RIT on Weather, Chaos, and Data Assimilation Archives for Fall 2024 to Spring 2025
Transition of the NCEP GDAS to JEDI and the path to coupled global data assimilation
When: Mon, September 11, 2023 - 2:00pmWhere: Atlantic Building 3400
Speaker: Cory Martin ( NOAA/NWS/NCEP/EMC) -
Abstract: Currently, the Global Data Assimilation System (GDAS) at the National Centers for Environmental Prediction (NCEP) uses the Gridpoint Statistical Interpolation (GSI) software for computing operational analyses. As part of the transition to the Unified Forecast System (UFS), GSI is slated to be replaced by a Unified Data Assimilation (UDA) framework based on the Joint Effort for Data assimilation Integration (JEDI). Led by the Joint Center for Satellite Data Assimilation (JCSDA), groups from NOAA, NASA, the US Navy and Air Force, and the UK MetOffice all contribute to development and testing of JEDI software. Efforts are underway at NCEP on replacing all GSI-based assimilation components with JEDI-based components for a future implementation of the GDAS. An overview of initial results from a low-resolution prototype JEDI-based atmosphere-only cycled GDAS will be presented. Technical developments and challenges including workflow and observation processing as well as scientific validation of system components are discussed. Concurrent with the transition to new software infrastructure, NCEP is also developing capabilities for coupled Earth system modeling and data assimilation (atmosphere, ocean, sea-ice, land, and composition), with the hope that UFS and JEDI will accelerate scientific innovations on this front. A brief summary of progress and plans for coupled Earth system DA in GDAS is also presented.
A Review of Ensemble Forecasting
When: Mon, September 25, 2023 - 1:30pmWhere: Atlantic Building 4301
Speaker: Zoltan Toth (NOAA Global System Laboratory) - https://gsl.noaa.gov/profiles/zoltan.toth
Abstract: Ensembles are widely used, with demonstrated value. They are believed to sample a small subspace where error preferentially falls; with a much lower error in their mean, to contain genuinely more information about future weather; and with case-dependent variations in their distribution and cloud, to enhance probabilistic forecast performance and capture the dynamical evolution of the real atmosphere. Instead, an analysis of operational, perfect, and statistically generated ensembles reveal a different picture. This talk will provide a review of ensemble forecasting from both theoretical and operational perspectives.
Using Ensemble Modeling and Data Assimilation to Understand Convection-Environment Interactions
When: Wed, October 4, 2023 - 3:00pmWhere: Atlantic Building 3400
Speaker: Derek Posselt (NASA/JPL, Caltech) - https://science.jpl.nasa.gov/people/posselt/
SAMs: Summarizing NWP Forecast Assessment Metrics
When: Mon, October 16, 2023 - 2:00pmWhere: Atlantic Building 3400
Speaker: Ross N. Hoffman (UMD/CISESS & NOAA/NESDIS) - https://scholar.google.com/citations?user=1Hq20YIAAAAJ&hl=en
Data Assimilation Development for the Next Global Forecast System (GFSv17)
When: Mon, October 30, 2023 - 2:00pmWhere: Atlantic Building 3400
Speaker: Catherine Thomas (NOAA/NWS/NCEP/EMC) -
Evaluation of Arctic Clouds in GFS/UFS via Field Campaigns
When: Mon, November 13, 2023 - 2:00pmWhere: https://umd.zoom.us/j/98855947730?pwd=blNQcnpWUVNRSDE2eTNSREIrM3BjZz09
Speaker: Nick Szapiro (UMD and NOAA/NCEP/EMC) - https://www.linkedin.com/in/nicholas-szapiro-b68027228
Abstract: Polar predictions and projections in weather and climate models suffer from misrepresented mass, momentum, and energy processes and limited representations of uncertainties. Supercooled cloud water exemplifies the challenges, with ice-friendly chains of processes leading to commonly gross underestimates and regime errors propagating through coupled systems. Over October 2019 to September 2020, detailed observations made during the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) expedition offer a wealth of local in-situ, integrated, and retrieved observations to benchmark and improve polar processes and predictions.
While scale mismatches complicate comparisons with cloud remote sensing and surface flux observations, systematic underestimates and errors in supercooled cloud liquid are evident in illustrative cases and aggregated populations broadly over the central Arctic and locally over MOSAiC. Sensitivities to initialization and physical parameterizations yield gross, qualitative errors in liquid-bearing cloud occurrence and associated surface longwave radiative fluxes. These regime errors in forecast cloud occurrence and composition imply gross errors in the Arctic surface energy budget for Earth system component coupling. Future cycled prototypes and commonly conditioned single column experiments are expected to further disentangle sensitivities due to conditioning and physical parameterizations and aid development.
Assimilation Design in 10 Steps... for Planetary Atmospheres and Beyond!
When: Mon, December 4, 2023 - 2:00pmWhere: Atlantic Building 3400
Speaker: Steven Greybush (Penn State) - https://www.ems.psu.edu/directory/steven-greybush
Using data assimilation to train a machine-learning forecast model
When: Mon, April 22, 2024 - 2:00pmWhere: ATL 3400
Speaker: Brian Hunt (University of Maryland) - https://www.math.umd.edu/~bhunt/