5/7/2020 – Ensemble Data Assimilation and Prediction Development for the UFS
Presenter: Jeffrey S. Whitaker, NOAA PSL and University of CO/CIRES
Abstract: This talk presented recent results from NGGPS-funded work at the Physical Sciences Laboratory focused on improving the use of ensembles in the UFS for data assimilation and prediction. In order to provide an accurate estimate of the background-error covariance used in data assimilation, and to provide reliable probabilistic predictions, the ensemble prediction system must account for both errors in the initial state and the prediction system itself. PSL has been working on improvements to the stochastic parameterization suite for UFS weather application to better capture uncertainty in the land states, and to improve the consistency of the representation of uncertainty in atmospheric physics tendencies and fluxes across model component interfaces (land/atmosphere and ocean/atmosphere). PSL has also been working to improve the Ensemble Kalman Filter (EnKF) data assimilation system used to initialize the NOAA GFS – including updating the system to work with the a much higher model top (80km instead of 50km), improving assimilation of satellite radiances, and reducing the differences between solutions provided by the variational and EnKF algorithms. Recent progress in both of these areas was be presented.
5/21/2020 – Implementation of Global Ensemble Forecast System (GEFSv12) as the First UFS Sub-Seasonal Weather Application
Presenter: Vijay Tallapragada, NOAA/NWS/NCEP/EMC
Abstract: NCEP has implemented the first version of the Finite Volume Cubed Sphere (FV3) dynamical core based Global Forecast System (GFS v15) into operations in June 2019, replacing the spectral model based GFS. This is the first instantiation of NOAA’s Unified Forecast System (UFS) for Medium Range Weather Application in operations. The next major upgrade is for the Global Ensemble Forecast System (GEFSv12) which will use the same FV3 based global model with advanced stochastic physics perturbations, and 2-tiered SSTs, and for the first time, will be extending ensemble based weather predictions for sub-seasonal scales to 35 days. GEFSv12 comes with 20-year reanalysis and 31-year reforecasts to support stakeholder needs for calibration and validation, and 2.5 year retrospective forecasts provide basis for the scientific evaluation of GEFSv12. GEFSv12 will also unify the wave ensembles and aerosol capabilities with this implementation in support of simplifying the NCEP Production Suite. Development of GEFSv12 was a multi-year project involving collaborations with various community partners and stakeholders. This webinar describes the science changes for GEFSv12 and comprehensive evaluation of the ensemble performance for medium and extended range (weeks 3&4) weather predictions along with results from the evaluation of GEFS-Wave ensemble and GEFS-Aerosol components.
6/4/2020 – The Unified Forecast System Short-Range Weather Application for Convection Allowing Model Forecasts
Presenter: Curtis Alexander, NOAA/OAR/GSL
Abstract: Over the past several years, NOAA’s numerical weather prediction (NWP) efforts have organized around a vision of a community-based model system unification, i.e. Unified Forecast System (UFS), across domains from global to regional mesoscale to CONUS-scale convection-allowing (and ultimately cloud-resolving) forecasts. This webinar will focus on describing the development and plans for a UFS Short-Range Weather / Convection Allowing Model (CAM) application including the establishment of a stand-alone regional version of the FV3 dynamic core with extensible grid generation capabilities, an interface with both data assimilation systems and lateral boundary forcings provided from external models and an end-to-end workflow. The webinar will also highlight the plan to consolidate/replace many existing operational CAM systems with a UFS CAM application known as the Rapid Refresh Forecast System (RRFS) including metrics that will be used in the research-to-operations transition process along with physical parameterizations (suite) to be used in the RRFS. Public release(s) of the RRFS components will be described that will help facilitate community engagement in future development efforts. Finally, the webinar will describe some scientific challenges at CAM scales including documented biases in the depiction of convective-scale processes and other CAM initialization challenges.
