The $AKARI$ Far-Infrared All-Sky Survey Maps

We present a far-infrared all-sky atlas from a sensitive all-sky survey using the Japanese $AKARI$ satellite. The survey covers $>99$% of the sky in four photometric bands centred at 65 $\mu$m, 90 $\mu$m, 140 $\mu$m, and 160 $\mu$m with spatial resolutions ranging from 1 to 1.5 arcmin. These data provide crucial information for the investigation and characterisation of the properties of dusty material in the Interstellar Medium (ISM), since significant portion of its energy is emitted between $\sim$50 and 200 $\mu$m. The large-scale distribution of interstellar clouds, their thermal dust temperatures and column densities, can be investigated with the improved spatial resolution compared to earlier all-sky survey observations. In addition to the point source distribution, the large-scale distribution of ISM cirrus emission, and its filamentary structure, are well traced. We have made the first public release of the full-sky data to provide a legacy data set for use by the astronomical community.


Introduction
Infrared continuum emission is ubiquitous across the sky, and is attributed to the thermal emission from interstellar dust particles.These interstellar dust particles are heated by incident stellar radiation field of UV, optical, and near-infrared wavelengths, which together have an energy peak at around 1 µm (interstellar radiation field: ISRF; Mathis et al. 1983).The heated dust radiates its thermal energy at longer wavelengths in the infrared to mm wavelength range.The spectral energy distribution (SED) of the dust continuum emission has been observed from various astronomical objects including the diffuse interstellar medium (ISM), star-formation regions and galaxies.The observed SEDs show their peak at far-infrared (30 -300 µm; FIR) wavelengths, with approximately two thirds of the energy being radiated at λ ≥ 50 µm (Draine 2003; also see Compiègne et al. 2011).It is thus important to measure the total FIR continuum emission energy as it is a good tracer of total stellar radiation energy, which is dominated by the radiation from young OB-stars, and thus a good indicator of the star-formation activity (Kennicutt 1998).
The first all-sky survey of infrared continuum between 12 and 100 µm was pioneered by the IRAS satellite (Neugebauer et al. 1984).It carried out a photometric survey with two FIR photometric bands centred at 60 µm and 100 µm achieving a spatial resolution of ∼ 4 ′ .The IRAS observation was designed to detect point sources, but not designed to make absolute photometry (Beichman et al. 1988), leaving the surface brightness of diffuse emission with spatial scales larger than ∼ 5 ′ to be only relatively measured.It has been possible, however, to create large-area sky maps (the IRAS Sky Survey Atlas: ISSA; Wheelock et al. 1994), by filtering out detector sensitivity drifts.An improvement to the efficacy of the IRAS diffuse emission data was made by combining absolute photometry data with lower spatial resolution from COBE/DIRBE observation (Boggess et al. 1992;Hauser et al. 1998), and by using an improved image destriping technique (Improved Reprocessing of the IRAS Survey: IRIS; Miville-Deschênes & Lagache 2005).An all-sky survey at sub-millimetre wavelengths from 350 µm to 10 mm were made by the Planck satellite with an angular resolution from 5 ′ to 33 ′ (Planck Collaboration et al.  2014a).
The observed FIR dust continuum SEDs are reasonably well fitted by modified black-body spectra having spectral indices β ∼ 1.5 -2, with temperatures ∼15 -20 K (Boulanger et al. 1996;Lagache et al. 1998;Draine 2003;Roy et al. 2010;Planck Collaboration et al. 2011;Planck Collaboration et al. 2014b;Meisner & Finkbeiner 2015).Dust particles with relatively larger sizes (BG: big grains; a ≥ 0.01 µm with a typical size of a ∼ 0.1 µm) are considered to be the main source of this FIR emission (Désert et al. 1990;Mathis 1990;Draine & Li 2007; Compiègne et al. 2011).The BGs are in thermal equilib-Y.Doi et.al. [Vol., rium with the ambient ISRF and thus their temperatures (T BG ) trace the local ISRF intensity (I ISRF ; Compiègne et al. 2010;Bernard et al. 2010).Measuring the SED of the BG emitters both above, and below its peak, is therefore important to determining T BG .For the diffuse ISM, we can assume a uniform T BG along the line of sight and thus can convert the observed T BG to I ISRF .The opacity of the BGs (τ BG ) of the diffuse clouds can be evaluated from the observed FIR intensity and T BG .Assuming that dust and gas are well mixed in interstellar space, τ BG is a good tracer of the total gas column density including all the interstellar components: atomic, molecular and ionized gas (e.g.Boulanger & Perault 1988;Joncas et al. 1992;Boulanger et al. 1996;Bernard et al. 1999;Lagache et al. 2000;Miville-Deschênes et al. 2007;Planck Collaboration et al. 2011).It is also important to use estimates of τ BG as a proxy to infer the foreground to the cosmic microwave background (CMB) emission that is observed in longer wavelengths (Schlegel et al. 1998;Miville-Deschênes & Lagache 2005;Compiègne et al. 2011;Planck Collaboration et al. 2014a;Planck Collaboration et al. 2014b;Meisner & Finkbeiner 2015).
Since the IRAS observations cover only the wavelengths at, or shorter than 100 µm, it is important to note that emission from stochastically heated smaller dust grains becomes significant at shorter wavelengths.Compiègne et al. (2010) estimated the contribution of smaller dust emission to their observations with the PACS instrument (Poglitsch et al. 2010) on-board the Herschel satellite (Pilbratt et al. 2010) based on their adopted dust model (Compiègne et al. 2011).They estimated the contribution to be up to ∼ 50% in the 70 µm band, ∼ 17% in 100 µm band, and up to ∼ 7% in the 160 µm band, respectively and concluded that photometric observations at ≤ 70 µm should be modelled by taking into account the significant contribution of emission from stochastically heated grains.To make accurate T BG determination without suffering from the excess emission from stochastically heated smaller dust grains, the IRAS data thus need to be combined with longer wavelength COBE/DIRBE or Planck data (Schlegel et al. 1998;Planck Collaboration et al. 2014b;Meisner & Finkbeiner 2015) or to be fit with a modelled SED of the dust continuum (Planck Collaboration et al. 2014c).However, COBE/DIRBE data have limited spatial resolution of a 0.7 • -square field-of-view (Hauser et al. 1998).Planck data have a spatial resolution that is comparable with that of IRAS, but the data do not cover the peak of the dust SED.Observations cover the peak wavelength of the BG thermal emission with higher spatial resolutions are therefore required.
Ubiquitous distribution of diffuse infrared emission (cirrus emission) was first discovered by IRAS observations (Low et al. 1984).Satellite and balloon-observations have revealed that the cirrus emission has no characteristic spatial scales, and is well represented by Gaussian random fields with a power-law spectrum ∝ k −3.0 (Gautier et al. 1992;Kiss et al. 2001Kiss et al. , 2003;;Miville-Deschênes et al. 2002, 2007;Jeong et al. 2005;Roy et al. 2010), down to sub-arcminute scales (Miville-Deschênes et al. 2010;Martin et al. 2010).Recently, observations with Herschel with a high-spatial resolution (12 ′′ -18 ′′ at 70, 100, 160 and 250 µm) resolved cirrus spatial structures in nearby star-formation regions, showing that they were dominated by filamentary structures.These filament structures have shown typical widths of ∼ 0.1 pc, which is common for all the filaments of those observed in the nearby Gould Belt clouds (cf.Arzoumanian et al. 2011, also see Arzoumanian et al. 2013 and a comprehensive review by André 2013).Prestellar cores are observed to be concentrated on the filaments (Könyves et al. 2010;André et al. 2010), suggesting that the filamentary structure plays a prominent role in the star-formation process (André et al. 2014).To reveal the mechanism of the very first stages of the star-formation process, whose spatial scale changes from that of giant molecular clouds (≥ 100 pc) to the pre-stellar cores (≤ 0.1 pc), a wider field survey with a high spatial resolution is required so that we can investigate the distribution of the cirrus emission across a broad range of spatial scales, study its nature in more detail, as to remove its contribution as a foreground to the CMB.
For these reasons described above, we have made a new all-sky survey with an infrared astronomical satellite AKARI (Murakami et al. 2007).The sky coverage of this survey was 99%, with 97% of the sky covered with multiple scans.The observed wavelength coverage spans 50 -180 µm continuously with four photometric bands, centred at 65 µm, 90 µm, 140 µm, and 160 µm having spatial resolutions of 1 ′ -1 ′ .5.The detection limit of the four bands reaches 2.5 -16 [MJy sr −1 ] with relative accuracy of < 20%.
In this paper, we describe the details of the observation and the data analysis procedure.We also discuss the suitability of the data to measure the SED and estimate spectral peak of the dust emission (so as to investigate the total FIR emission energy), and to determine the spatial distribution of the interstellar matter and star-formation activity at a high spatial resolution.
In §2, we describe the observation by AKARI with our FIR instrument.The details of the data analysis are given in §3, including subtraction of foreground Zodiacal emission, correction scheme of transient response of FIR detectors, and image destriping.Characteristics of the produced images and their quality are described in §4.In §5, we describe the capability of the AKARI image to make the detailed evaluation of the spatial distribution and the SED of the interstellar dust emission.The caveats that we have about remaining artefacts in the images, and our future plans to mitigate the effects of these are described in §6.Information of the data release is given in §7.
In §8, we summarise the results.

