Basic Results

3.1 Luminosity of galaxies

First of all, we calculated the luminosity of sample galaxies from their measured flux density. The flux density S _ν at the frequency ν and the monochromatic luminosity L _ν (the luminosity per frequency) of an object are related through the following equation:

()(1))

where d_L(z) is the luminosity distance of an object at red-shift z. Given the Hubble parameter H(z) at redshift z and the velocity of light c, d_L (z) is expressed as

((2))

where

((3))

Then we obtain,

((4))

where

((5))

The AKARI- and GALEX-band luminosities are calculated with the following formulae because of the different definitions of AKARI and GALEX photometry.

((6))

((7))

Here, Δν (AKARI band) stands for the frequency range of the AKARI bands, and ν(@GALEX band) stands for the effective frequencies of the GALEX bands.

3.2 Total IR luminosity

We first obtained the total IR (TIR) luminosity L_tir from AKARI FIS bands. Hirashita et al. (2008) proposed a formula to estimate the TIR luminosity at λ = 3–1000 μm using the flux densities of three AKARI bands: N60, WIDE-S, and WIDE-L, for the AKARI pointed observation sample of dwarf star-forming galaxies and blue compact galaxies. Their formula was examined and confirmed to be valid for the AKARI FIS All-Sky Survey galaxies by Takeuchi et al. (2010). However, since the flux density at N60 was noisier than in the widebands, Takeuchi et al. (2010) found that an alternative formula which makes use of only WIDE-S and WIDE-L flux densities gives a tighter relation than that for three bands. Hence, in this study we adopted the formula of Takeuchi et al. (2010) to estimate TIR luminosity:

((8))

where

((9))

The conversion formula from to L_TIr is

((10))

The distributions of the FUV and TIR luminosities are shown in Fig. 1. We see that the two distributions are significantly different from each other. The first reason for the difference is that our sample is AKARI 90 μm and IRAS IIFSCz 60 μm selected, i.e., they are inclined to dustier galaxies. Then, the UV continuum of the sample galaxies is extinguished by dust, leading to lower luminosities in the NUV and FUV bands. Another reason is because of the well-known difference in the shape of the luminosity function between UV and IR (Buat and Burgarella, 1998; Takeuchi et al., 2005a), especially at the highest luminosities. The peak luminosities at FUV and TIR are also significantly different. A more detailed analysis on the luminosity functions at UV and IR is presented in Takeuchi et al. (2013).

Fig. 1.

L_TIR and L_FUV distributions of our sample. Dashed and solid histograms represent the distributions of L_FUV and L_TIR, respectively.

3.3 Redshift and luminosity distributions

The redshift distribution of our sample is shown in Fig. 2. We see that most of the sample galaxies are at low redshifts of z ≲ 0.5. The peak of the redshift distribution is located at z ≃ 0.02, almost the same as that of the IIFSCz.

Fig. 2.

Redshift distribution of the sample galaxies. Redshift data are taken from the Imperial IRAS FSC Redshift Catalog (IIFSCz) (Wang and Rowan-Robinson, 2009).

Figure 3 shows the redshift-TIR luminosity distribution. The selection boundary due to the flux detection limit is clearly seen. In statistical terminology this case is referred to as “truncated” meaning that we cannot know if there would be objects below the detection limit.

Fig. 3.

The redshift-total IR (TIR) luminosity distribution.

We show the relation between the redshift and FUV luminosity in Fig. 4. The small squares represent the detection at FUV, while the downward arrows show the upper limits of the luminosities. Due to the fact that our sample is primarily selected at 90 μm (see Fig. 3) some galaxies detected in IR will be invisible at UV bands (see Fig. 4). This case is referred to as “censored”. This difference is important when we try to estimate a luminosity function. We discuss this issue elsewhere (Takeuchi et al., 2013).

Fig. 4.

The redshift-FUV luminosity relations. Open squares represent the FUV luminosities and the downward arrows show galaxies which have FUV flux densities below the detection limit of GALEX AIS. The FUV detection limit is 19.9 mag in AB system (Morrissey et al., 2007).