6/18/2020 – The NCEP Global Ensemble Forecast System version 12: Reanalysis and Reforecast
Presenter: Tom Hamill, NOAA Physical Sciences Laboratory
Abstract: The upcoming implementation of version 12 of the NCEP Global Ensemble Forecast System (GEFSv12) will be accompanied by a 20-year reanalysis and reforecast. The reanalysis uses a reduced-resolution approximation of the operational global data assimilation system to produce reanalyses with statistical characteristics that approximate those of the real-time system. The reanalyses are used primarily for reforecast initialization. The primary reforecast data spans the period 2000-2020. Every day during this period there were 5-member reforecasts computed to +16 days lead time, and once per week an 11-member ensemble was computed to +35 days lead. These reforecasts will be used for the statistical calibration of precipitation/temperature/freezing level for hydrologic forecasts, for 6-10 and 8-14 day forecasts, and in the future for a broader range of postprocessed products including National Blend of Models precipitation inputs, week 2 fire weather forecasts, and weeks 3-4 S2S forecasts including Atlantic hurricane cyclone energy.
To facilitate research across the enterprise, approximately 200 fields from the reanalysis and reforecast data will be made available on NOAA disk servers and at Amazon Web Services via the NOAA Big Data Program.
The seminar will provide more details on the reanalysis/reforecast, review their statistical characteristics, and will briefly discuss the anticipated future GEFSv13 reanalysis/reforecast, which we envision will utilize a weakly coupled data assimilation system and which may be computed using cloud resources.
7/2/2020 – Assessing the Influence of UFS Tropical Forecast Errors on Higher Latitude Predictions Using Nudging Experiments
Presenter: Juliana Dias, NOAA OAR/ESRL/PSL
Abstract: The atmospheric response to variations in tropical latent heating extends well beyond its source region, and therefore it is thought that a reduction of tropical forecast errors should also benefit subsequent forecasts over the extratropics. In this presentation, we employ the use of “relaxation experiments” to quantify the remote influence of tropical forecast errors, which is implemented on the National Centers for Environmental Prediction (NCEP) unified forecast system (UFS). This approach involves nudging forecasts towards reanalyses over a tropical region, while allowing the model to run freely elsewhere. By comparing nudged to free running forecasts, this type of experiment generally shows that midlatitude forecasts are improved in association with reducing tropical forecast errors. For example, Week 2-4 forecast errors over the North Pacific and North America in particular are reduced by tropical nudging. The sensitivity of changes in remote forecast errors to nudging parameters is discussed with focus on the location of the nudging region as well as on which state variables are nudged. In addition, potential modulations of the pattern and amplitude of remote error reductions by ENSO as well as by the Madden Julian oscillation are investigated.
7/16/2020 – The UFS Research-to-Operations (R2O) Project
Presenter: Jim Kinter COLA, George Mason University
Abstract: The Unified Forecast System (UFS) is a community-based, coupled, comprehensive Earth modeling system. It can be configured into multiple applications, which span local to global domains and predictive time scales from sub-hourly analyses to seasonal predictions. The UFS is already being used for a wide range of research and operational prediction applications. In early 2020, the National Weather Service and the Office of Oceanic and Atmospheric Research teamed up to sponsor a project specifically targeting research to address critically important operational prediction issues and bring that research to a pre-implementation level of readiness in anticipation of transition to operations. This presentation will describe this UFS R2O Project, which got underway on 1 July 2020.
8/13/2020 – The Earth System Modeling Framework (ESMF) in the UFS Architecture
Presenter: Rocky Dunlap, NCAR/ESMF
Abstract: The Unified Forecast System (UFS) coupled model architecture is based on the Earth System Modeling Framework (ESMF) and the National Unified Operational Prediction Capability (NUOPC) Interoperability Layer. ESMF and the NUOPC Layer provide a unified, standardized approach to implementing coupled models in the UFS, including a single Earth system driver used across applications, and a flexible approach to supporting different configurations of atmosphere, ocean, sea ice, wave, aerosol, and other model components. This talk will provide a technical overview of how ESMF/NUOPC are used in UFS applications and define key concepts important for UFS users, such as “NUOPC cap”, “Mediator,” and “Connector.” This talk will also provide an overview of capabilities provided by the multi-agency ESMF/NUOPC framework and its role in solving technical challenges in building modular, high-performance Earth system models.
