Source author record

Ryan Keisler

Ryan Keisler appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

astro-ph.CO Computer Vision Machine Learning physics.ao-ph

Catalog footprint

What is connected

5works

4topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

Forecasting Global Weather with Graph Neural Networks

We present a data-driven approach for forecasting global weather using graph neural networks. The system learns to step forward the current 3D atmospheric state by six hours, and multiple steps are chained together to produce skillful forecasts going out several days into the future. The underlying model is trained on reanalysis data from ERA5 or forecast data from GFS. Test performance on metrics such as Z500 (geopotential height) and T850 (temperature) improves upon previous data-driven approaches and is comparable to operational, full-resolution, physical models from GFS and ECMWF, at least when evaluated on 1-degree scales and when using reanalysis initial conditions. We also show results from connecting this data-driven model to live, operational forecasts from GFS.

preprint2020arXiv

Visual search over billions of aerial and satellite images

We present a system for performing visual search over billions of aerial and satellite images. The purpose of visual search is to find images that are visually similar to a query image. We define visual similarity using 512 abstract visual features generated by a convolutional neural network that has been trained on aerial and satellite imagery. The features are converted to binary values to reduce data and compute requirements. We employ a hash-based search using Bigtable, a scalable database service from Google Cloud. Searching the continental United States at 1-meter pixel resolution, corresponding to approximately 2 billion images, takes approximately 0.1 seconds. This system enables real-time visual search over the surface of the earth, and an interactive demo is available at https://search.descarteslabs.com.

preprint2013arXiv

Prospects for measuring the relative velocities of galaxy clusters in photometric surveys using the kinetic Sunyaev-Zel'dovich Effect

We consider the prospects for measuring the pairwise kinetic Sunyaev-Zel'dovich (kSZ) signal from galaxy clusters discovered in large photometric surveys such as the Dark Energy Survey (DES). We project that the DES cluster sample will, in conjunction with existing mm-wave data from the South Pole Telescope (SPT), yield a detection of the pairwise kSZ signal at the 8-13 sigma level, with sensitivity peaking for clusters separated by ~100 Mpc distances. A next-generation version of SPT would allow for a 18-30 sigma detection and would be limited by variance from the kSZ signal itself and residual thermal Sunyaev-Zel'dovich (tSZ) signal. Throughout our analysis we assume photometric redshift errors, which wash out the signal for clusters separated by <~50 Mpc; a spectroscopic survey of the DES sample would recover this signal and allow for a 26-43 sigma detection, and would again be limited by kSZ/tSZ variance. Assuming a standard model of structure formation, these high-precision measurements of the pairwise kSZ signal will yield detailed information on the gas content of the galaxy clusters. Alternatively, if the gas can be sufficiently characterized by other means (e.g. using tSZ, X-ray, or weak lensing), then the relative velocities of the galaxy clusters can be isolated, thereby providing a precision measurement of gravity on 100 Mpc scales. We briefly consider the utility of these measurements for constraining theories of modified gravity.

preprint2013arXiv

The Statistical Significance of the "Dark Flow"

We revisit the statistical significance of the "dark flow" presented in Kashlinsky et al. (2009). We do not find a statistically significant detection of a bulk flow. Instead we find that CMB correlations between the 8 WMAP channels used in this analysis decrease the inferred significance of the detection to 0.7σ.

preprint2012arXiv

How Massless Neutrinos Affect the Cosmic Microwave Background Damping Tail

We explore the physical origin and robustness of constraints on the energy density in relativistic species prior to and during recombination, often expressed as constraints on an effective number of neutrino species, Neff. Constraints from current data combination of Wilkinson Microwave Anisotropy Probe (WMAP) and South Pole Telescope (SPT) are almost entirely due to the impact of the neutrinos on the expansion rate, and how those changes to the expansion rate alter the ratio of the photon diffusion scale to the sound horizon scale at recombination. We demonstrate that very little of the constraining power comes from the early Integrated Sachs-Wolfe (ISW) effect, and also provide a first determination of the amplitude of the early ISW effect. Varying the fraction of baryonic mass in Helium, Yp, also changes the ratio of damping to sound-horizon scales. We discuss the physical effects that prevent the resulting near-degeneracy between Neff and Yp from being a complete one. Examining light element abundance measurements, we see no significant evidence for evolution of Neff and the baryon-to-photon ratio from the epoch of big bang nucleosynthesis to decoupling. Finally, we consider measurements of the distance-redshift relation at low to intermediate redshifts and their implications for the value of Neff.