SMICA

SMICA, short for Spectral Matching Independent Component Analysis, is a statistical method used in astrophysical data analysis to separate multiple sky components in multi-frequency observations, especially the cosmic microwave background (CMB). It is designed to extract the CMB signal from foreground emissions such as galactic dust, synchrotron radiation, and extragalactic sources.

Method: SMICA operates in harmonic space by analyzing the frequency dependence of sky signals. The observed data across frequency channels are modeled as a linear mixture of independent components with distinct spectral energy distributions, plus instrumental noise and beam effects. SMICA fits a model to the cross-spectral covariances of the data, using a maximum-likelihood or quasi-likelihood approach to estimate the CMB component maps and their angular power spectrum, while also characterizing the foreground components. The technique is semi-blind, exploiting known features of the CMB spectrum (blackbody) and minimal assumptions about foregrounds, enabling separation without requiring precise priors for every foreground.

Applications: SMICA has been applied to data from the Planck mission and is one of the primary component-separated CMB maps produced in Planck's data releases, alongside other methods such as NILC, SEVEM, and Commander. It provides cleaned CMB maps and estimates of the CMB power spectrum, along with models of residual foregrounds.

Strengths and limitations: SMICA effectively leverages spectral diversity across frequency channels but relies on accurate modeling of noise and beam properties; residual foreground contamination can affect the recovered CMB, particularly in polarization or at low multipoles.