8/27/2020 – Hybridization of Physics-Based Modeling with Machine Learning in Numerical Weather/Climate Prediction Systems
Presenter: Vladimir Krasnopolsky, NOAA/NWS/NCEP/EMC
Abstract: During the last decade, machine learning (ML) began to play an important role in advancing scientific discovery in domains traditionally dominated by physically based (first principle) models. The use of ML models is particularly promising in scientific problems involving processes that are not completely understood, or where it is computationally infeasible to run physically based models at desired resolutions in space and time. Numerical Weather/Climate Prediction Systems (NWPS) represent one of the most complex systems that deal with such problems. Attempts to completely substitute physically based models with even the state-of-the-art black box ML models have often met with limited success in scientific domains due to inability to provide a meaningful physical understanding of underlying processes, their large data requirements, and their limited generalizability to out-of-sample scenarios. Given that neither an ML-only nor a physically based-only approach can be considered sufficient for complex scientific and operational applications, the research community explores the continuum of hybrids of physically based and ML models, where both scientific knowledge and data are integrated in a synergistic manner. This paradigm is fundamentally different from mainstream practices in the ML community that can only work with simpler forms of heuristics and constraints. This presentation is focused on hybrid NWPSs incorporating a deeper coupling of ML methods with physical knowledge. Advantages and limitations of such an approach are discussed.
9/10/2020 – Model Validation Using GOES-16 Brightness Temperatures
Presenter: Sarah Griffin, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin-Madison
Abstract: In this presentation, infrared brightness temperatures (BTs) from the GOES-16 Advanced Baseline Imager are used to examine the accuracy of cloud forecasts using two different model approaches. The first approach will be ensemble-based, comparing simulated BTs from a 5-member ensemble where a stochastic perturbed parameter methodology is applied to the widely-used Thompson-Eidhammer cloud microphysics scheme to a 5-member ensemble with white noise perturbations added to the potential temperature fields at initialization time. The second approach compares simulated BTs from several microphysics and planetary boundary layer (PBL) schemes, as well as land surface models and surface layers.
This presentation will utilize both pixel-based and object-based statistics. Some validation metrics include the mean absolute error and mean bias error, as well as the Object-Based Threat Score and Mean-Error Distance calculation. Objects are identified using the Method for Object-Based Diagnostic Evaluation.
9/24/2020 – Almost Resolving Convection, but not quite…Challenges for Convective Parameterizations
Presenter: Georg A. Grell, NOAA, Global Systems Laboratory
Abstract: Convection Parameterizations (CPs) are components of atmospheric models that aim to represent the statistical effects of a sub-grid scale ensemble of convective clouds. This is done in models in which the spatial resolution is not sufficient to resolve the associated convective circulations. Although CPs have been under development for over 50 years, many challenges remain. These parameterizations often differ fundamentally in closure assumptions and parameters used to solve the interaction problem, leading to a large spread and uncertainty in possible solutions. Additionally, more complexity is being added with almost every new development. On the other hand, increasing resolution in Numerical Weather Prediction models introduced additional challenges, since models can now partially resolve convection. We will discuss basic ideas and constraints of parameterizations, challenges with treating gray-scales (when convection is partially resolved) and new ideas for future developments. Simulations with FV3GFS, GEOS-5, and WRF (with and without the Grell-Freitas convection scheme) will be used to illustrate some of the issues.
