NSSL researcher wins honors in AMS Artificial Intelligence competition

Kim Elmore
Kim Elmore

NSSL/CIMMS researcher Kim Elmore received second place in the American Meteorological Society Artificial Intelligence Competition.  This is the third year of the contest.
To compete, participants are given a data set, then tasked with defining or detecting some weather-related phenomena based the provided data.  The entrants are also requested to present a paper on the method used.

This year’s task was to make probability forecasts of moderate or greater turbulence for airline flights using over 100,000 observations and 130 variables. Elmore, along with co-participant University of Oklahoma School of Meteorology Professor Mike Richman, used an ensemble tree regression method to solve the problem, and were awarded second place.

Elmore and Richman co-chaired last year’s competition, and was invited to compete in the 2009 event.  As a result of Elmore’s efforts, he has been invited to become a member of the AMS Committee on Artificial Intelligence.

Share this:

Storm Tracker tool in development

An online interactive tool to automatically identify and track convective clusters from satellite and radar data has been developed by NSSL’s Bob Rabin and Tom Whittaker from the Space Science and Engineering Center /Cooperative Institute for Mesoscale Satellite Studies (CIMSS) at the University of Wisconsin-Madison with assistance from V. Lakshmanan at the Cooperative Institute for Mesoscale Meteorological Studies (CIMMS).  The system is designed for research and operational communities to have remote access to both near real-time and archived multi-sensor data including radar, satellite, lightning, and model analyses and forecasts.

Storm Tracker
Storm Tracker

Storm Tracker expands on the concepts of radar tracking algorithms, but uses satellite images instead of radar to track cells.  This capability allows it to operate on large storms such as a Mesoscale Convective System (MCS) or tropical storms.  Multi-sensor data is also available to study the characteristics of each particular cell.  Each storm feature has an identification number, and retains that ID as the cell moves.

Storm Tracker is currently being used to study which (if any) multisensory characteristics can be used to anticipate growth rates and longevity of MCS.  This project is in collaboration with the NESDIS CREST (NOAA-Cooperative Remote Sensing Science and Technology Center) at the City College of NY.

Current implementation of the tool is available at:  stormtrack.nssl.noaa.gov

This work has been funded in part by the NOAA HPCC program.

Share this:

Tropical Storm Ida gives CI-FLOW research opportunity

Tropical Storm Ida gave the Coastal and Inland – Flooding Observation and Warning project (CI-FLOW) team a valuable research opportunity this week to demonstrate, in real-time, the capability to use NSSL’s real-time gridded quantitative precipitation estimates (QPE) in the CI-FLOW river models.

NSSL is leading CI-FLOW, an interdisciplinary multi-institutional team working to combine existing monitoring technology and new techniques to forecast and warn of coastal storm effects such as heavy rainfall, storm surge, and the subsequent river conditions in coastal North Carolina.  CI-FLOW will ultimately provide a total water-level product for any location in the watershed.

For this demonstration, the newly developed CI-FLOW computing environment collected hourly multi-sensor quantitative precipitation estimates from the NSSL Q2 system (nmq.ou.edu) and gridded quantitative precipitation forecast products from NOAA’s Hydrometeorological Prediction Center (HPC).  The data was fed into one of two CIFLOW models, the NWS HL-RDHM (Hydrologic Laboratory Research Distributed Hydrologic Model), to generate 10-day forecasts of streamflow from multiple points in the Tar-Pamlico and Neuse River basins.

CI-FLOW also tested NOAA nowCOAST (nowcoast.noaa.gov) CIFLOW visualization capabilities, supported by NOAA Southeastern- Caribbean Regional Team (SECART) funding, to explore how data can be displayed for stakeholder outreach by NOAA SeaGrant and NWS offices as well as internally for science assessments by CI-FLOW team members.

As CIFLOW closes out the 2009 Atlantic hurricane season, CIFLOW partners will continue to leverage a NOAA Integrated Ocean Observing System (IOOS) research effort to stabilize the Advanced Circulation model (ADCIRC) grid to accommodate inland rivers and bays in the North Carolina tidal zones.  This will allow CIFLOW to complete a demonstration of the CIFLOW coupled model system to produce simulations of total water level for the lower portions of the Tar-Pamlico and Neuse Rivers and coastline of the Pamlico Sound using past storm events including Hurricane Isabel.

Share this:

GOES-R Proving Ground Activities at NSSL and SPC kicks off

A new program to extend the use of geostationary satellite data in the operational environment has kicked off this spring in the NOAA Hazardous Weather Testbed at the National Weather Center in Norman, Okla.  The GOES-R PG will involve the operational forecast community in the assessment and development of techniques for the next generation GOES satellites (GOES-R).