Observation
We performed an all-sky survey observation with the AKARI satellite, a dedicated satellite for infrared astronomical observations (Murakami et al. 2007).Its telescope mirror had a diameter of φ = 685 mm and was capable of observing across the 2 -180 µm near to far-infrared spectral regions, with two focal-plane instruments (FPIs): the InfraRed Camera (IRC: Onaka et al. 2007) and the Far-Infrared Surveyor (FIS: Kawada et al. 2007).Both of the FPIs, as well as the telescope, were cooled down to 6 K with liquid Helium cryogen and mechanical double-stage Stirling coolers as a support (Nakagawa et al. 2007), reducing the instrumental thermal emission.The satellite was launched in February 2006 and the all-sky survey was performed during the period April 2006 -August 2007, which was the cold operational phase of the satellite with liquid Helium cryogen.
The satellite was launched into a sun-synchronous polar orbit with an altitude of 700 km, an inclination angle of 98 • .2, and an orbital period of 98 minutes, so that the satellite revolved along the day-night boundary of the earth.During the survey observations, the pointing direction of the telescope was kept orthogonal to the sun-earth direction and was kept away from the centre of the earth.Consequently, the radiative heat input from the sun to the satellite was kept constant and that from the earth was minimised (see Figure 4 of Murakami et al. 2007).
With this orbital configuration, the AKARI telescope continuously scanned the sky along the great circle with a constant scan speed of 3 ′ .6 sec −1 .Due to the yearly revolution of the earth, the scan direction is shifted ∼ 4 ′ per satellite revolution in a longitudinal direction or in the cross-scan direction.Consequently it was possible to survey the whole sky during each six month period of continuous observation.
Dedicated pointed observations of specific astronomical objects were interspersed with survey observations (Murakami et al. 2007).The survey observation was halted for 30 minutes during a pointed observation to allow for 10 minutes of integration time as well as the satellite's attitude manoeuvre before and after the pointed observation.The regions that had been left un-surveyed due to pointed observations were re-scanned at a later time during the cold operational phase.
The all-sky survey observations at far-IR wavelengths were performed by using the FIS instrument, which was dedicated for photometric scan and spectroscopic observations across the 50 -180 µm spectral region (Kawada et al. 2007).Four photometric bands were used for the all-sky survey.The spectral responses of the bands are shown in Figure 1 and the characteristics of the bands are summarised in Table 1.Two of four bands had a broader wavelength coverage and continuously covered the whole waveband range: the WIDE-S band (50 -110 µm, centred at 90 µm) and the WIDE-L band (110 -180 µm, centred at 140 µm).The other two bands had narrower wavelength coverage and sampled both the shorter and the longer ends of the wavebands: the N60 band (50 -80 µm, centred at 65 µm) and the N160 band (140 -180 µm, centred at 160 µm).
The pixel scales of the detectors were 26 ′′ .8× 26 ′′ .8 for the short wavelength bands (N60 and WIDE-S) and 44 ′′ .2× 44 ′′ .2 for the long wavelength bands (WIDE-L and N160; Table 1).The cross-scan width of the detector arrays were ∼ 8 ′ for N60 and WIDE-S and ∼ 12 ′ for WIDE-L and N160 (see Figure 3 of Kawada et al. 2007).These array widths corresponded to two or three times the shift of the scan direction per satellite revolution (∼ 4 ′ , see above).With a continuous survey observation, as a result, regions close to the ecliptic plane were surveyed at least twice with the N60 and WIDE-S bands and three times with the WIDE-L and N160 bands.Regions at higher ecliptic latitudes had increasingly greater exposure times and numbers of confirmatory scans.
The detector signals were read out by Capacitive Trans-Impedance Amplifiers (CTIAs) that were specially developed for this satellite mission (Nagata et al. 2004).The CTIA is an integrating amplifier so that the detector signal was read out as an integration ramp, with a slope that is proportional to the detector flux.The signal level was sampled at 25.28Hz (N60, WIDE-S) and 16.86Hz (WIDE-L, N160), which corresponded to about three samples in a pixel crossing time of an astronomical source (Kawada et al. 2007).
The integrated electrical charge was reset periodically so that the output voltage remained well within the dynamic range of the CTIA.We applied regular reset intervals of either 2 seconds, 1 second, or after every sampling (correlated double sampling: CDS) for the all-sky survey observations depending upon the sky brightness, with reference to the sky brightness at 140 µm observed by CORBE/DIRBE from Annual Average Map (Hauser et al. 1998).A reset interval of 2sec was normally applied, although it was shortened to 1-sec for the brighter celestial regions with > 60 [MJy sr −1 ].The CDS mode was applied for sky with > 210 [MJy sr −1 ], which mainly corresponds to the inner Galactic plane.
An additional calibration sequence was periodically carried out during the survey observations to calibrate the absolute and relative sensitivities of the various detector channels.The FIS was equipped with a cold shutter to allow absolute sensitivity calibration (Kawada et al. 2007).The shutter was closed for 1 minute after every 150 minutes of observations, or after ∼ 1.5 orbits of the satellite, so as to estimate the dark-sky condition, and to measure the absolute zero level of the detector signal.A constant illumination calibration flash was also applied for 30 seconds while the shutter was closed so that we could calibrate the responsivity of the detector channels in addition to the dark measurement.This calibration sequence was performed when the regions near the north or the south ecliptic pole were surveyed as the regions were repeatedly surveyed with many survey scans.The observed regions that were affected by this calibration sequence were intentionally scattered in the pole regions so that we could maximise the on-sky observing time.
In addition to the absolute calibration sequences mentioned above, a calibration light was flashed every minute during the survey observations to calibrate sensitivity drifts whose time variations were shorter than 150 minutes.The flash simulated the profile of a point source crossing the FOV of each detector pixel.
High-energy particles hitting the detector cause glitches or sudden jumps in the detector signals (Suzuki et al. 2008), which must be restored or removed from analyses to eliminate false detection of astronomical objects.However, the high frequency of high-energy particle hits prevents continuous observations, and may even cause significant drift of the detector responsivities.A region above the earth's surface over the South Atlantic is known to have anomalous geomagnetic field which leads to a higher density of solar protons (the South Atlantic Anomaly: SAA).We characterized this region from in-flight data of particle hit rates as shown in Figure 2. We paused the survey observation during satellite passages through this region.
During the SAA passages, the response of each detector pixel was significantly changed due to high frequency hitting of high-energy particles, which led to excitation of charged particles in the detector material (Suzuki et al. 2008).To restore the detector responsivity, we applied a high electrical voltage (on the order of 1 V) for 30 seconds to each detector pixel to flush out the charged particles (bias-boosting).We performed a bias-boost operation after every SAA passage waiting 1 minutes after exit from the SAA region, re-starting the survey observations two minutes after the bias-boost procedure.
The scattered light of bright objects from the telescope structures at the periphery of the telescope's field of view can also affect the survey data, and emission from the moon was found to be the dominant contributor to this contamination.Thus we measured the scattered light pattern as shown in the Figure 3 and eliminated the contaminated data from the image processing ( §3).The eliminated region is indicated in Figure 3 by the green lines.The scattered light from bright planets (Jupiter and Saturn) also needs to be considered, but we have not currently applied a correction for these signals, and will defer this for a later reprocessing of the image data ( §6.2).
Interruption of the survey observations due to SAA passages, interference from the moon, and pointed observations left gaps in the sky coverage in the first six-month's survey coverage.These gaps were filled in at later times using AKARI's attitude control, which allowed cross-scan offsets < ±1 • .
A summary of the survey coverage and completeness is shown in Figure 4 and Table 2.After about 17 months of the survey period (the cold operational phase), 97 % of the sky was multiply surveyed in the four photometric bands.

