8 Decades of Weather Radar History: The Technology in Operational Use at the KMA By 2025 Ken Crawford, Vice Administrator Korea Meteorological Administration International Weather Radar Workshop Daegu, Korea November 1, 2012 With the grateful assistance of: Don Burgess, Retired NSSL In The Beginning. Part of Ken Crawford s Career Sound Mirrors After World War I, the threat of an air attack was recognized in England. Sound mirrors were developed to detect incoming aircraft. But the increased speed of the airplanes meant warning times of only ~4 minutes. During a demonstration setup for the Air Ministry in England, officials were nearly deafened by a horse-drawn milk cart in 1935. Sound mirrors were abandoned. The 1930s 1
Development of the Magnetron The US had developed sophisticated receivers and antenna, but lacked a good transmitter. The UK magnetron completed the system, but it was Churchill who decided to make a full disclosure offer to the US with no strings attached. A 10-cm magnetron offered the potential for better resolving power, smaller antennas, and lighter weight equipment. The MIT Radiation Laboratory was formed in November 1940 and cream-of-the-crop physicists were recruited. The 1940s Post World War II The 1940s Secrecy was no longer important. Project Stormy Weather in Canada begun in 1943 First led By J. Stewart Marshall o Met Walter Palmer while working in Ottawa o Investigated Z-R Relationships o Z = 200*R 1.6 by Marshall and Palmer in 1948 (other researchers were suspicious ) U.S. Air Force Flying Weather Radars begun in 1945 David Atlas was among the first to lead Weather Radar Research Project begun at MIT in February 1946 Alan Bemis, Initial Director Thunderstorm Project in Florida/Ohio in held 1946-47 1st Weather Radar Conference at MIT March 1947 U.S. Weather Bureau obtained 25 Air Force Weather Radars; Renamed them WSR-1s, WSR-3s, etc. The 1940s 2
Hook Echo: The First Radar Tornado Signature April 9, 1953 Illinois State Water Survey Champaign, IL Obtained by Accident 1950s radar research led to development of the WSR-57 radars (1957 1995) WSR-57: 10 cm wavelength, 2 deg beamwidth NWS (Weather Bureau) began issuing tornado warnings in 1956 and 1957 Early on, hook echoes occurred with only 11% of all tornadoes tornado warnings not very good APS-15 (Airborne Radar from WWII) The 1950s 3 cm wavelength, 3 deg beamwidth WSR-57 Radar Phosphorus Display: image disappeared in a few seconds Grease Pencil: outline cells/lines at different times to calculate movement RHI: stop antenna, take RHI, no volume scan The 1950s 3
Research Doppler Radar 1958 Weather Bureau/NSSL (1958-1970) 1970) Navy Radar (WWII), 3 cm wavelength, 1.8 deg beamwidth The 1960s Norman Doppler Observation of the Union City Tornado May 24, 1973 First Real Time Tornado Detection First Color Doppler Display Display (not real time) Coordination with NSSL Storm Intercept Project The 1970s 4
Joint Doppler Operational Project (1976-1979) 1979) NSSL, NWS, USAF, and FAA personnel Norman Doppler Radar Simulated warnings compared to NWS/USAF warnings 1977-19791979 POD improved by 28% FAR improved by 31% LT improved by 16 min The 1970s The 1980s 1990s 5
Role of Weather Radar Training In The USA: HOW? 1990 2010 Role of Weather Radar Training In The USA: WHY?/WHERE? During the next 10 years, Dual Polarization will be followed by optimized scan strategies and by techniques to combine Doppler scans and reflectivity scans into a single best scan. These concepts plus the routine interpretation of Dual-Pol images illustrated on a another slide provide compelling evidence as to Why rigorous training is needed. The WDTB / the ROC work closely together to ensure that upgraded software/hardware components & the training regime are compatible. In Korea, the KMA should establish a disciplined, experienced, and well-staffed training unit that will: o On a regular basis, work very closely with the WRC staff; o Critique software modifications the WRC wishes to implement to ensure the proposed modifications are internally consistent; o Ensure that the on-going training remains compatible with the system software developed at the WRC. 2000 2015 6
Role of Weather Radar Training For Korea: HOW? The training technology will use (example next slide): o Distance learning technology; o Simulators at the primary training unit in Seoul 1 ; o The SOO 1 located at each Regional Forecast Office; o Live, web-based instructors; and o A new approach featuring the Storm of the Season (i.e., best cases captured by a given Forecast Office). With the use of these new features in training, the Regional Forecast Offices will listen in as new knowledge is shared to enable the KMA staff to receive in-depth training during the next several years. Related seminars and symposiums will be conducted during which the various Forecast Offices will share their vision of best practices. 2010 2020 Reflectivity Versus Correlation Coefficients (A Tool to Discriminate Efficiently Between Snow, Ground Clutter, & Chaff) SNOW ~0.85-1.00 Reflectivity (Z h ) 2010 2020 CLUTTER ~0.5-0.85 CHAFF ~0.2-0.5 Correlation Coefficient (ρ hv ) 7
