Deblending: Plans and Challenges

LSST will detect a multitude of galaxies out to 26th magnitude and beyond, producing unprecedentedly dense sky images for a survey of this type. To take complete advantage of this dataset we must be able to robustly infer single-galaxy properties from image regions in which multiple galaxies appear to significantly overlap. This introduces a challenging source of systematic uncertainty that affects object detection, shape measurements, and ultimately inference methods for cosmological parameters, such as dark energy, that depend on a maximally large sample of well-measured galaxies.

In this session we’ll discuss the challenges above along with strategies for deblending and deblended measurements, particularly in the context of dark energy science. We’ll review the current Project plans for detection and Scarlet deblending in the Data Releases, along with community-built tools for blending studies. It would also be good to hear any concrete plans from the wider community for implementing alternative deblending methods and for sharing the resulting data products. Anyone with questions, suggestions, or volunteer contributions touching on any of the above topics is encouraged to contact the session chair.


James Buchanan – Overview
Erfan Nourbakhsh – Effect of unrecognized blends on cosmic shear inference
Fred Moolekamp – Scarlet Lite: Status and plans
Ismael Mendoza – BlendingToolKit, and probabilistic catalogs
General discussion

James J. Buchanan
Suggested Audience: 
community scientists, project staff
Data Management
Applicable to: 
Thursday 08/11
3:30 - 5:00pm
Tortolita A

Upload Slides

Please enter the agenda item in the “Subject” field, choose your file, and click save.


imendoza's picture