The “GOES-R Proving Ground” (PG) is sponsored by the NESDIS GOES-R program office and will take place over the next several years at the Storm Prediction Center in collaboration with NSSL and the Norman, Okla. National Weather Service Forecast Office.

During the first year, the GOES-R PG expects to lay foundational relationships and develop test methods that will lead to optimal testing of suites of products in following years.  Techniques to be tested include:

  • short-term forecasting of developing thunderstorms based on observations of cloud top cooling and development of ice in cloud tops (Cooperative Institute of Meteorological Satellite Studies (CIMSS) at the University of Wisconsin-Madison)
  • simulated observations expected to be available from the optical lightning mapper on the GOES-R satellites
  • short-term forecasts of hail probabilities based on cloud top information and environmental conditions (CIRA/NESDIS)

Most of these products will also be available for evaluation by the Experimental Warning Program and VORTEX2 forecasters this spring.

Background: The GOES-R is scheduled for launch in 2016.  The availability of GOES-R products pre-launch will demonstrate a portion of the full observing capability of the GOES-R system.

Share this:

2009 HWT Spring Experiment: Experimental Forecast Program

The Experimental Forecasting Program (EFP) branch of the NOAA Hazardous Weather Testbed conducted its annual 2009 Spring Experiment, organized by the SPC and the National Severe Storms Laboratory, from May 4 through June 5.   Approximately 40 visiting scientists, model developers, operational forecasters, faculty members and graduate students from the U.S., Canada, and the United Kingdom collaborated with SPC and NSSL forecasters and scientists during the five-week program.

As in recent years, the primary focus was the utilization of convection-allowing model forecasts as guidance for the prediction of severe convective weather.   EFP participants explored a variety of model output for severe storm prediction and aviation impacts. The model guidance was generated by the University of Oklahoma’s Center for Analysis and Prediction of Storms (OU-CAPS), the National Centers for Environmental Prediction Environmental Modeling Center (NCEP-EMC), the National Center for Atmospheric Research (NCAR), the National Severe Storms Laboratory (NSSL), the Air Force Weather Agency (AFWA), and the Earth System Research Laboratory Global System Division (ESRL/GSD).

EFP Participants looked at products derived from both ensembles and deterministic forecasts.  This revealed sensitivities in the model output to mesoscale and larger scale initial and lateral boundary conditions, radar data assimilation, model physics, model dynamic cores, model resolution, and updates to initial conditions (e.g., 1200 vs. 0000 UTC initialization).  Some of these sensitivities were assessed subjectively, but we also utilized the Developmental Testbed Center’s Model Evaluation Tools for real-time and post-experiment analysis.

The combination of the unique framework of the HWT and synergy with VORTEX II and the new GOES-R Proving Ground made the 2009 Spring Experiment an exciting and scientifically stimulating event.

Share this:

NOAA Hazardous Weather Testbed 2009 Spring Experiment

Each year dozens of visiting scientists, model developers, faculty members and graduate students from around the world gather for the NOAA Hazardous Weather Testbed Spring Experiment.  This is the ninth year of the program designed by National Severe Storms Laboratory (NSSL), the Storm Prediction Center and the National Weather Service to foster improved severe weather forecasts and warnings.

The Hazardous Weather Testbed (HWT) currently has two branches, the Experimental Forecasting Program (EFP) and the Experimental Warning Program (EWP).  During Spring Experiment operations each functions on a slightly different schedule with different but complementary roles and goals

The EFP Spring Experiment will run from May 4 through June 5, and will focus on using convection-allowing model forecasts as guidance for the prediction of severe convective weather.

The experimental models will be generated by a number of collaborators including the University of Oklahoma’s Center for Analysis and Prediction of Storms (OU-CAPS), the NOAA National Centers for Environmental Prediction Environmental Modeling Center (NCEP/EMC), the National Center for Atmospheric Research (NCAR), NSSL, the Air Force Weather Agency (AFWA), and the NOAA Earth System Research Laboratory Global System Division (ESRL/GSD).  Participants will assess strengths and weaknesses of the models using various verification approaches, including new methods being developed at the cooperative Developmental Testbed Center.  Furthermore, they will explore new data assimilation strategies and their potential impact on forecasting.  Unique to this year will be the synergy with a project to study tornadoes, VORTEX2 and the new GOES-R Proving Ground project both located in the National Weather Center.