3.4 Sample completeness

Now we examine the completeness of the final sample. From number counts of the sample, we expect that the flux limit is S₉₀ = 0.45 Jy. The redshift completeness of the sample is tested by V/V_max statistics (Rowan-Robinson, 1968; Schmidt, 1968). Here, V is the volume enclosed in a sphere whose radius is the distance to a certain source in the sample, and V_max is the volume enclosed in a sphere whose radius is the largest distance at which the source can be detected. If the sample is complete, V/V_max of the sample galaxies is expected to be distributed uniformly between 0 and 1, with an average ⟨V/V_max⟩ = 0.5 and a standard deviation (12)^−1/2. For our final sample, we obtain ⟨V/V_max⟩ = 0.51 ± 0.27, i.e., the sample can be regarded as complete down to 0.45 Jy. Thus, even if we have set additional conditions to have redshifts from IIFSCz, our sample is complete above this flux limit. The following analyses are all based on this limit.

3.5 Star formation luminosity

Now we compare FUV luminosity (mainly from massive stars) with TIR luminosity (mainly from dust) for the sample galaxies. Figure 5 shows that our sample galaxies are much more luminous at FIR than at FUV. Downward arrows in Fig. 5 represent galaxies which have UV flux densities below the detection limit of GALEX.

Fig. 5.

Relation between L_FUV and L_TIR. The diagonal dotted line represents the case if L_FUV equals L_TIR. Black dots represent the data from Takeuchi et al. (2010). Downward arrows represent galaxies which have UV flux densities below the detection limit of GALEX.

Here, we define the star formation luminosity, L_sf, as the total luminosity contributed only by massive stars. The SF luminosity is expressed as

((11))

where η is the fraction of IR emission produced by dust heated by old stars, which is not related to the current SF. We adopted 30% for this fraction (Hirashita et al, 2003) in the Local Universe. Buat et al. (2011) have shown that η is almost constant for a wide range of L_tir for galaxies. Their result supports the assumption of a constant η.

Figure 6 shows the contributions of FUV- and SF-related dust luminosities to the SF luminosity. Here, the SF dust luminosity stands for (1 − η)L_tir. The left panel in Fig. 6 shows the contribution of FUV luminosity to the SF luminosity, and the right panel shows the contribution of the SF dust luminosity. It is clearly seen that the galaxies are significantly above the diagonal line, and the scatter is large in the left panel. This comes directly from the fact that L_tir is much larger than L_FUV. In contrast, the right panel of Fig. 6 shows a very tight correlation between (1−η)L_tir and L_sf. We should note that, by definition, (1 − η)L_tir < L_sf. However, as seen in Fig. 6, the galaxies with upper limits exactly delineate the diagonal line. For these galaxies, the SF activity is almost completely hidden by dust. We discuss this hidden SF in more detail in the next section.

Fig. 6.

Contributions of FUV and SF-related dust luminosities to SF luminosity. The left panel shows the contribution of FUV luminosity to the SF luminosity, and the right panel shows the contribution of the dust luminosity. Downward and left-pointing arrows represent galaxies which have UV flux densities below the detection limit of GALEX.

4. Discussion

4.1 Star formation rate

Since we measure the emission from massive stars, we need to convert the number of massive stars to the total number of stars. We use an initial mass function (IMF) for this conversion. The IMF represents the number of newly-formed stars per mass. We assume the Salpeter IMF (Salpeter, 1955).

With the spectral evolutionary synthesis model Star-burst99 (Leitherer et al., 1999), and assuming a constant SFR over 10⁸ yr, solar metallicity, and the Salpeter IMF (mass range 0.1M_⊙−100M_⊙), we obtain the relation between the SFR and L_FUV as (Iglesias-Páramo et al., 2006)

((12))

The relation between SFR and TIR luminosity is

((13))

where we have assumed that all stellar UV light is absorbed by dust. And we obtain the formula under the same assumption for both the star formation history (SFH) and the IMF as those of the FUV. We obtain the following formula to calculate the total SFR:

((14))

where SFR_FUV is the SFR estimated from directly observable UV luminosity, and SFR_dust is that estimated from the dust luminosity.

The fraction of SFR_FUV to the total SFR for the sample galaxies is shown in Fig. 7. The scatter of the fraction is very large at SFR < 20 M_⊙yr⁻¹. However, there is a sudden drop at SFR > 20 M_⊙ yr⁻¹. This means that the fraction of the hidden SF strongly depends on the SFR, and galaxies with higher SFRs are more strongly extinguished by dust.

Fig. 7.