10/8/2020 – Climate reanalysis at ECMWF: From research to operational services
Presenter: Dick Dee, Joint Center for Satellite Data Assimilation (JCSDA)
Abstract: This presentation will focus on recent developments in climate reanalysis at ECMWF. We will explain the increasing role of reanalysis activities in operational products and services in Europe in the context of the Copernicus Program. Some results from the latest ECMWF atmosphere reanalysis, ERA5, will be presented. We will also discuss the approach to coupled climate reanalysis being explored for ERA6, to be produced in the next phase of Copernicus program.
10/22/2020 – Development and Evaluation of NCEP’s Global Forecast System GFSv16
Presenter: Fanglin Yang, NOAA/NWS/NCEP/EMC
Abstract: The National Centers for Environmental Prediction (NCEP/NOAA) will upgrade its Global Forecast System (GFS) to version 16 in February 2021. Development of the GFSv16 started one and a half years ago, building upon the implementation of version 15, which featured a new FV3 atmospheric model dynamic core, in June 2019. This talk will present several upgrades included in GFSv16, their impact on downstream applications, and testing and evaluation results from pre-implementation real-time and retrospective parallel experiments. GFSv16 is an implementation of the Unified Forecast System (UFS) featuring an increase in the number of model vertical layers from 64 to 127, whereas the model top is extended from the upper stratosphere to the mesopause (~80 km height). Major upgrades in model physics include: (1) employing a new scheme to parameterize sub-grid scale stationary and non-stationary gravity waves; (2) adopting a scale-aware TKE-EDMF scheme to better represent the PBL processes; and (3) updating the RRTMG radiation package. Major changes in data assimilation include: (1) spinning up an offline land model with observed precipitation to provide improved land initial conditions, (2) using LETKF with model space localization and linearized observation operator to replace the Ensemble Square Root Filter, (3) employing the 4-Dimensional Incremental Analysis Update technique, (4) adopting SKEB perturbation technique in the ensemble forecast component, updating variational quality control, applying Hilbert curve to aircraft data, and inter-channel correlated observation error for CrIS and IASI observations, and (5) assimilating new satellite observations. In addition, GFSv16 includes a one-way coupled wave component that will replace the current operational stand-alone global deterministic wave model.
11/5/2020 – The MOM6 Community Ocean Model and Its Operational Use Across NOAA
Presenter: Robert Hallberg, NOAA/GFDL
Abstract: The Modular Ocean Model, version 6 (MOM6) is the latest in a long line of NOAA-supported community ocean models. With the advent of a new Open Community Development paradigm, MOM6 draws upon a broader range of precursor ocean models, and there is now engagement in the development and deployment of MOM6 by several major groups from academia and agencies in the U.S. and abroad. Extensive testing and modern version control make this Open Community Development possible, while giving each scientific group the control it needs to ensure the quality of its MOM6-based model configurations. This talk will describe some of the notable features of MOM6 (such as the use of Lagrangian Vertical Dynamics to facilitate the use of range of different vertical coordinates and cost-efficient simulations in tracer-rich configurations) that make MOM6 well suited for wide range of research and operational applications. MOM6 being used in a number of different operational or pre-operational applications in NOAA, which this talk will also describe. Examples of such applications range from centennial-scale Earth System projections, to seasonal-to-interannual global coupled forecasts, to high-resolution global near-term forecasts and regional simulations with extensive marine ecosystem components for fishery-related studies.