The EWP Spring Experiment is focusing on shorter-term convective weather warning needs of forecasters and will run for six weeks:  April 27- May 22, and June 1-12.  The EWP will test and evaluate emerging technologies and science for WFO severe convective weather warning operations in Weather Forecast Offices (WFO).  There will be four projects geared toward WFO warning decision-making applications:

  • An evaluation of experimental multiple-radar/sensor gridded severe weather algorithm products using the NSSL Warning Decision Support System II (WDSSII);
  • An evaluation of the 3D Lightning Mapping Arrays (LMA) in Central Oklahoma, Northern Alabama, the Washington D.C. Metro Area, and possibly East-Central Florida;
  • An evaluation of networked 3-cm radars (CASA) in Central Oklahoma;
  • An evaluation of the phased array radar (PAR) in Norman, Okla.
Share this:

The Shadow Forecast Program at the NOAA Hazardous Weather Testbed

NSSL scientists are shadowing NOAA Storm Prediction Center (SPC) operational forecasters this spring to immerse themselves in the front-line operational and scientific challenges associated with forecasting mesoscale hazardous weather.

The program enables NSSL scientists to observe and interact with SPC forecasters at various forecast desks on an occasional basis, during a variety of operational forecasting scenarios. These can range from quieter convective days when there can be more opportunity for in-depth discussion, to more active severe weather days when direct interaction may be minimized but valuable insights can still be gained through observation of the operational decision-making processes.

Shadow shift participants will be looking at SPC forecasting responsibilities, division of duties at various forecast desks and the SPC’s team approach to forecasting. They will also study specific SPC scheduled and event-driven forecast products and their content. They will examine how forecasters utilize scientific knowledge, observational and model datasets, and technological tools to form and issue SPC forecast products.

Background: A culture of collaboration has existed for many years between NSSL researchers and National Weather Service forecasters, beginning with the development of Doppler radar technologies in the 1970s.

Significance: The shadow forecast program immerses scientists in the operational forecasting environment, giving them the insight needed to enhance the operational relevance of their research efforts and accelerate the transfer of science and technology from research to operations.

Share this:

Operational users evaluate new NSSL radar products

An upgrade to the national network of WSR-88D (NEXRAD) weather radars will be completed by 2010. The improvements will include a dual-polarization capability, allowing the radar can transmit pulses in both horizontal and vertical orientations (the WSR-88D currently only transmits horizontal pulses). Dual-polarization will provide significant improvements in rainfall estimation, precipitation classification (rain, hail, snow, sleet), and weather hazard detection.

NSSL is working to extend the effectiveness of this upgrade by developing products to assist users in operational decision-making during severe local storms, including flooding and winter storms. The products were evaluated by operational users, and results were published in a paper for the 33rd Conference on Radar Meteorology this fall.

Field evaluators consisted of a variety of users ranging from experienced users of dual-pol products to forecasters who frequently use WSR-88D data during the daily weather-related operations.

Seven dual-pol WSR-88D datasets were chosen to cover a variety of high-impact weather events including significant winter storms, severe thunderstorms, mixed precipitation phases and heavy rainfall. Evaluators were asked to provide feedback on the candidate dual-pol base products and associated algorithms to estimate how they might benefit operational meteorologists.

Positive results included:

– The “Rain/hail” category was useful to operations as a “flag” of locations where the meteorologist should further scrutinize base products

– Precipitation classification and melting layer capabilities may be very helpful in forecast operations, particularly in winter storms

– Major improvements to remote quantitative precipitation estimation are likely, particularly in flash flood warning decision making and forecasting

– Instantaneous rainfall rate information was found to be potentially useful in flash flood warning operations

Background: NSSL has been a leader and major contributor to the scientific and engineering development of polarimetric weather radar for the past 20 years.

Significance: The additional information from vertical pulses provided by dual-polarized Doppler radar will greatly improve the quality and accuracy of many different types of forecasts and warnings for hazardous weather.

Share this:

NSSL’s new radar “slice and dice” tool helps in storm warnings

NSSL’s prototype Four-Dimensional Stormcell Investigator (FSI), a 3D/4D base radar data display tool, was alpha-tested in three NWS Weather Forecast Offices (WFOs) this summer – Melbourne, Fla., Omaha, Neb., and Huntsville, Ala. FSI is based on NSSL’s Warning Decision Support System – Integrated Information (WDSS-II), and allows users to create and manipulate dynamic cross-sections (both vertical and at constant altitude). Forecasters can “slice and dice” storms and view these data in three dimensions and across time.

Feedback was positive – with one forecaster commenting: “Overall, we found the FSI utility to be quite useful in the Warning Decision Making process. Specifically, the cross-section and 3D display capabilities did a very nice job showing the intensifying structure and increasing hail potential with the storm that tracked through Madison County.”

FSI will be implemented at all WFOs in AWIPS Build OB8.2 from Dec. 07 through Mar. 08.

Background: The FSI is one of the many activities being coordinated between NSSL and the NWS Meteorological Development Laboratory to assist NWS meteorologists in making hazardous weather warning decisions.

Significance: This is another successful example of OAR research transferred to NWS operations.


Share this: