Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userFar-IR Detection Limits II: Probing Confusion including Source Confusion
84
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
We present a comprehensive analysis for the determination of the confusion levels for the current and the next generation of far-infrared surveys assuming three different cosmological evolutionary scenarios. We include an extensive model for diffuse emission from infrared cirrus in order to derive absolute sensitivity levels taking into account the source confusion noise due to point sources, the sky confusion noise due to the diffuse emission, and instrumental noise. We use our derived sensitivities to suggest best survey strategies for the current and the future far-infrared space missions Spitzer, AKARI (ASTRO-F), Herschel, and SPICA. We discuss whether the theoretical estimates are realistic and the competing necessities of reliability and completeness. We find the best estimator for the representation of the source confusion and produce predictions for the source confusion using far-infrared source count models. From these confusion limits considering both source and sky confusions, we obtain the optimal, confusion limited redshift distribution for each mission. Finally, we predict the Cosmic Far-Infrared Background (CFIRB) which includes information about the number and distribution of the contributing sources.
rate research
Read More
Fluctuations in the brightness of the background radiation can lead to confusion with real point sources. Such background emission confusion will be important for infrared observations with relatively large beam sizes since the amount of fluctuation tends to increase with angular scale. In order to quantitively assess the effect of the background emission on the detection of point sources for current and future far-infrared observations by space-borne missions such as Spitzer, ASTRO-F, Herschel and SPICA, we have extended the Galactic emission map to higher angular resolution than the currently available data. Using this high resolution map, we estimate the sky confusion noise due to the emission from interstellar dust clouds or cirrus, based on fluctuation analysis and detailed photometry over realistically simulated images. We find that the confusion noise derived by simple fluctuation analysis agrees well with the result from realistic simulations. Although the sky confusion noise becomes dominant in long wavelength bands (> 100 um) with 60 - 90cm aperture missions, it is expected to be two order of magnitude smaller for the next generation space missions with larger aperture sizes such as Herschel and SPICA.
We present detailed predictions for the confusion noise due to extragalactic sources in the far-IR/(sub)-millimeter channels of ESA/ISO, NASA/Spitzer, ESA/Herschel and ESA/Planck satellites, including the contribution from clustering of unresolved SCUBA galaxies. Clustering is found to increase the confusion noise, compared to the case of purely Poisson fluctuations, by 10-15% for the lowest frequency (i.e. lowest angular resolution) Spitzer and Herschel channels, by 25-35% for the 175 micron ISOPHOT channel, and to dominate in the case of Planck/HFI channels at nu>143GHz. Although our calculations make use of a specific evolutionary model (Granato et al. 2004), the results are strongly constrained by the observed counts and by data on the redshift distribution of SCUBA sources, and therefore are not expected to be heavily model dependent. The main uncertainty arises from the poor observational definition of the source clustering properties. Two models have been used for the latter: a power-law with constant slope and a redshift-independent comoving correlation length,r_0, and the standard theoretical model for clustering evolution in a LambdaCDM universe, with a redshift-dependent bias factor. In both cases, the clustering amplitude has been normalized to yield a unit angular correlation function at theta_0=1-2 arcsec for 850 micron sources fainter than 2 mJy, consistent with the results by Peacock et al. (2000). This normalization yields, for the first model, r_0=8.3$ Mpc/h, and, for the second model, an effective mass of dark matter haloes in which these sources reside of M_halo=1.8*10^{13} M_sun/h. These results are consistent with independent estimates for SCUBA galaxies and for other, likely related, sources.
The Laser Interferometer Space Antenna (LISA) will detect thousands of gravitational wave sources. Many of these sources will be overlapping in the sense that their signals will have a non-zero cross-correlation. Such overlaps lead to source confusion, which adversely affects how well we can extract information about the individual sources. Here we study how source confusion impacts parameter estimation for galactic compact binaries, with emphasis on the effects of the number of overlaping sources, the time of observation, the gravitational wave frequencies of the sources, and the degree of the signal correlations. Our main findings are that the parameter resolution decays exponentially with the number of overlapping sources, and super-exponentially with the degree of cross-correlation. We also find that an extended mission lifetime is key to disentangling the source confusion as the parameter resolution for overlapping sources improves much faster than the usual square root of the observation time.
Gravitational microlensing surveys target very dense stellar fields in the local group. As a consequence the microlensed source stars are often blended with nearby unresolved stars. The presence of `blending is a cause of major uncertainty when determining the lensing properties of events towards the Galactic centre. After demonstrating empirical cases of blending we utilize Monte Carlo simulations to probe the effects of blending. We generate artificial microlensing events using an HST luminosity function convolved to typical ground-based seeing, adopting a range of values for the stellar density and seeing. We find that a significant fraction of bright events are blended, contrary to the oft-quoted assumption that bright events should be free from blending. We probe the effect that this erroneous assumption has on both the observed event timescale distribution and the optical depth, using realistic detection criteria relevent to the different surveys. Importantly, under this assumption the latter quantity appears to be reasonably unaffected across our adopted values for seeing and density. The timescale distribution is however biased towards smaller values, even for the least dense fields. The dominant source of blending is from lensing of faint source stars, rather than lensing of bright source stars blended with nearby fainter stars. We also explore other issues, such as the centroid motion of blended events and the phenomena of `negative blending. Furthermore, we breifly note that blending can affect the determination of the centre of the red clump giant region from an observed luminosity function. This has implications for a variety of studies, e.g. mapping extinction towards the bulge and attempts to constrain the parameters of the Galactic bar through red clump giant number counts. (Abridged)
Few-shot object detection is a challenging but realistic scenario, where only a few annotated training images are available for training detectors. A popular approach to handle this problem is transfer learning, i.e., fine-tuning a detector pretrained on a source-domain benchmark. However, such transferred detector often fails to recognize new objects in the target domain, due to low data diversity of training samples. To tackle this problem, we propose a novel Context-Transformer within a concise deep transfer framework. Specifically, Context-Transformer can effectively leverage source-domain object knowledge as guidance, and automatically exploit contexts from only a few training images in the target domain. Subsequently, it can adaptively integrate these relational clues to enhance the discriminative power of detector, in order to reduce object confusion in few-shot scenarios. Moreover, Context-Transformer is flexibly embedded in the popular SSD-style detectors, which makes it a plug-and-play module for end-to-end few-shot learning. Finally, we evaluate Context-Transformer on the challenging settings of few-shot detection and incremental few-shot detection. The experimental results show that, our framework outperforms the recent state-of-the-art approaches.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