Data Analysis
The data obtained during the observations were preprocessed using the AKARI pipeline tool originally optimised for point source extraction (Yamamura et al. 2009).This included corrections for the linearity of the CTIA amplifiers and sensitivity drifts of the detectors referenced against the calibration signals ( §2), rejection of anomalous data due to highenergy particle (glitches), signal saturation, and other instrumental effects as well as dark-current subtraction (Kawada et al. 2007;Yamamura et al. 2009).The orientation of the detector FOV was determined using data taken by a NIR star camera (Yamamura et al. 2009).
Following this initial pre-processing, a separate pipeline was then used to accurately recover the large-scale spatial structures.The main properties of the diffuse mapping pipeline are as follows: 1) transient response correction of the detector signal, 2) subtraction of zodiacal emission, 3) image pro-cessing, 4) image destriping, and 5) recursive deglitching and flat-fielding to produce the final image.Processes 1) and 2) are performed on the time-series data and 3) -5) are performed on image plane data.In the following section, we describe the details of these procedures.

Time-series data analysis with transient response correction
The time-series signal from each detector pixel was processed to recover the true sky flux.Firstly we eliminated anomalous data that were detected during pre-process including glitches and data for which the output of the CTIAs became non-linear because of signal saturation.The data collected during CTIA resets and calibration lamp flashes were also removed (see §2 for the amplifier reset and calibration lamp operation).
The data were then converted to astronomical surface brightness [MJy sr −1 ] from the raw output signal values, multiplied by calibration conversion factors (Takita et al. 2015).Different conversion factors were applied to the data taken in the nominal integration mode, and in the CDS mode ( §2) so that these data are treated separately (Takita et al. 2015).
The transient response, due to non-linear behaviour of the detector, is a major cause of distortion of the detector timeseries signal (Kaneda et al. 2009) and its correction is particularly important for the recovery of high quality diffuse structure (Shirahata 2009).Kaneda et al. (2009) have previously discussed the characterisation of detector slow-response effects and their mitigation, and we take a practical approach here to correct the all-sky survey data so that high quality diffuse maps can be recovered with realistic computational overheads (see Doi et al. 2009 for the details of the computation scheme).
Figure 5 top panel shows an example detector signal that was taken during an in-flight dedicated calibration sequence.
The cold shutter was closed to achieve the absolute zero level ( §2), meanwhile constant illumination with internal calibration lights was performed with two different luminosities (t = 80 − 200 [sec] and t = 320 − 440 [sec]).The output signal should be rectangular wave (the black broken line in the figure) if the detector had an ideal time response, while the observed signal showed significant distortion due to the transient response.Upward steps of the observed signal showed slow time response that took several tens of seconds to reach a stable signal level.On the other hand, downward steps showed relatively quick response with significant overshoot at the falling edges.
Since these signals are step response functions to the incident light, we can evaluate frequency response functions of the detector as the Fourier transformation of the detector signal (Figure 5 middle panel).The response became low at higher frequencies due to slow-response effect of the detector.In addition to that, downward response showed signal amplification at < 1 [Hz], which corresponds to the overshoot of the detector signal.
The transient response of the raw detector signal can be corrected by referring these frequency response functions.The correction function should have a frequency response that is inverse of that of the detector, so the correction function is obtained by deriving reciprocal of the detector frequency response and then make Fourier inverse transformation of the reciprocal data.Since the response function for upward steps and that for downward steps were significantly different, we derived both upward and downward transient correction functions and applied both to the same detector signal (Figure 5 bottom panel).A reasonable correction is achieved by taking the larger signal at each sampling, which is the upper envelope of the two corrected signals.
Because the high-frequency signal component is amplified with the transient response correction, the high-frequency noise is also amplified.To mitigate this side effect, we suppressed the high-frequency part of the numerical filtering to filter out the frequency components having higher frequencies than the signal crossing time of the detector FOV.The residual glitch noises should be eliminated by a deglitching process described below ( §3.3; §3.5).