Contribution of the fraction of SFR_fuv to the total SFR. The red and blue symbols are the same as in Fig. 6. The black crosses represent the quasars and active galaxies cross-matched to the objects by Véron-Cetty and Véron (2010). The dotted line represents the total SFR where SFR_FUV/SFR drops suddenly for increasing SFR.

We also find some outliers which have high SFRs and high SFR_FUV fractions. A possible explanation is that they might harbor quasars/AGNs. Since the UV energy of these objects is generated from accretion disks around central black holes, the efficiency of the energy release is much higher than for the usual SF. As such, quasars tend to have a pointlike strong UV source in the center. In this case, L_UV is not related to the SF activity in a simple way so we do not discuss this further in this work. We crossmatched the galaxies to the quasars/AGNs of Véron-Cetty and Véron (2010) and the identified galaxies are represented by the black crosses on the red or blue symbols in Fig. 7. We find that 246 objects in our sample are known quasars/AGNs. The ratio of quasars/AGNs to galaxies with SFR ≤ 20 M_⊙ yr⁻¹ is 5.9%, and the ratio of quasars/AGNs to galaxies with SFR > 20M_⊙yr⁻¹ is 11%. We also find that 6/9 galaxies which have SFR > 20 M_⊙ yr⁻¹ and SFR_FUV/SFR > 0.3 are quasars/AGNs. Thus, we can safely conclude that normal star-forming galaxies (without contamination of UV flux from AGNs) have small SFR_FUV/SFR ratios.

Figure 8 shows the ratio SFR_FUV/SFR_dust as a function of L_sf. The similar ratio using SFR_nuv was presented in Iglesias-Páramo et al. (2006). We can relatively safely compare their result with ours, because SFR_nuv and SFR_FUV are calibrated from the same model, and the total IR luminosity in Iglesias-Páramo et al. (2006) is estimated with a formula proposed by Dale et al. (2001), which was proved to be a close approximation to our L_tir (see Takeuchi et al., 2005b). We observe a larger scatter in Fig. 8 than Iglesias-Páramo et al. (2006). This may be because our sample is much larger and contains more extremely dusty objects.

Fig. 8.

Ratio between SFR_FUV and SFR_TIR as a function of star-formation luminosity L_SF. The red and blue symbols are the same as in Fig. 6.

Figure 9 shows the relation between SFR and L_tir. We find a very tight correlation. Comparing our result with the result of Bothwell et al. (2011), there are more galaxies in the regime with higher L_tir (L_tir ≳ 10¹² L_⊙) and higher SFR (SFR ≳ 10² M_⊙ y^r−1), and few galaxies in the lower L_tir (L_tir ≲ 10⁸ L_⊙) and lower SFR (SFR ≲ 10⁻¹ M_⊙ yr⁻¹) regime. This is because Bothwell et al. (2011) added low-luminosity local volume galaxies to their IR- and UV-selected samples, and they made a volume correction on these data, while our sample is only IR-selected.

Fig. 9.

Relation between SFR and L_TIR. The red and blue symbols are the same as in Fig. 6. The solid line represents the L_TIR-SFR scaling given by Eq. (13). The dotted line represents the L_FIR-SFR scaling given by Iglesias-Páramo et al. (2006).

We also see that galaxies with FUV flux densities below the GALEX detection limit delineate the same trend which was reported by Bothwell et al. (2011). Further, galaxies with SFR < 10⁻¹ M_⊙ yr⁻¹ start to deviate from the general trend, also consistent with Bothwell et al. (2011). This confirms that galaxies with low SFR have larger contributions from SFR_FUV.

4.2 Dust extinction

The relation between L_sf and the infrared excess IRX ≡ L_tIr/L_fuv, known as an indicator of extinction as a function of IR luminosities, is shown in Fig. 10. Figure 10 has the trend that galaxies which have larger L_sf have high IRX, consistent with figure 7 of Buat et al. (2007).

Fig. 10.

Relation between L_sf and the infrared excess IRX ≡ L_tir/L_FUv. Upward and left-pointing arrows represent galaxies which have UV flux densities below the detection limit of GALEX. The scatter of the IRX is large.

Rather naturally, we expected strong dust extinction for the current sample because they are IR-selected. We find a similar result to that of Buat et al. (2005) and Takeuchi et al. (2010).

