SANEPIC: A Map-Making Method for Timestream Data From Large Arrays

TL;DR

SANEPIC is a novel map-making method for large array timestream data that efficiently handles correlated noise, demonstrated on submillimeter telescope data, improving map quality with manageable computational resources.

Contribution

It introduces SANEPIC, a maximum likelihood map-making approach explicitly modeling detector correlations, suitable for large datasets with modest computational demands.

Findings

Successfully applied to BLAST data from 2005.

Outperforms simpler filtering-based map-makers.

Two efficient implementations demonstrated.

Abstract

We describe a map-making method which we have developed for the Balloon-borne Large Aperture Submillimeter Telescope (BLAST) experiment, but which should have general application to data from other submillimeter arrays. Our method uses a Maximum Likelihood based approach, with several approximations, which allows images to be constructed using large amounts of data with fairly modest computer memory and processing requirements. This new approach, Signal And Noise Estimation Procedure Including Correlations (SANEPIC), builds upon several previous methods, but focuses specifically on the regime where there is a large number of detectors sampling the same map of the sky, and explicitly allowing for the the possibility of strong correlations between the detector timestreams. We provide real and simulated examples of how well this method performs compared with more simplistic map-makers based on filtering. We discuss two separate implementations of SANEPIC: a brute-force approach, in which the inverse pixel-pixel covariance matrix is computed; and an iterative approach, which is much more efficient for large maps. SANEPIC has been successfully used to produce maps using data from the 2005 BLAST flight.