Subtraction of Zodiacal emission
The zodiacal light emission (ZE) is the thermal emission from the interplanetary dust (IPD) and the dominant diffuse radiation in the mid-infrared (MIR) to FIR wavelength regions.Since the dust around 1 AU from the sun has a thermal equilibrium temperature of ∼ 280 K, the ZE has a spectral peak around 10 -20 µm, and it dominates the MIR brightness in the diffuse sky.Even in the FIR, the ZE contribution is not completely negligible at high galactic latitude.Therefore the ZE component should be segregated from the observed signal to make astronomical sky maps.
Many efforts have been devoted to describe the structure of the zodiacal dust cloud.In particular, number of models have been developed using the infrared satellite data, such as IRAS and COBE/DIRBE.The ZE model most commonly used to date is the one based on the DIRBE data (e.g.Kelsall et  [Vol., reported an inconsistency of 20% for the intensity of the ring component between the AKARI observations and the Kelsall model prediction in MIR.From these points of view, we only subtracted the smooth cloud component of the ZE based on the Gorjian model at this stage.The contribution of the asteroidal dust bands and the MMR component will be described in a separate paper (Ootsubo et al. 2015), which will provide a ZE model including all components for AKARI FIS all sky images.The Gorjian model evaluates the expected ZE brightness in the DIRBE bands.Based on this model evaluation, we estimated the SED of the diffuse ZE component by interpolating the expected ZE brightness in 25, 60, 100, 140, and 240 µm DIRBE bands using a cubic Hermite spline.Then the expected diffuse ZE brightness in each AKARI band was estimated by multiplying this SED by each AKARI band (Figure 1).The earth's orbital position in the solar system at the time of the observation and the viewing direction from the satellite were also considered in the model estimation to take the non-uniform spatial distribution of the IPD cloud in the solar system into account.
The intensities of dust band and MMR components remaining in the released AKARI FIR images are less than about 5 [MJy sr −1 ] for all bands.Detailed values and caveats about the ZE are given in §6.1.

Image processing from time-series signals
The processed time-series data described above were used to produce a preliminary intensity image.To evaluate the intensity of an individual image pixel, we selected data samples near the pixel centre.The selection radius was set at three times the half power beam width (Shirahata 2009;Takita et al. 2015).
Outlier data were removed from the samples.The outliers are mainly due to 1) glitch residuals, 2) noise signal amplified by the transient response correction, and 3) base-line offsets due to long-term sensitivity variations of the detector.The outliers 1) and 2) were eliminated by using a sigma clipping method, and at this level about half of the data samples were eliminated.As for the outliers 3), a string of data was eliminated by referring whose intensity distribution function was distinct from other data.These eliminated data were restored with an additional gain and offset adjustment in a later stage of the data processing ( §3.5).
The residual data were weighted by distance from the pixel centre by assuming a Gaussian beam profile, whose full width at half maximum was 30 ′′ for N60 and WIDE-S, and 50 ′′ for WIDE-L and N160, so that the beam width was comparable to the spatial resolution of the photometric bands and did not therefore degrade the spatial resolutions of resultant images significantly.The weighted mean of the residual data was then taken as the preliminary intensity of the image pixel.The weighted standard deviation, sample number, and spatial scan numbers were also recorded.

Image destriping
The resultant images show residual spatial stripes due to imperfect flat-fielding caused by long-term sensitivity variations of the detector.To eliminate this artificial pattern, we developed a destriping method based on Miville-Deschênes & Lagache ( 2005), who developed their destriping method to ob-tain the Improved Reprocessing of the IRAS Survey (IRIS) map.The details of our procedure will be described in Tanaka et al. (2015).We describe a brief summary in this paper.
The destriping method by Miville-Deschênes & Lagache (2005) can be summarised as follows: 1. Image decomposition: An input image is decomposed into three components; a large-scale emission map, a small-scale emission map, and a point sources map.2. Stripe cleaning: The small-scale emission map is Fourier-transformed to obtain a spatial frequency map.Spatial frequency components that are affected by the stripes are confined in the spatial frequency map along the radial directions that correspond to the stripe directions in the original intensity map.The affected spatial frequency components are replaced with magnitudes that have averaged values along the azimuthal direction (see Figure 1 of Miville-Deschênes & Lagache 2005).

Image restoration: The stripe-cleaned map is inversely
Fourier-transformed and combined with the other decomposed images (a large-scale emission map and a point-sources map).
We find that the stripe cleaning process of the small-scale emission map described above causes strong artefact in some of our images containing bright molecular clouds (e.g., Orion clouds), since such regions contain small-scale and largeamplitude fluctuations of emission.Such large intensity fluctuations contaminate the derived spatial spectrum and degrade the outcomes of subsequent destriping processes.Therefore, we masked the regions that showed large intensity variance in a small-scale map and eliminated those regions from the stripe cleaning process, in addition to a large-scale emission map and a point-sources map.As an indicator of the variance, we calculated the local Root Mean Square (RMS) of the pixel values on a small-scale map and eliminated the regions whose local RMS was greater than a threshold value (2×RMS of entire region).We need to apodise the discontinuity of pixel values at the boundaries of the eliminated regions to suppress spurious spectrum patterns in the estimated spatial frequency spectrum.Thus we multiplied the pixel values in the eliminated regions by (1 + (s − 1) 2 ) −1 , where s is (local RMS)/threshold, instead of filling the regions with zero.
The derived small-scale map was then Fourier-transformed to obtain a two-dimensional spatial frequency map.We applied the stripe cleaning process to the images in Ecliptic coordinates, in which the scanning direction of the AKARI survey is parallel to the y-axis of the images (see §2 for the spatial scan during the all-sky survey observations).This is beneficial because the direction of the spatial stripe patterns caused by the spatial scans become parallel to the y-axis, then noises caused by the stripes are confined along the x-axis of the twodimensional spatial frequency map as shown in Figure 7 (a).The stripe cleaning can be achieved by suppressing the power excess along the x-axis.While Miville-Deschênes & Lagache (2005) made substitution of the pixel values in their noisecontaminated regions with the azimuthal-averaged values, we multiplied the spatial frequency map by an attenuation factor A(k), where k is a wavenumber vector, to suppress the power excess.This attenuation factor is similar to a filter in signal pro-cessing, which is used for converting an input spectrum F (k) to an output spectrum G(k) = A(k)F (k).Since a discontinuous filter function such as box-car function brings ripples on the result of Fourier transform, a smooth function is better for suppressing artefact.Therefore, we calculated the attenuation factor as follows.First we obtained a power spectrum P (r, θ), where (r, θ) is a position on the two-dimensional spatial frequency map in the polar coordinate system.Next, we calculated the azimuthal average of P (r,θ) as Pa (r), where a region around the x-axis was excluded from the averaging.Then, we calculated the local average of P (r, θ) as Pl (r, θ), in the range of r ± √ r/2 [pixels] along the radial direction of the spatial frequency map.Finally, the attenuation factor was calculated as: where R A is a region around the x-axis defined by a particular range in θ and y.The result shows that the spatial frequency is attenuated down to the level of the azimuthal average.We tested our destriping method in a similar way as Miville-Deschênes & Lagache (2005), using simulated images composed of the ISM with a spectral power-law index of γ = −3 as well as random and stripe noises.The result is shown in Figure 8.Our simulation shows that our method does not modify the power spectrum by more than 1% at small scales.

Recursive deglitching and flat-fielding by referring processed image
Stripes were eliminated from the preliminary image by the above mentioned process, which was a consequence of longterm sensitivity drift of the detectors.To restore those data and to perform better deglitching, we reprocessed the time-series data by referring against the preliminary image.
An additional gain and offset adjustment was made for each set of time-series data by each detector pixel to get better correction of the flat-fielding, by fitting each set to the preliminary image.At the same time, outlier data rejection was also made by eliminating the data that exceed 10 times of the standard deviation from the reference image.
The final image was then processed from the time-series data and was cleaned by the image destriping process described above.
We processed the whole sky image in 6 • × 6 • image patches with 5-degree separations.These processed images were then combined into larger images using the Montage software package (Berriman et al. 2008).

Results
Intensity maps from the all-sky survey in the four photometric bands are shown in Figure 9. Detailed photometry with high spatial resolution has been achieved for the whole sky.The far-infrared intensity distribution of the emission shows a clear wavelength dependence, concentrating more to the Galactic plane at shorter wavelengths with significant extension towards higher galactic latitudes being seen at longer wavelengths.The concentration to the Galactic plane indicates a tighter connection to the star-formation activity by tracing high ISRF regions at and around star-formation regions.The extended emission seen at longer wavelengths traces spatial distribution of low temperature dust in high latitude cirrus clouds.This noticeable dependence of the spatial distribution on the wavelength difference between 65 µm and 160 µm shows the importance of the wide wavelength coverage of AKARI to observe the FIR dust emission at the peak of its SED where it has the largest dependence on the temperature difference of the dust particles.This difference in concentration to the Galactic plane is also visible in a zoom-up image of the Galactic plane at l = 280 • -300 • (Figure 10).
The emission ranging from the tenuous dust in high Galactic cirrus clouds to the bright inner galactic plane, has been suc-Y.Doi et.al.
[Vol. ,   for N160.Takita et al. (2015) also estimates the full width at half maximum of point spread function as 63 ′′ , 78 ′′ , and 88 ′′ for N60, WIDE-S, and WIDE-L.Although no evaluation of the point spread function is available for N160 due to the lower sensitivity of the N160 band (Takita et al. 2015), comparable point spread function is expected for N160 as WIDE-L referring to the more sensitive slow-scan data (Shirahata 2009).Improvement of the spatial resolution comparing to IRAS images is demonstrated in Figure 11.The residual of the Zodiacal emission (ZE) along the ecliptic plane is visible in shorter wavelength images, especially in the N60 image.This is because we have subtracted the spa-tially smooth component from the data but have not subtracted the dust band component, which has smaller scale structures that are for the first time revealed by the high spatial resolution AKARI observation ( §3.2).The former models of the ZE cannot reproduce this spatial distribution properly so that we need to develop our own model to remove the emission from the celestial image.Our current estimation of the ZE is briefly described in §6.1 and will be investigated in detail in a future paper (Ootsubo et al. 2015).

Spatial Power Spectra
Our survey images are the first-ever FIR images that cover the whole sky with arc-minute spatial resolution.A major advantage of the AKARI data is that it enables us to obtain the global distribution of the ISM with higher spatial resolutions.In other words, the large spatial dynamic range of the data is one of the key characteristics of the AKARI FIR survey.This characteristic can be examined by calculating the spatial power spectra of the cirrus distribution taken from the AKARI all sky survey, as the cirrus power spectra can be well represented by power-law spectra ( §1).
Miville-Deschênes et al. ( 2010) derived the IRAS/IRIS 100 µm and Herschel/SPIRE 250 µm combined spatial power spectrum at the Polaris flare region and confirmed that the power-law nature of the cirrus power spectrum is kept down to sub-arcminute scales, with a power-law index γ = −2.65±0.10 on scales 0.025 < k < 2 [arcmin −1 ] (also see Martin et al. 2010).Since the Polaris flare is a high Galactic latitude cirrus cloud with virtually no sign of star-formation activity (Ward-Thompson et al. 2010;Martin et al. 2010), the region is well suited to assess the nature of the cirrus power spectrum.
The AKARI FIR image of the Polaris flare region and its spatial power spectra are shown in Figure 12 (also see Doi et al. 2012).Good linearity is found from a small scale of ∼ 3 ′ to the larger scales beyond 10 • , which is consistent with the measurements of many other authors (Miville-Deschênes et al. 2010 and the references therein).The power law index is estimated as γ 140µm = −2.61± 0.01 by fitting the WIDE-L 140 µm spectrum in k = 0.0014 -0.21 [arcmin −1 ] wavenumber range.An excess component from the fitted power-law spectrum is found in the wavenumber range ∼ 0.2 − 0.4 [arcmin −1 ] (the right lower panel of the Figure 12; also see Figure 3 of Doi et al. 2012), which is due to the noise in the data.So we limit the wavenumber range for the fit as < 0.21 [arcmin −1 ] and eliminate all the data above this wavenumber limit.This excess component should be negligible in the power spectra of brighter regions.The deviation from the power-law spectrum at the scales smaller than ∼ 3 ′ can be attributed to the spatial resolution of the WIDE-L band, whose point spread function (PSF) is shown as the dotted line in the Figure 12.
The estimated γ is in good agreement with that by Miville-Deschênes et al. (2010) shown above.The spectra clearly show the wide dynamic range of our observation, as the spatial power component is well retrieved from > 10 • large scale distribution to < 5 ′ small scale structures.
One advantage of the AKARI FIR data compared to the IRAS data is illustrated in Figure 12, which shows deviation from the power-law distribution well above the spatial scales that the AKARI FIR data shows the deviation to be, due to the improved spatial resolution of the AKARI data.The advantage of the improved spatial resolution of the AKARI FIR data is thus clearly indicated.13 to their nearest filamentary structures.Distance to the Taurus region is assumed as 137 pc (Torres et al. 2007).All the 30 YSOs show spatial coincidence with the extracted filaments within the range of our spatial resolution (< 0.05 pc).TTSs show moderate concentration to the filaments with the distance < 0.4 pc.The gray bars show an expected distribution of the uniformly distributed sources in the region.The significance of the difference between the distribution of TTSs and that of uniformly distributed sources is estimated as p-value = 0.004 by the chi-square test.