Figure 11 shows the relation between SFR and IRX. Again, we compare it with figure 10 of Bothwell et al. (2011). Since we plot the sample galaxies directly without a volume correction, it is not straightforward to compare the distribution of galaxies on this figure. Even so, we can see that we have less galaxies in the low SFR (SFR < 10⁻¹ M_⊙ yr⁻¹), and small attenuation (IRX < 1), regime compared with Bothwell et al. (2011). As for Fig. 9, this comes from the effects of sample selection and our sample is inclined to more luminous galaxies because of the IR-selection. The global trend is, however, almost the same for Bothwell et al. (2011).

Fig. 11.

Relation between SFR and IRX. The red and blue symbols are the same as in Fig.10. The dotted line represents the IRX-SFR relation derived for star forming z ~ 0 galaxies by Overzier et al. (2011).

Figure 12 shows the relation between the FUV–NUV color and IRX. The galaxies which have UV flux densities below the detection limit are not plotted. We show the dependence on the two different IR luminosities, L_IRAS60μm and L_tir, for comparison with various previous studies. In general, L_tir is smaller than L_{IRAS60 μm} as can also be seen in Fig. 9. We can compare the left panel of Fig. 12 with figure 15 of Takeuchi et al. (2010) directly, and find a similar trend. By contrast, L_tir is more directly connected to the definition of IR luminous galaxies, LIRGs (10¹¹L_⊙ ≤ L_tir < 10¹²L_⊙) and ultra IR luminous galaxies, ULIRGs (10¹²L_⊙ ≤ L_tir). Indeed, more galaxies in the right panel are identified as the IR luminous galaxies according to the criterion of L_tir dependence. Solid curves in each panel represent the revised IRX-β relation obtained from the GALEX-AKARI measurement of the same UV-luminous starbursts as (Meurer et al., 1999), proposed by Takeuchi et al. (2012). Most IR luminous galaxies are above the curve in each panel. This trend was discovered by Goldader et al. (2002) and followed by subsequent studies (e.g., Buat et al., 2005; Takeuchi et al., 2010, among others). On the other hand, non-IR luminous galaxies, i.e. lower extinction galaxies, follow this curve. This may be explained as follows: Takeuchi et al. (2012) used the same original galaxy sample of Meurer et al. (1999), which was selected for central UV-luminous intense star-bursts. Since these galaxies tend to have lower dust attenuation than LIRGs/ULIRGs, they are similar to the lower-IR luminosity galaxies in our sample having significant UV fluxes, as we have seen in Fig. 9. Hence, we conclude that low-luminosity IR galaxies have a common attenuation strength with UV-luminous starbursts.

Fig. 12.

Relation between FUV–NUV color and IRX. The left panel shows the relation as a function of IRAS 60 μm luminosity, L_{iras 60 μm}. The right panel is the same but with total IR luminosity, L_tir. Dots: galaxies with L_ir < 10¹¹ L_⊙, crosses: galaxies with 10¹¹ L_⊙ ≤ L_ir < 10¹² L_⊙, triangles: galaxies with L_ir ≤ 10¹² L_⊙. Here, L_ir stands for L_{iras 60 μm} in the left panel and L_tir in the right panel. The solid curve in each panel represents the IRX–UV-color relation of UV-luminous starbursts (Takeuchi et al., 2012).

5. Conclusions

We analyzed star formation-related properties of Local galaxies using AKARI and GALEX data. The summary and conclusions of this study are as follows:

• (1) The star formation luminosity, L_sf, is dominated by the emission from dust related to SF activity, (1 − η) L_TIR.

• (2) The contribution of ultraviolet luminosity, L_FUV, has a larger scatter than that of the contribution of (1 − η ) L_tir.

• (3) It is difficult to estimate the star formation activity only from the relation between L_sf and L_FUV because of the small contribution of L_FUV.

• (4) Galaxies with higher SF activity (SFR > 20 M_⊙ yr⁻¹) have a higher fraction of their SF hidden by dust.

• (5) We examined the relation between IRX, L_tir/L_FUV, and FUV–NUV color. Among the current sample, low-IR luminosity galaxies (<10¹¹L_⊙) follow the relation for UV-luminous starbursts proposed by Takeuchi et al. (2012).

These conclusions are consistent with those of Takeuchi et al. (2010). However, we find that the dispersion in various relations they obtained suffered from the noise of the AKARI BSC β-1 catalog. In this study, the S/N is three times better so we do not share the same problem. Thus, we can safely conclude that the above properties are general features of Local star-forming dusty galaxies.