Presentation on theme: "Why I am CORRECT far more often than TV...NWS... and The WEATHER CHANNEL."— Presentation transcript:
Why I am CORRECT far more often than TV...NWS... and The WEATHER CHANNEL
The WXRISK PHILOSOPHY To strike the perfect balance between CW - Conventional Wisdom - and New Ideas… without coming across as too contrarian. Use of new lessor known weather models and techniques When everyone is thinking the same… someone isnt thinking… George S Patton
NOTEABLE SUCCESSFUL WxRisk FORECASTS S American Heat Wave / Drought FEB - MARCH 2005 Hurricane Season 2004...FLORIDA Charley Frances Ivan Jeanne S American Heat Wave / Drought FEB - MARCH 2004 2006 HURRICANE SEASON -- only well known PWSIP to forecast BELOW NORMAL HURRICANE ACTVITY
NOTEABLE SUCCESSFUL WxRisk FORECASTS SUPER MILD WINTER OF 2005-2006... Consensus called for Cold snowy winter over Eastern US 2006-07 EL NINO WINTER.... MILD.... SEVERE WINTER PATTERN "FLIP" JAN 15 - MAR 15 ACTIVE HURRICANE SEAS0N 2010 … NO MAJOR US LANDFALLS: Accu wx ( Accu- Blunder) 7 US Landfalls 3 Intense Hurricanes SEVERE EAST COAST WINTER 2009-10
NOTEABLE VA SUCCESS
1998 SEVERE ICE STORM CENTRAL VA --- UKROPS HURRICANE ISABEL -- 7 day forecast that was within 100 miles of actual landfall...1 st time WRVA 1140AM SOUTHEAST VA MAJOR SNOWSTORM 12/25/04 NWS 1-3 for Norfolk VA beach... 3-6 for Newport News / Williamsburg... WXRISK 6-10 Norfolk VA Beach.... 12 plus Peninsula. 16 Williamsburg 14 Newport News 9 Norfolk
19-20 DEC 2009 MAJOR VA /MD / DE SNOWSTORM.... 5 DAYS OUT called for 10-20" NWS and local TV 3-6"....6-12 N & W NIGHT BEFORE STORM
DEC 19-20 SNOWSTORM
30 JAN 2010.... WXRISK calls for 8-15" FIVE DAYS OUT... FRONT PAGE OF RTD... FEB 5-6...WXRISK 3 days out 3-6" RIC 6-10" N & W NWS calls for ALL RAIN 3 DAYS OUT... BLIZZARD FOR DCA DEC 25-26 2011.... 6 DAYS OUT... WTVR
DEC 25-26 SNOWSTORM DEC 18-25 192,000 HITS DEC 19- JAN 20 1,.741,671 hits
WHY ARE MAJOR WEATHER EVENTS MISSED IN THE DAY 3 - 10? SEVERAL REASONS OLD SCHOOL * Dont bother past 3 days * All the glory is on MESO scale weather MODERN WEATHER SCHOOLS not teaching ANYTHING about Day 3-14. EXPERIENCE HAS LESS VALUE Those forecasters who have been doing this for years are unfamiliar with new models & methods.
SEQUENCE OF EVENTS SPECIFIC ORDER A+ B + C+ D = FORECAST... IF B does not show up... FORECAST cannot remain the same WHAT IS A WEATHER FORECAST ?
WHAT ARE WEATHER MODELS? Computer Representations of what the atmosphere looks like and what it SHOULD look like over the next several days. – Very Much similar to the Exit polling Debacle in 2000 and 2004 Prez elections… GIGO … Better results CONNECT THE DOTS! … that is how the computer models do it.
CHAOS Theory -- Dr Lorenz in the 60s from MIT… was developed by a Meteorologist. Ironic isnt it? If you run a model from the same starting point with all of the exact same data you SHOULD get the exact same results... NOT in WEATHER. WHY??! – Bird flaps its wings in Tokyo and it rains in NY 10 days later…. WHAT ARE WEATHER MODELS?
The early weather models in 1960s 70s and 80s had Grid Points 250 miles apart WHAT IS GRID POINT? – Specific Place where all the data for 1 location goes into the Model… Weather Balloons Upper Air reports Surface Data Ship reports Satellite data etc There is a lot of atmosphere in between those DOTS/ Grid points - PACIFIC OCEAN!!! Missing Data causes wild swings in the model with very different looking weather maps / solutions by day 3 say 5 Day 7 and beyond WHAT ARE WEATHER MODELS?