Filamentary
Recent observations by the Herschel satellite shows that cirrus clouds consists from filamentary structures, whose typical width is ∼ 0.1 pc ( §1).This typical width corresponds to 3 ′ .4-1 ′ .1 at the distance of 100 -300 pc from the sun.Thus these filaments in the local clouds are detectable in the AKARI images, and the global distribution of the filamentary structures can be revealed by the AKARI all-sky survey (Doi et al. 2014).Figure 13 shows an example of the filamentary structure extraction by applying DisPerSE algorithm (Sousbie 2013) to an AKARI image of the Taurus molecular cloud (Doi et al. 2014).Ubiquitous distribution of the filamentary structure is displayed.
We also plot candidate sources of young stellar objects (YSOs; Tóth et al. 2014) and known T-Tauri stars (TTSs; Takita et al. 2010) in Figure 13 to check their spatial correlation with the filamentary structures (also see Figure 14).All the YSO candidates (30 out of 30 sources) in the region show spatial coincidence with the filamentary structure within the range of our spatial resolution.This correspondence is con- sistent with the Herschel observations ( §1), who found > 70% of prestellar cores in the filaments (André et al. 2014), indicating a strong connection of the filamentary structures and star-formation activities.TTSs show less concentration to the filaments, but still significantly higher concentration than that expected for uniformly distributed sources in the region with a p-value of 0.004.Figure 15 shows the correlation between the distance of the TTSs to their nearest neighbour TTSs, which we take as an indicator of the crowdedness of the TTSs' distribution, and the distance of the TTSs to their adjacent filaments.Since we find no correlation in the plot, we conclude that the moderate concentration of the TTSs found in Figures 13 & 14 is not because that TTSs tend to distribute in the dense cirrus regions, which have crowded filamentary structures.This moderate concentration of the TTSs can be explained as an age evolution of their distribution.A proper motion of 0.1 [km s −1 ] of the TTSs against their natal filaments for 1 -5 × 10 6 years gives distance of 0.1 -0.5 [pc], which is consistent with our observed projected distance of the TTSs from their adjacent filaments.
The large spatial dynamic range of the AKARI observation enables us to reveal global distribution of the filamentary structures, and to study their correlation with young stellar sources of various evolutionary stages.

Spectral Energy Distribution
The SED of the cirrus emission can be estimated from the square root of the amplitude of the spatial power spectra (Roy et al. 2010).We estimate the wavelength dependence of the amplitude in the Polaris Flare region, whose area is shown in Figure 12.The power law index of the spectrum scales is assumed as γ = −2.61for all the four AKARI FIR bands as well as the ancillary data of IRIS and Planck.The estimated relative amplitudes and their fitting errors are shown in Figure 16.
In addition to the spatial dynamic range of the AKARI data, another key aspect is its ability to sample across the spectral peak of the dust SED with four photometric wavebands.To demonstrate this aspect, we performed a model fit to the AKARI FIR data with the Compiègne et al. (2011) dust model, using the DustEM numerical tool (http://www.ias.upsud.fr/DUSTEM/),as indicated in Figure 16.It is clear that we can accurately reproduce the FIR dust SED with the AKARI FIR data, as the fitted model SED agrees well with all the ancillary data.
The colour temperature from the WIDE-S / WIDE-L intensity ratio (90 µm / 140 µm) is estimated to be T = 16.8 [K].A modified black body spectrum of the estimated temperature is shown in Figure 16 as the red dashed line.A modified black-Fig.13.Filamentary structures extracted from the AKARI all-sky survey image of the Taurus region (Doi et al. 2014).The skeleton of the filamentary structures, which is extracted from the AKARI image by applying DisPerSE algorithm (Sousbie 2013), is superposed on the AKARI WIDE-L all-sky survey image.The red circles are 30 candidate sources of young stellar objects identified in the AKARI bright source catalogue (Tóth et al. 2014).The blue triangles are 303 known T-Tauri stars in the region (Takita et al. 2010), a compilation of published catalogues by Strom et al. (1989), Beckwith et al. (1990), Wichmann et al. (1996), Magazzu et al. (1997), Li & Hu (1998), Güdel et al. (2007), Kenyon, Gómez, Whitney (2008), andRebull et al. (2010).The AKARI FIR data is therefore a good tracer of the temperature, and the total amount of BGs, and as a result of the ISM as a whole.Together with the much improved spatial resolution from the former all-sky survey data in the FIR wavebands, the newly achieved AKARI FIR high-spatial resolution images of the whole sky should be a powerful tool to investigate the detailed spatial structure of ISM and its physical environment.

Caveats and plans for future improvements
Although we have produced a sensitive all sky FIR images, the data still contains some artefacts, and there is scope to make further improvements.In the following, we describe remaining caveats about the efficacy of the data and our plans to mitigate these remaining problems.

