Michael Medford
  • Home
  • Bio
  • Experience
  • Resume
  • Contact

Geospatial Software Engineer
Technical Lead of the Planet Fusion Monitoring Pod
Planet (San Francisco, CA)

Planet has the largest collection of imaging satellites and daily imaging activities of any company in the world. This makes it a natural fit for my expertise in building massive data processing pipelines that solve intractable problems at scale. After my three month internship where I assembled an automated image processing pipelines leveraged Planet's collection of lunar images for camera calibration and validation, I became the Technical Lead for the Planet Fusion Monitoring Pod. My team is responsible for serving our customers the highest quality satellite imagery products by building scalable and efficient computational infrastructure. We harmonize petabytes of imagery from the world's largest earth imaging satellite constellation with publicly available data to give our customers the actionable insights that they need.

Picture

Removing Atmospheric Fringes using Principal Component Analysis
w/ ​Dr. Peter Nugent (Lawrence Berkeley National Laboratory)

Publications of the Astronomical Society of the Pacific, June 2021

Picture
Large scale synoptic surveys have made the generation of optical images efficient to obtain and consequently ubiquitous. Such data flows necessitate that the reduction of raw images into science images must be systematic and require minimal human intervention, all while removing the contamination of sources of noise that can be transient and significant. One such significant source of noise in long wavelength optical imaging are atmospheric emission lines. This work constructs physical models of atmospheric noise using principal component analysis that are used to identify and extract this noise. This pipeline has been assembled into the parallelized open source Python package fringez that is currently executed on all i-band images taken at the Zwicky Transient Facility (ZTF) .

PASP Paper | fringez Documentation

Gravitational Microlensing Event Statistics for the Zwicky Transient Facility
​w/ Dr. Jessica Lu (UC Berkeley) and Dr. Peter Nugent (Lawrence Berkeley National Laboratory)

The Astrophysical Journal, 4 June 2020


128 Microlensing Events from the Three Years of Zwicky Transient Facility Phase One
​w/ Dr. Jessica Lu (UC Berkeley) and Dr. Peter Nugent (Lawrence Berkeley National Laboratory)

Submitted: The Astrophysical Journal, December 2021

Massive stars end their lives in either massive explosions or self-collapse, creating an object whose density is so large that not even light can escape its grasp. These exotic remnants are known as black holes. Models predict that there should be millions of them in the Milky Way Galaxy. Yet astronomers have only been able to discover a few dozen, all of which are dancing with binary partners and emitting strong X-rays.

​Photometric mircolensing is the only method for detecting isolated black holes and may be the key to discovering the lost population of dark objects swimming amongst the stars. I search through the terabytes of data generated by the Zwicky Transient Facility (ZTF) for moments when black holes appear to pass in front of background stars and act as gravitational lenses, warping the spacetime so strongly that the background stars appear to shine brighter for longer.

In my first paper, I generated galaxy simulations to estimate the event rate of microlensing in optical image surveys. In my second paper, I constructed a massive data processing pipeline to search for microlensing event in real Zwicky Transient Facility data. We found 128 microlensing events over three years, providing valuable targets for astrometric follow-up. These papers formed the core of my thesis: Scalable Methods for Detecting Microlensing Black Holes with the Zwicky Transient Facility​.

Astrophysical Journal Abstract | ​Arxiv Abstract
Picture

Calculating Photometric Transformations into the Zwicky Transient Facility Filter System
w/ ​Dr. Jessica Lu (UC Berkeley)

Published: Research Notes fo the American Astronomical Society, 13 March 2020

Picture
​The Zwicky Transient Facility (ZTF) is an optical time-domain survey that surveys the visible Northern sky every few nights in ZTF g-band and r-band. Transformations between the photometric systems of ZTF, Pan-STARRS1 (PS1), and Johnson–Morgan–Cousins (UBV) are essential for extending both the time-baseline and wavelength coverage of ZTF catalogs with information from other catalogs. We use cross-matching catalogs and simple stellar population synthesis models to derive photometric transformations from PS1 and UBV to ZTF. The results of this work, including a set of python functions that executes the UBV to ZTF transformations, can be found at here.

The Hunt for Planet Nine: Detection of trans-Neptunian Objects via Wide-Sky Shift-and-Add Method
w/ ​Dr. Peter Nugent (Lawrence Berkeley National Laboratory)

Trans-Neptunian Objects are asteroids beyond Neptune left over from the initial formation of the solar system. The traditional method to detect these transients is to continuously observe a patch of night sky for a moving piece of light. The proposed "Planet Nine", theorized by Mike Brown and Konstantin Batygin, may be so far away from our sun that it is undetectable by these usual methods. We are developing a novel approach which leverages large-scale computational techniques to possibly discover the planet in existing images. The Palomar Transient Factory (PTF) contains millions of images of the Northern sky, many which fall right on top of the most likely location of the planet. We calculate the trillions upon trillions of possible paths of the planet, and then combine PTF images along these arcs across the sky. By combining many images together, instead of searching for a signal on a single image, we hope to see deeper into the night sky and possible discover Planet Nine.

Read about our search in this article in IEEE Spectrum magazine.
Picture

Exploration of Massive Stellar Core Boundary Definitions in MESA
w/ Dr. Vicky Kalogera (Northwestern University Haven Professor of Physics & Astronomy, Director of CIERA)

We explored the various methods of defining the core boundaries for massive star models with the Modules for Experiments in Stellar Astrophysics (MESA). The currently accepted understanding is that core boundary definitions widely vary within modern star modeling, leading to inconsistent calculations for envelope ejection in binary star formation and it's associated parameters (Tauris 2001). We believe that these results may only apply to low-mass stars and that massive stars may have coherent boundary definitions in certain circumstances. Discovery of coherent core boundary definitions for massive stars may lead to more accurate calculations of binary envelope ejection energies and associated parameters. In order to increase the rate of our work, I have programmed submission and analysis scripts (Python & IDL) to enable parallel MESA simulations on the Northwestern Quest/Grail high performance computing cluster.
Picture
Picture

Near-infrared and Millimeter-wavelength Observations of Mol 160:
A Massive Young Protostellar Core
w/ Drs. Michael Smutko and Grace Wolf-Chase

Published: The Astrophysical Journal, 6 Jan 2012

Picture
The evolution of massive protostars is believed to involve mass accretion followed by outflow jets. In this paper, we measured the collision of outflow jets with molecular hydrogen gas in a potentially very early phase massive protostellar system. I developed data reduction software that automated the process of cleaning, combining and analyzing years of data. This 40-fold increase in efficiency left me the time to investigate previously uncorrected over-subtractions in the NICFIPS imager. I programmed inverse correction frames to cancel these instrumental biases, a new correction technique which led to previously undiscovered shock-excited gas in Mol 160. These shocks provided evidence of outflow jets and became the central focus of our paper.

Astrophysical Journal Abstract