Fall 2023

Group meetings are for students and faculty at CMU and UPitt.
The STAMPS webinars are open to everyone.

These meetings are Fridays 1:30-2:30 PM. Meetings will be in-person in Baker Hall (BH) 232M (second floor of the StatDS Department in Baker Hall). If that plan is to change, we will send out an alert via the mailing list.

September 1 - No Meeting #

September 8 - Welcome meeting #

Welcome newcomers and returning participants!

September 15 - Alex Shen (CMU StatDS) #

September 22 - Josh Speagle (University of Toronto) #

September 29 - LANL Statistical Sciences Group #

October 6 - No Meeting #

October 13, 1:30PM EDT - Public Webinar : Laurence Perault-Levasseur (Université de Montréal / Mila)

October 20 - No Meeting (Fall Break) #

October 27, 1:30PM EDT - Public Webinar : Michael Wehner (Lawrence Berkeley National Laboratory)

November 3 - No Meeting #

November 10, 1:30PM EST - Public Webinar : Michael Kagan (SLAC National Accelerator Laboratory)

November 17 - Maggie Johnson (JPL) #

Title: Tracking plant stress from space: Improving estimates of evapotranspiration through multisensor spatiotemporal data fusion
Description: Evapotranspiration (ET) has been identified as a most important science measurement to better understand the impacts of water loss, for drought prediction, and for agricultural water management. Estimation of ET requires near daily and high spatial resolution (<100m) satellite-derived surface reflectance imagery to characterize highly heterogeneous vegetated surfaces. However, no single, non-commercial space mission currently provides such data due to limitations of the spatial, temporal and spectral resolutions of individual instruments. In this talk, we propose a scalable, spatiotemporal data fusion methodology to combine measurements from multiple remote sensing instruments to produce daily, high spatial resolution surface reflectance products with associated uncertainty estimates. Space-time dynamic linear models are used to leverage spatial and temporal dependence for gap-filling between high resolution images, and a local Kalman filter/smoother is implemented to facilitate processing of billions of measurements on regional to global scales. Finally, we illustrate the impact of the fused products on resulting ET estimates.
Bio: Dr. Maggie Johnson is a statistician at the NASA Jet Propulsion Laboratory (JPL). She has a PhD in statistics from Iowa State University and was a postdoctoral fellow in the SAMSI program on Mathematical and Statistical Methods for Climate and the Earth System before joining JPL in 2018. Her research interests are in spatiotemporal modeling, time series, uncertainty quantification, and environmental statistics. Her current research largely focuses on developing scalable data fusion methodologies and uncertainty quantification for NASA satellite Earth science missions.

November 24 - No Meeting (Thanksgiving) #

December 1 - Lei Hu (McWilliams Center for Cosmology) #

Title: Optimal Image Differencing Algorithm for Detecting Stellar Explosions in the Universe
Description: Image difference analysis is the primary enabling technique in time-domain astronomy. The weather condition, photometric scaling, and sky background generally vary with time and across the field of view for astronomical imaging data taken with optical or near-infrared telescopes. To unveil the stellar explosions as the death of the stars, it is essential to homogenize the observed images using image differencing algorithms for the detection of true brightness change. I will introduce a novel GPU-powered algorithm SFFT, designed to conduct image differencing in Fourier space to attain exceptional computational efficiency and flexibility. The SFFT has been applied in several ongoing time-domain surveys and some transient analyses, including JWST supernova searches and DECam time-domain programs for transient survey and gravitational waves follow-ups. I will also discuss the challenges in developments of statistically optimal methods for future time-domain surveys.

December 8, 1:30PM EST - Public Webinar : Haruko Wainwright (MIT)