Zodiacal emission
Although we have subtracted the smooth cloud component of the ZE from the raw data, we have not yet subtracted the asteroidal dust band component, nor the MMR component from the data, with the consequence that the contributions from these remain in the images, and is recognisable at the shorter wavelength bands (N60 and WIDE-S).This is because small scale structures of the ZE are not adequately reproduced by the former Zodiacal emission models ( §3.2).Asteroidal dust bands   (Compiègne et al. 2011).The black dotted lines are the decomposition of the modelled SED in BGs, SGs, and PAHs.The relative fraction of the excess emission from SGs against that from BGs estimated from the fitted model SED are 70.0%for N60, 16.4% for WIDE-S, 2.6% for WIDE-L, and 2.0% for N160.The red dashed line indicates the estimation of colour temperature (T = 16.8 [K]) from the intensity ratio between WIDE-S and WIDE-L data assuming a modified black-body spectral index β = 1.62.
appear as pairs of parallel bands equally spaced above and below the ecliptic plane.The continuous and smooth distribution of the dust bands and the MMR component along the ecliptic plane can be seen in the AKARI N60 and WIDE-S maps (Figure 9).
The intensity and the ecliptic latitudes of the peak positions of these components change depending on the ecliptic longitude.We briefly evaluate the spatial distribution and the latitudinal profiles of these components at the shorter wavelength bands, N60 and WIDE-S.The contribution of the asteroidal dust bands and the MMR component will be discussed in more detail in Ootsubo et al. (2015).Figure 17 shows example latitudinal profiles of the residual ZE contribution from the dust bands and the MMR component observed in the WIDE-S band in two regions, where the ZE contributions are strong (175 • < λ < 185 • and −5 • < λ < 5 • ).The estimated intensities of the emission that are left in the AKARI FIR images become maximum near the ecliptic plane and are < 5 [MJy sr −1 ] for N60 and < 4 [MJy sr −1 ] for WIDE-S, respectively.Although the ZE contribution cannot be clearly seen in the WIDE-L and N160 images, the estimated intensities are < 1 [MJy sr −1 ] for WIDE-L and N160 at most, if we assume that these components have the dust temperature T = 200 -300 K.

Moving bodies
Planets and asteroids have not been masked during our image processing as the detection of the faint sources are not fully investigated and the scattering pattern of the bright sources are yet to be analysed.So the images near the ecliptic plane may contain these solar system bodies and will need further consideration in the future.Meanwhile, we list positions of the planets and 55 major asteroids that are scanned during the AKARI FIR all-sky survey observation in Table 3.

Earth shine
During the survey observation, the pointing direction of the telescope was kept orthogonal to the sun-earth direction and away from the centre of the earth, so that we can minimise the heat input from the earth to the telescope ( §2).However, since the inclination angle of the satellite's orbit was not equal to 90 • , the satellite direction was not precisely in the opposite direction to the centre of the earth and had a small time variation.This resulted in a small variation in the viewing angle of the earth limb from the telescope.The viewing angle became minimum in the observation around the north ecliptic pole in the summer solstice period, caused the illumination of the earth's thermal radiation to the top of the telescope struc- ture.This thermal radiation was detected as the excess emission in the all-sky survey image, and is most conspicuous in the WIDE-S band images with a maximum intensity of ∼ 1.5 [MJy sr −1 ].The contaminated regions are spread like a fan around the north ecliptic pole at around λ: −30 • -+20 • , β ≥ 71 • and λ: +150 • -+210 • , β ≥ 54 • .The cross-section of the spatial profile of the excess emission is shown in Figure 18.Since the spatial profile is not well determined as the scattering path of the earth shine in the telescope is not fully studied yet, we leave the emission in the production image and ask an attention of the users.Assuming 300 [K] black-body spectrum for the earth shine, the excess emission of 1.5 [MJy sr −1 ] in the WIDE-S band corresponds to 2.2, 0.13, and 0.05 [MJy sr −1 ] in N60, WIDE-L, and N160 bands, respectively.

Data Release
We have made the first data release of our all-sky survey data to the public in December 2014.The data are released as 6 • × 6 • FITS format intensity image tiles that cover the whole sky, with supporting data showing the standard deviation of intensity, data sample numbers, and number of spatial scans in the same 2-dimensional FITS files.All the data can be retrieved from the following web site: http://www.ir.isas.jaxa.jp/AKARI/Observation/

Conclusion
We provide full sky images of the AKARI FIR survey at 65 µm, 90 µm, 140 µm and 160 µm.Together with the > 99%  coverage of the whole sky, the high spatial resolution from 1 ′ to 1.5 ′ of the AKARI FIR survey reveals the large-scale distribution of ISM with the great detail.Comprehensive wavelength coverage from 50 µm to 180 µm with four photometric bands provides SED information at the peak of the dust continuum emission, enabling us to make precise evaluation of its temperature, which leads to a detailed investigation of the total amount of dust particles and its irradiation environment.The AKARI FIR images are a new powerful resource from which to investigate the detailed nature of ISM from small scales to the full sky.

Fig. 1 .
Fig. 1.Spectral responses of the four AKARI FIR bands centred at 65 µm (N60), 90 µm (WIDE-S), 140 µm (WIDE-L), and 160 µm (N160).Note that AKARI has continuous wavebands covering 50 -180 µm so that we can make precise evaluation of the total FIR intensity from the in-band flux of the AKARI observations.Spectral responses of IRAS and COBE/DIRBE are also shown for comparison.

Fig. 3 .
Fig. 3. Scattered light around the moon observed in the Wide-L band.The moon position is at the centre of the figure.We closed the shutter near the moon < 5 • and thus no data are available at the region around the moon.The broad diffuse horizontal feature in the middle of the figure is the Zodiacal dust band component, which has not been removed from the data (see §3.2).The green lines indicate the regions which are contaminated by the scattered light.The data observed in these regions were eliminated from the image processing.

Fig. 2 .
Fig. 2. The spatial distribution of the high-energy particle hit rate evaluated from in-flight data of the FIS detector signal.The dashed line indicates the region where we stopped the FIS observation during the satellite passages.

Fig. 4 .
Fig. 4. Sky coverage of the AKARI all-sky survey.Spatial scan numbers are displayed in different celestial coordinates.

Fig. 5 .
Fig. 5. Top panel -a step function of the detector signal taken in-flight by an internal calibration lamp illumination.The cyan line shows the detected signal and the black broken line shows a schematic of the lamp illumination pattern.Middle panel -spectral response functions of a detector pixel.The blue line shows a spectral response that is estimated from the upward step function and the red line shows a spectral response that is estimated from the downward step function.Correction functions for the transient response of the detector are evaluated from these spectral responses (see text).Bottom panel -signal correction by the correction functions evaluated from the above spectral responses.The cyan line is the raw detector signal, which is the same signal shown in the top panel.The blue line is a corrected signal by applying the upward correction function and the red line is a corrected signal by applying the downward correction function.A reasonable signal profile is reproduced by taking the upper envelope of the two corrected signals.