11/19/2020 – Development of Data Assimilation and Ensemble Forecasting Capabilities for Rapid Refresh Forecast System at CAPS
Presenter: Ming Xue, Center for Analysis and Prediction of Storms (CAPS) and School of Meteorology, University of Oklahoma
Abstract: Under the support of NOAA JTTI, Warn-on-Forecast and GOES-R program fundings, CAPS has been developing capabilities for directly assimilating radar reflectivity, radial velocity, and GOES-R Geostationary Lightning Mapper (GLM) observations directly into the GSI EnKF and hybrid EnVar systems, for HRRR-like CAM forecasting systems including the future FV3-based Rapid Refresh Forecast System (RRFS). Special techniques and treatments have to be devised and implemented in GSI hybrid EnVar to be able to effectively assimilate radar reflectivity data directly within the variational framework due to the high nonlinearity of reflectivity observation operator. The operator also needs to be consistent with the preferred multi-moment microphysics scheme used. For GLM lightning flesh extent density (FED) data, tuned observation operators based on graupel mass and graupel volume are implemented within GSI and experiments show that the assimilation of FED data can achieve similar level of impacts as assimilating radar data. To effectively assimilate observations sampling synoptic (e.g. rawinsonde) through convective (e.g., radar) scales on continental-scale CAM grids utilizing ensemble error covariances, a multi-scale algorithm is developed and tested with GSI EnKF. Assimilation and forecasting results with above schemes with individual cases in an extended period will be presented.
Most recently, the direct radar reflectivity assimilation capabilities in GSI have been tentatively implemented within the first public release of JEDI, and single-time 3DVar and En3DVar analyses of hydrometeors yield reasonable results, on a local FV3-LAM grid or a stretched global FV3 grid. The presentation will also briefly report on results of FV3-LAM forecasts with multiple physics configurations for the HMT realtime forecast experiments.
12/03/2020 – Hurricane Analysis and Forecast System (HAFS): A Unified Forecast System Hurricane Application
Presenter: Avichal Mehra – Chief, Dynamics and Coupled Modeling Group, Modeling and Data Assimilation Branch, NOAA/NWS/NCEP/EMC
Abstract: NOAA/NCEP/EMC has embarked on advancing the next generation operational Hurricane Analysis and Forecast System (HAFS) at NWS as a Unified Forecast System (UFS) application with active participation from other NOAA Laboratories (AOML, GFDL and ESRL), NCAR and operational centers (NHC and AOC). FV3-based HAFS will be a multi-scale model and data assimilation package capable of providing analyses and forecasts of the inner core structure of the Tropical Cyclones (TC) out to 7 days, which is key to improving size and intensity predictions, as well as the large-scale environment that is known to influence the TC’s motion. It will provide an advanced analysis and forecast system for cutting-edge research on modeling, physics, data assimilation, and coupling to earth system components for high-resolution TC predictions to address the outlined objectives of the UFS and the Hurricane Forecast Improvement Plan (HFIP). HAFS will provide Hurricane forecasters with reliable, robust and skillful guidance on TC track and intensity (including RI), storm size, genesis, storm surge, rainfall and tornadoes associated with TCs.
A number of different experiments based on alternate HAFS configurations were run in real-time for the 2020 Hurricane season which included stand-alone-regional domains; nested domains within global models; alternate grid projections and ensembles. In this presentation, performance of these real-time configurations will be compared and contrasted and details discussed along with plans for Hurricane model improvements in the next two to five years at NWS/NCEP.
12/17/2020 – Using AI to Create Situational Intelligence for Storm Responders **
Presenters: Vijay Jayachandran, CEO, ACW Analytics and Peter Watson is the CTO of ACW Analytics
Abstract: Extreme weather events cause billions of dollars of damage every year. Response during storms and restoration of damaged infrastructure afterwards is a perennial problem in many sectors. In the immediate aftermath of an extreme weather event, obtaining accurate ground truth information about “what just happened” is often a time consuming and manual process. This leads to operational delays and inefficiencies, which in turn drive higher costs and poor customer satisfaction.
ACW Analytics is attempting to drastically shorten the post-storm damage assessment window by using AI to confidently estimate the impacts of events in real-time. We have developed an analytical system that applies machine learning algorithms to a blend of NOAA nowcasting, analysis, and observations (HRRR, RTMA, NEXRAD) together with infrastructural and environmental data to produce detailed estimates of the severity and location of storm damages. This information can help infrastructure managers gain the situational intelligence they need to quickly adapt to severe weather as it happens, and ensure a fast and efficient return to normal.
** No Slides or Webinar recording is available