FY22 Competitions - OSTI Modeling
(Click on each description for more information)
Title | PIs and Co-PIs | Affliation |
---|---|---|
Tropical Cyclone Wind Gust Guidance for NWS Operations | PI: Mark DeMaria | CIRA/Colorado State University |
Jun Zhang, Andrew Hazelton | University of Miami | |
Lixin Lu | CIRA/Colorado State University |
Abstract
Tropical Cyclone Wind Gust Guidance for NWS Operations
Authors:Mark DeMaria (PI, CIRA/Colorado State University), Jun Zhang (University of Miami), Andrew Hazelton (University of Miami) and Lixin Lu (CIRA/Colorado State University)
Abstract: Problem Introduction and Rationale: The National Weather Service (NWS) is tasked with forecasting the sustained surface winds and wind gusts from tropical cyclones (TCs) over water and land. The wind gusts forecasts are critical in decision support briefings to emergency managers since high wind gusts can occur outside of areas with hurricane and tropical storm watches and warnings, which are based on sustained winds, and because structural damage, power outages, and other wind hazards are typically associated with gusts. The National Hurricane Center (NHC) official forecasts include gust estimates out to 5 days based upon the maximum sustained winds while local NWS Weather Forecast Offices (WFOs) provide gridded fields of sustained winds and wind gusts out to 7 days. Despite its importance, operational wind gust forecasts rely on climatological gust factors because there is no reliable dynamical model guidance available. For example, 99% of the operational NHC wind gust forecasts for the past three seasons were based on a lookup table that is a simple function of the sustained wind. The lack of TC wind gust information is a significant gap in NWS modeling capabilities and is the primary rationale for this proposal.
Summary of Proposed Research: In this project, we propose to develop more accurate gust parameterizations for the Hurricane Analysis and Forecast System (HAFS) and Global Forecast System (GFS) through a hierarchy of methods and to demonstrate these new capabilities to NWS forecasters at NHC and WFOs. The methods range from simple statistical techniques based on updated climatological relationships, more sophisticated empirical methods based on machine learning with input from the HAFS and GFS forecasts, and an improved version of the physically based gust model from the fifth generation ECMWF reanalysis (ERA5), where a diagnostic parameterization produces instantaneous gust magnitudes from the turbulent and convective components of gusts. The project leverages surface (10 m) wind databases of sustained wind and gusts that were constructed for landfalling TCs from 2011-2021 under previous NOAA-supported projects, which will continue to be expanded during the project to include additional cases. These databases will be used to examine the dependence of the ratio of gusts to sustained winds (gust factors) on multiple variables, including the sustained wind speed and direction, the distance from the storm center, surface roughness and low-level vertical shear and static stability to guide the development of the gust algorithms.
The NOAA collaborations with NWS forecasters and modelers (letters of support are included) will provide valuable feedback on the operational utility of the gust algorithms and to ensure consistency with the operational Unified Forecast System (UFS) to set the stage for operational transition. The coordination with AOML and NESDIS will leverage data analysis methods, hurricane model diagnostics and machine learning experience.
Title | PIs and Co-PIs | Affliation |
---|---|---|
JEDI-based Multi-resolution and Multiscale Hybrid 4DEnVar to Improve UFS GFScoupled data assimilation and prediction |
PI: Xuguang Wang | University of Oklahoma |
Abstract
JEDI-based Multi-resolution and Multiscale Hybrid 4DEnVar to Improve UFS GFS coupled data assimilation and prediction
Author:Xuguang Wang (PI, University of Oklahoma)
Abstract:GFSv16 with the FV3 dynamic core was operationally implemented for the UFS Medium-Range Weather prediction beginning March 2021. GFSv16 adopts the 4DEnVar data assimilation method. Under the support of UFS-R2O, NOAA and its partners are making efforts to further develop GFSv17 toward a weakly coupled data assimilation (DA) and forecast system by FY24. A JEDI infrastructure is developed by JCSDA and its partners to bring all components of the earth system into one DA system. As a critical component of the GFSv17, EMC in collaboration with JCSDA and NASA GMAO is currently developing a JEDI based prototype for the cycled GFS atmospheric DA, eventually replacing the legacy GSI system. With the increase of model resolution and the ability of assimilating all sky satellite radiance data, a large range of scales are resolved by the operational global DA system. Efforts are therefore needed to advance the global hybrid 4DEnVar for an effective multi-scale DA.
The OU team in close collaboration with the NOAA/NCEP/EMC, NOAA/ESRL/PSL and JCSDA collaborators proposes to (1) Implement the MR-ENS-SDL for 4DEnVar in JEDI; (2) Design and perform experiments to determine the optimal multi-resolution and multiscale configuration for GFS 4DEnVar using capabilities developed in (1); (3) Systematically evaluate the experiments in (2) on the analyses and subsequent global and hurricane predictions in the weakly coupled DA context and document the experiment results; (4) Transition the development and research to NCEP operational GFS hybrid DA system if results warrant.
The proposed work is in direct response to the NGGPS advances in data assimilation priority. The multi-resolution and multiscale capability will be tested along with the operational FV3GFS model and observations in the operational data stream, directly benefit operational forecasts, and involve important players from the NOAA research and operational centers and JCSDA.