Fig. 6 .
Fig. 6.An example of the transient response correction that is applied to M33 WIDE-L images.The top panel shows the image without the correction for the comparison to the image with the correction in the bottom panel.An image destriping process ( §3.4) has been applied for both images.fied the parameters of the Wright model.Although the most part of the ZE structure in AKARI FIR images can be well reproduced with the Kelsall or the Wright/Gorjian models, there are discrepancies at small-scale structures.In particular, the intensity and the ecliptic latitudes of the peak positions of the asteroidal dust bands cannot be precisely reproduced with these models.Pyo et al. (2010) also Figure 7 (b) shows an attenuation factor map obtained for a two-dimensional spatial frequency map shown in Figure 7 (a).Since Pl (r, θ) is the local average of P (r, θ), A(r, θ) has smooth distribution in the radial direction.
Figure 7 (b) also shows that A(r,θ) is close to unity if |y| > 0.05 [arcmin −1 ], which means that our destripe process is not sensitive to the definition of R A .The resultant destriped spatial frequency map is shown in Figure 7 (c), which is obtained by multiplying the spatial frequency map (a) by the attenuation factor (b).

Fig. 7 .
Fig. 7. Panel (a) -A spatial frequency map before destripe, obtained with Fourier transformation of an original intensity map containing stripe noises.The real part of the complex Fourier moduli is plotted for panels (a) and (c).Panel (b) -A map of an attenuation factor, A(r, θ).See text for the definition of A(r,θ).Panel (c) -A destriped spatial frequency map, obtained by multiplying (a) by (b).We show details of the maps around the x-axis (x ≥ 0 [arcmin −1 ], |y| ≤ 0.11 [arcmin −1 ]), as the excess component caused by the stripe noises is confined along the x-axis (see text).The colour scales of the images are from -0.03 to 0.03 in linear scale for (a) and (c) and from 0.2 to 1.0 in logarithmic scale for (b).Colour bars are shown at the bottom of the figure.

Fig. 8 .
Fig. 8.A verification of the image restoration with our destriping method.Panel (a) -A synthetic image of ISM with random noise.Panel (b) -An input image for destriping made by adding stripe noises to (a).Panel (c) -A destriped image.Panel (d) -Residual error of (c) minus (a).The colour scales are same for the four images.Bottom panel -Relative error between the spatial spectra of (c) and (a) derived by [(c) − (a)]/(a).

Fig. 11 .
Fig. 11.Zoom-up images of the η Carinae region.Upper panel -a three-colour composite of N60 (blue), WIDE-S (green), and WIDE-L (red) images.Lower panel -a composite image of IRIS 60 µm (cyan) and IRIS 100 µm (red) images for a comparison of spatial resolutions between IRAS and AKARI images.

Fig. 14 .
Fig. 14.A histogram of the distance from young stellar objects (YSOs; red bars) and T-Tauri stars (TTSs; blue bars) in the region shown in Figure13to their nearest filamentary structures.Distance to the Taurus region is assumed as 137 pc(Torres et al. 2007).All the 30 YSOs show spatial coincidence with the extracted filaments within the range of our spatial resolution (< 0.05 pc).TTSs show moderate concentration to the filaments with the distance < 0.4 pc.The gray bars show an expected distribution of the uniformly distributed sources in the region.The significance of the difference between the distribution of TTSs and that of uniformly distributed sources is estimated as p-value = 0.004 by the chi-square test.

Fig. 12 .
Fig. 12. Left panel -WIDE-L intensity image of the Polaris Flare region in Galactic coordinates.Right upper panel -Spatial power spectrum of WIDE-L intensity (the blue solid line) observed in the Polaris flare region shown in the left figure.A power-law fitting of the spectrum at the scales of 0.0014 < k < 0.21 [arcmin −1 ] is shown as the blue dashed line with the power index of −2.61 ± 0.01.The blue dotted line shows the spectrum of the point spread function, which is arbitrary shifted in the vertical direction for the comparison with the WIDE-L power spectrum.The red solid line shows the spatial spectrum of the IRIS 100 µm data of the same region for comparison.Right lower panel -Relative error of WIDE-L spatial spectrum from the power-law fitting shown in the right upper panel (the blue solid line).Standard error of the WIDE-L spatial spectrum estimation is denoted as the blue dotted line.Relative error of the IRIS 100 µm data is also shown as the red solid line for comparison.
body spectral index is assumed as β = 1.62, which is a mean value over the whole sky estimated by Planck Collaboration et al. (2014b) by fitting IRIS 100 µm and Planck 353, 545, and 857 GHz data with modified black body spectra.The estimated colour temperature spectrum is consistent with the FIR -submm part of the BGs' emission spectrum as well as the Planck observations.

Fig. 16 .
Fig.16.Spectrum Energy Distribution at the Polaris Flare region shown in Figure12by fitting the spatial power spectra of each observational band by assuming the power low index of -2.61, which is estimated for the WIDE-L data.The square root of the relative amplitude of the spatial power spectra and their fitting errors are shown for the four AKARI FIR data as well as the ancillary IRIS and Planck data.The blue solid line indicates a model fitting of the four AKARI FIR data by applying DustEM SED model(Compiègne et al. 2011).The black dotted lines are the decomposition of the modelled SED in BGs, SGs, and PAHs.The relative fraction of the excess emission from SGs against that from BGs estimated from the fitted model SED are 70.0%for N60, 16.4% for WIDE-S, 2.6% for WIDE-L, and 2.0% for N160.The red dashed line indicates the estimation of colour temperature (T = 16.8 [K]) from the intensity ratio between WIDE-S and WIDE-L data assuming a modified black-body spectral index β = 1.62.

Fig. 15 .
Fig. 15.Correlation between the distance of the TTSs to their nearest neighbour TTSs and the distance of the TTSs to their adjacent filaments.Arrows are the sources who show positional correspondence to filaments within our spatial resolution.The nearest neighbour distance is an indicator of the crowdedness of the TTSs' distribution.We find no dependence of the TTSs' distance to their adjacent filaments whether they are in the crowded regions or in an open field.

Fig. 17 .
Fig. 17.Example latitudinal profiles of the dust bands and the MMR component observed in the WIDE-S band in two regions, 175 • < λ < 185 • (the solid red line) and −5 • < λ < 5 • (the solid blue line).The shaded areas denote the standard error of the profile.The intensity of the residual ZE contribution in the AKARI map are less than 5 [MJy sr −1 ] at the shorter wavelength bands.

Fig. 18 .
Fig. 18.Top panel -WIDE-S image of the north ecliptic pole.A fanlike distribution of the excess emission due to the earth shine is recognisable.Middle and bottom panels -cross-sections of the excess emission in WIDE-S image.The shaded areas denote the standard deviation of the profile.

Table 2 .
Scan coverage of the survey observations.

Table 3 .
List of planets' and 55 major asteroids' positions that are scanned during the AKARI FIR all-sky survey observation.