Another Set of Examples Z Z DR 2010 2020 Future Evolution of Warning Decision-Making Science Present 2010 (± 2 yr) 2017 (± 5 yr) 2025 (± 10 yr) WSR-88D USA Dual-Polarization Radar Phased Array Radar Existing storms Newly initiated convection Forecast convection (does not yet exist) 2005 2020 8
WoF 2020 Vision For USA: Ultra High-Resolution Radar Data and Storm-scale Models for Numerical Weather Prediction 4 of 5 WSR-88D TDWR Data Quality Control Data Assimilation Storm-scale Analysis Storm-scale Forecast Visualization WSR-88D TDWR PAR Model Analysis + Forecast! PAR Reflectivity 2015 2025 NSSL Warn on Forecast Briefing March 5, 2007 Radial Velocity Initial threat area 30 min. threat probability 1 hr threat swath (accum) Probabilistic Warning Guidance Future building blocks: Dual-pol radar output Expanded enhanced severe storm verification Scientific advancement: Straight-line winds Climatological storm type statistics Greater computational ability Estimated time of arrival 2015 2025 9
Science and Technology Enables Storm-scale Numerical Weather Prediction 1975 NWP Model Grid North Texas Region (DFW area) 1975: was represented by a single grid point in our NWP models North Texas Region (DFW area) 2005: now can be represented by ~ 10,000 grid points Can resolve a single tornadic storm.. 1975 2007 DFW Model grid resolution ~ 200 km Tornadic Supercell Model grid resolution ~ 2 km 30 years of atmospheric research + computers which are 10 million times faster NSSL Warn on Forecast Briefing March 5, 2007 2010 2020 in Korea Warn on Forecast Forecast in the 2020s Now: NWS warning operation warnings (~ 13 min lead time) 3-6 hours: Large-scale weather prediction model output 1-3 hours: Monitor surface and satellite for initiation 0-1 hour: Warn from a single Doppler radar, spotter reports 2022: NWS warning operations (~30+ min lead time?) 3-6 hours: Ultra-high resolution model output (e.g., CONUS@1 km) 1-3 hours: State and local mesonets, micronets, profilers, ultra-high-res satellite Other exotic data sources (e.g., UAV s?) 0-1 hour: Multiple radars in metropolitan areas (public and private) 5-10x more data per radar (PAR, polarization variables) Live streaming video from news media and public Overwhelming Gap filling, mobile radars (public and private?) Complexity! NWS PAR Surface Ob Storm-Scale NWP Probabilistic 0-1 hour forecasts of individual storms, potential damage swaths from tornadoes, Spotter Public reports Manageable Complexity? Severe weather warnings 2025: Mesoscale Analysis/Fcst Fire Forecaster Increased lead time Hydrant GapF radar having dense Data Stream TV Radar Tuesday, November 20, wind, and hail GIS information content 2012 Satellite OAR PPES Briefing Information 20 NSSL Warn on Forecast Briefing March 5, 2007 2020 2030 10
Warn on Forecast in 2025: What might it look like? Radar and Initial Forecast at 2100 CST Radar at 2130 CST: Accurate Forecast An ensemble of storm-scale NWP models predict the path of a potentially tornadic supercell during the next 1 hour. The ensemble is used to create a probabilistic tornado warning. Probabilistic tornado warning: Forecast looks on track, storm circulation (hook echo) is tracking along centerline of highest tornadic probabilities Developing thunderstorm 50% 30% Most Likely Tornado Path 50% 30% Most Likely Tornado Path 70% 70% T=2150 T=2140 T=2130 T=2120 CST T=2200 CST T=2200 CST T=2150 T=2140 T=2130 T=2120 CST NSSL Warn on Forecast Briefing March 5, 2007 2020 2030 By The Year 2030. (100 years after the first sound mirrors) Radar data quality will improve so much over today s initial Dual Polarization data that it will be fed directly into our best storm scale models. The POD will exceed 0.9 for advanced notice of flood producing rainfall on the 0-1 hour observation scale and on the 2-6 hour modeling scale. QPEs from 0-24 hours will have improved enough to be fed directly into our best hydrologic models. As a result of improved QPEs, reliable decision support systems will enable the management of water resources. Those in this audience today who are younger than 40 years old will contribute to these accomplishments in the name of improved services. 11
Summary Long History of Radar Development Has Helped Improve Forecasts/Warnings/Safety Current Skill is Relatively Good Attractive Paths Exist for Future Development Ambitious Improvement Goals Have Been Set Are They Achievable? 12