Title | PIs and Co-PIs | Affliation |
---|---|---|
Strengthening NOAA’s Seasonal Prediction Capabilities by Improving the Ocean and Ice Components in the Unified Forecast System
|
PI: Rainer Bleck | CIRES, University of Colorado |
Shan Sun |
ESRL/GSL | |
Yuejian Zhu | NCEP/EMC | |
Benjamin W. Green | CIRES, University of Colorado |
Abstract
Strengthening NOAA’s Seasonal Prediction Capabilities by Improving the Ocean and Ice Components in the UFS
Authors:Rainer Bleck (PI, CIRES, University of Colorado), Shan Sun (ESRL/GSL), Yuejian Zhu (NCEP/EMC), Benjamin W. Green (CIRES, University of Colorado)
Abstract:Seasonal forecasts provide valuable information to many sectors of society. The Unified Forecast System (UFS) is designed to support the Weather Enterprise as a comprehensive system for NOAA's operational numerical weather prediction. The newly developed UFS-based Seasonal Forecast System (SFS) consists of six components including atmosphere, ocean, land, ice, waves and atmospheric composition. Substantial development and testing will still be required before this system can replace the current Climate Forecast System version 2 (CFSv2), which has been running operationally since 2011. The work proposed here is intended to contribute to this goal.
Specifically, we propose to improve the presentation of ocean and ice components carried over from the current UFS S2S Prototype that are suspected of limiting the seasonal forecast skills at this time. We will start with SFS version 0 (SFSv0), which is an extension of the UFS subseasonal forecast system S2S Prototype 8 (P8) and Ensemble Prototype 3 (EP3). Emphasis will be on (1) the mixed-layer parameterization and the vertical resolution in the mixed layer in the ocean model with an eye on the computation of air-sea fluxes and impact of SST uncertainty (especially in the tropics); (2) spatial grid resolution in the ocean and ice models; (3) fine-tuning of the ice model parameterizations; and (4) the sensitivity to the initial conditions, especially ice thickness. We will evaluate SFSv0 with the current operational CFSv2 as the baseline.
This work will strengthen NOAA’s seasonal prediction capabilities by identifying and reducing model biases in the ocean and ice components of the SFSv0. We plan to evaluate model biases, errors, and forecast skill of the model ensemble by running a set of retrospective runs. The runs will be designed to isolate shortcomings in the treatment of physical processes in the coupled system.
Title | PIs and Co-PIs | Affliation |
---|---|---|
Using Model Evaluation Tools (METplus) to Evaluate Process Related Precipitation Skill and Biases in the NOAA Seasonal Forecast System (SFS) over North America to Improve Climate Prediction Center (CPC) Operational Seasonal Forecasts | PI: Ben P. Kirtman | University of Miami |
PI: Tara Jensen | NCAR | |
Dan C. Collins, Johnna M. Infanti | NCEP/CPC |
Abstract
Using Model Evaluation Tools (METplus) to Evaluate Process Related Precipitation Skill and Biases in the NOAA Seasonal Forecast System (SFS) over North America to Improve Climate Prediction Center
Authors:Benjamin Kirtman (PI, Univ. of Miami), Tara Jensen (PI, NCAR) Dan Collins (CPC) Johnna Infanti (CPC)
Abstract:Integration of the NOAA Unified Forecast System (UFS) Seasonal Forecast System (SFS) into the seasonal research and forecasting communities, including University of Miami and the Climate Prediction Center (CPC) relies on assessment of skill and biases of precipitation over North America in both hindcasts and realtime forecasts. The National Center for Atmospheric Research (NCAR)’s enhanced Model Evaluation Tools (METplus) verification framework is intended to be used to verify the UFS, and is currently being onboarded for operational use at CPC due to its large library of verification metrics and community support approach. A currently funded collaborative effort between NCAR and CPC shows that METplus requires more development to seamlessly integrate with seasonal climate data, such as UFS-SFS and the North American Multi-Model Ensemble (NMME) (part a). Moreover, CPC seasonal forecasters rely on the state of primary climate drivers to forecast seasonal precipitation, and information on these drivers is imperative to the seasonal climate research and modeling communities. Thus, the assessment of the impact of El Niño Southern Oscillation (ENSO), decadal trends, etc. on North American precipitation variability is key to diagnosing the utility of any dynamical models used in seasonal forecasting and research. Though ENSO plays a key role in precipitation variability, other climate drivers should also be considered. For example, key internal forcing mechanisms such as thePacific Decadal Oscillation (PDO) and its impact on North American precipitation in seasonal forecast systems must be assessed, as well as the representation of in-situ drivers such as soil moisture and snow cover (part b). Collaboratively, we will create a verification framework utilizing METplus to allow streamlined assessment of probabilistic seasonal precipitation forecast skill, including hindcast and conditional skill related to the above key drivers within the UFS-SFS. An additional goal will be that it can be easily expanded to any climate model ensemble. The development, documentation, and demonstration of these process-based model capabilities will provide valuable feedback to the UFS model development team and community, with the potential to improve the key modes of variability that impact seasonal precipitation forecasts. Broader Impacts and Relevance: This proposal is relevant to Focus Area B. The creation of new METplus modules to facilitate the analysis of seasonal forecast verification metrics would: 1) Expand the capabilities of METplus to facilitate model evaluation at longer forecast time-scales and 2) Help guide the improvement of the SFS through improved understanding of the sources of biases. Focusing the analysis on ENSO variability and decadal variability and long-term trends would advance the ability of SFS to predict seasonal precipitation, potentially increase precipitation forecast skill and predictive lead time, and expand capacity for downstream applications. Such actions will improve the resilience of institutions that depend on accurate forecasts of seasonal precipitation.