Montage: a grid portal and software toolkit for science-grade astronomical image mosaicking

Montage is a portable software toolkit for constructing custom, science-grade mosaics by composing multiple astronomical images. The mosaics constructed by Montage preserve the astrometry (position) and photometry (intensity) of the sources in the input images. The mosaic to be constructed is specified by the user in terms of a set of parameters, including dataset and wavelength to be used, location and size on the sky, coordinate system and projection, and spatial sampling rate. Many astronomical datasets are massive, and are stored in distributed archives that are, in most cases, remote with respect to the available computational resources. Montage can be run on both single- and multi-processor computers, including clusters and grids. Standard grid tools are used to run Montage in the case where the data or computers used to construct a mosaic are located remotely on the Internet. This paper describes the architecture, algorithms, and usage of Montage as both a software toolkit and as a grid portal. Timing results are provided to show how Montage performance scales with number of processors on a cluster computer. In addition, we compare the performance of two methods of running Montage in parallel on a grid.

💡 Research Summary

The paper presents Montage, a portable, modular software toolkit designed to create custom, science‑grade astronomical image mosaics, and describes its deployment as a grid portal on the NSF‑funded TeraGrid. Montage accepts FITS images with World Coordinate System (WCS) metadata, reprojects them to a common projection, coordinate system, and spatial sampling, normalizes background levels across overlapping images, and co‑adds the results while preserving both astrometric and photometric fidelity.

The toolkit is built from a set of independent ANSI‑C modules: mImgtbl extracts geometry from FITS headers; mProject reprojects images; mAdd co‑adds the reprojected data; mOverlaps, mDiff, mFitplane, and mBgModel together perform background rectification by analyzing overlaps, computing pairwise differences, fitting planar corrections, and solving a global least‑squares problem; and auxiliary executables (mProjExec, mDiffExec, mFitExec, mBgExec) drive these modules sequentially or in parallel. This modularity enables users to invoke individual components, chain them in custom scripts, or run the full pipeline through the provided executables.

Two reprojection strategies are discussed. The original method maps the four corners of each input pixel to sky coordinates, then to output pixel coordinates, and computes the exact spherical polygon overlap using Girard’s theorem. This guarantees energy conservation and sub‑pixel astrometric accuracy but is computationally intensive. A later algorithm replaces the two‑step mapping with a direct linear‑equation transformation that approximates the mapping in a single step, achieving roughly a 30‑fold speed increase while retaining sufficient accuracy for most astronomical projections.

Background correction proceeds in three phases. First, mOverlaps identifies which images intersect on the sky. Second, mDiff creates difference images for each overlapping pair, assuming the images have already been reprojected so that pixels are aligned. Third, mFitplane fits a plane to each difference image (excluding outliers) to derive additive offsets. The mBgModel module then assembles all pairwise offsets into a global system of equations and solves for the optimal additive correction for each image, effectively placing all images on a common photometric baseline. The corrected images are finally passed to mAdd, which co‑adds them using area‑weighted pixel contributions, preserving total flux.

Montage can be executed on a single workstation, on a multi‑processor cluster using MPI, or on a distributed grid. On clusters, the MPI‑enabled executables distribute image‑level tasks across processors, yielding near‑linear scaling up to the number of cores tested. For grid execution, the authors built a portal that translates a user’s mosaic request into a workflow managed by standard grid middleware (Globus, Condor). Each Montage module becomes an independent job; the portal handles data staging, job submission, monitoring, and result retrieval. Two parallel strategies are compared: (1) a “tight‑coupled” MPI approach on a dedicated cluster, and (2) a “loosely‑coupled” grid workflow that tolerates higher latency and heterogeneous resources. Benchmarks on TeraGrid show that while the grid incurs additional overhead due to data transfer and job scheduling, it enables processing of terabyte‑scale datasets that would be impractical on a single cluster, and it provides on‑demand, web‑based access for the broader astronomical community.

The toolkit has been validated on three major sky surveys—2MASS (≈10 TB), DPOSS (≈3 TB), and SDSS DR4 (≈7.4 TB)—demonstrating compatibility with a range of wavelengths, resolutions, and projection systems. By adhering to FITS/WCS standards and the emerging Virtual Observatory (VO) protocols, Montage can be integrated into existing data archives and pipelines, or offered as a VO “science gateway” for end users.

Performance results indicate that on a 16‑core cluster, Montage’s total runtime for a 100‑image mosaic drops from ~30 minutes (single‑core) to ~2 minutes, confirming the efficiency of the parallel implementation. On TeraGrid, a comparable mosaic processed across distributed nodes takes ~3–4 minutes, with the extra time attributable mainly to network latency and remote data staging. The authors conclude that Montage successfully balances scientific rigor (preserving astrometry and photometry), computational efficiency (through modular design and parallel algorithms), and accessibility (via a web portal and VO compliance), making it a valuable tool for current and future large‑scale astronomical imaging projects.