SOLUTION Increased computing Power means GRID points can be much closer … Less Coarse Looking for the 1 perfect Weather model RESULT Better forecasts in the Short Range No change at all in skill past 3 or 4 days WHAT ARE WEATHER MODELS?
NATURE OF WEATHER CHAOS says It is easy to get a sunny day in middle of MAY or JUNE correct. Much harder to forecast winter rainstorm… harder still a East coast snowstorm; clouds, precip, jet streams, amounts of Cold air vs. Mild air SOME DATA FROM THE PACIFIC OCEAN… NOT ENOUGH.... and the MISSING Data is a killer MOST FORECAST ERROR LIES IN THE DAY 3 to DAY 10 PERIOD
SHORT RANGE MODELS 6 hours to 66 hours ETA...NGM...GFS (American)... MM5 RGEM... and newest one=WRF. High resolution -- the DOTS are close together so these models handle things like Thunderstorm clusters coastal fronts ETA-10km/ 20km GFS=
MEDIUM RANGE MODELS 72 hours to DAY 7 or 168 hours American (GFS) European (ECMWF) Canadian (GGEM) British (UKMET) Navy (Nogaps). Used around the world or Hemisphere UKMET 50km GGEM -60KM GFS=40km to day 7 ECMWF 20KM-- best in the world
STATE OF FORECASTING EARLY IN 21st CENTURY Huge increase in Number of Weather Models but Most forecasters are NOT familiar with these new Models CONSENSUS FORECASTING works great when its sunny in Mid May. For Major wx events such as landfalling hurricane or snowstorms -- NOT! 6 Medium Range models… some times 6 different Solutions! Which one is right?
ENSEMBLE FORECASTING Started in 1990s All these new weather models can often VERY different solutions. Which one is right? PREMISE-- with all these new super powerful computers we cab now run the SAME weather model 10, 12, 15 even 30 times. Not 1 or 2 solutions but a wide array of possible outcomes for forecasters to consider WARNING SIGN-- when a forecaster makes statement about the good old days of weather forecasting when there were just 3 or 4 models
ENSEMBLE FORECASTING Ensembles allows a forecaster to see a SPREAD of possible solutions. This gives a forecaster a degree of certainty and probability that did NOT exists before.
HOW MANY ENSEMBLES ARE THERE? AMERICAN MODEL…21 members which are run 4 times a Day CANADIAN.… 20 members 2x/ day EUROPEAN ….. 51 members 2x/ day NAVY …………. 10 members 2x / day SREF…. Short range ensembles to 66 hrs…run 4 times a day.
OCT 1 GULF HURRICANE - GFS
EUROPEAN MODEL SAYS ???
HYPOTHETICAL: SNOWSTORM FOR THE MID - ATLANTIC What does the overall Pattern say? What do the 5 Medium Range models say? Suppose 2 models show 6 or more inches for RIC 2 Models show NO snow and 1 model 2 Snow then Rain 22 member American Model; Suppose 14 have significant snow... 4 have snow/Ice... 2 show Rain...2 have NO event at all. GO with the 14 and see what the TRENDS are. Check Canadian, European, Navy etc
PRESIDENTS DAY STORM FEB 15 - 17 2003
ECMWF MODEL from FEB 10
WHAT WXRISK DOES DIFFERENTLY SOE -- Sequence of Events A form of Risk Assessment of Possible outcomes A Weather Forecast is a series of events that you expect to happen over a certain interval of time. A+ B + C + D = WEATHER FORECAST If B does Not occur then the rest of the Forecast could not possibly be correct
Instead of being over whelmed by massive increase in weather models over the past 10 years… S.O.E. allows a forecaster to develop several scenarios or varying probabilities Convey to clients in the forecast in the WHAT COULD GO WRONG section More importantly Forecasters are very RARELY SURPRISED by sudden model changes and you stay ahead of the Curve WHAT WXRISK DOES DIFFERENTLY
GFS American MODEL classic case of False East coast snowstorm forecasted for DEC 19-20 2004
AMERICAN MODEL ENSEMBLES
HURRICANE FRANCES 2004 CONSENSUS FORECAST – Miss FL turn up the East Coast – move into SE Canada and shove early season cold air mass into Upper Plains/ Midwest = FROST
Huge increase in various short Range Medium Range and Long Range Models demand constant attention NOT more confusion… tracking 3 or 4 scenarios as possible weather outcomes… sometimes its too early to make a decision… but with S.O.E. forecasters can figure out WHEN the answer will become clear SOE allows forecasters to react much faster instead of waiting for the next model … and FRONT RUN the weather curve…. SUMMARY