Forecast Skill and Major Forecast Failures over the Northeastern Pacific and Western North America Lynn McMurdie and Cliff Mass University of Washington.

Slides:



Advertisements
Similar presentations
Investigating medium-range forecast uncertainty and large error cases along the East coast using ensemble and diagnostic tools  Minghua Zheng, Brian Colle,
Advertisements

The South American Monsoon System: Recent Evolution and Current Status
PDO/PNA The PDO (Pacific Decadal Oscillation) is an index derived from North Pacific sea surface temperature anomalies and it has a high correlation to.
Madden/Julian Oscillation: Recent Evolution, Current Status and Forecasts Update prepared by Climate Prediction Center / NCEP February 20, 2006.
2012: Hurricane Sandy 125 dead, 60+ billion dollars damage.
Analysis of Precipitation Distributions Associated with Two Cool-Season Cutoff Cyclones Melissa Payer, Lance F. Bosart, Daniel Keyser Department of Atmospheric.
Planetary and Synoptic Analysis of Freezing Rain Events in Montreal, Quebec Gina M. Ressler, Eyad H. Atallah, and John R. Gyakum Department of Atmospheric.
An Overview of Environmental Conditions and Forecast Implications of the 3 May 1999 Tornado Outbreak Richard L. Thompson and Roger Edwards Presentation.
NCEP Model Comparisons over the East and West Coast of the US Lynn McMurdie Garrett Wedam Cliff Mass May 27, 2009 Wedam, McMurdie and Mass, 2009: Comparison.
Using ensemble data assimilation to investigate the initial condition sensitivity of Western Pacific extratropical transitions Ryan D. Torn University.
Performance Characteristics of a Pseudo-operational Ensemble Kalman Filter April 2006, EnKF Wildflower Meeting Greg Hakim & Ryan Torn University of Washington.
Revolutions in Remote Sensing Greatly Enhanced Weather Prediction from the 1950s Through Today.
An Investigation of Cool Season Extratropical Cyclone Forecast Errors Within Operational Models Brian A. Colle 1 and Michael Charles 1,2 1 School of Marine.
East Coast vs. West Coast: A Documentation of Model Forecast Failures for Eta, NAM, GFS, GEM, and ECMWF Garrett Wedam Lynn McMurdie, Cliff Mass.
United States Coast Guard 1985 Evaluation of a Multi-Model Storm Surge Ensemble for the New York Metropolitan Region Brian A. Colle Tom Di Liberto Stony.
An Examination of the Tropical System – Induced Flooding in Central New York and Northeast Pennsylvania in 2004.
CSTAR Update: New Tools for More Efficient Use of Ensembles in Operations Brian A. Colle, Minghua Zheng, and Edmund K.M. Chang, School of Marine and Atmospheric.
A Brief History of Weather Forecasting. The Stone Age Prior to approximately 1955, forecasting was basically a subjective art, and not very skillful.
National Weather Service Model Flip-Flops and Forecast Opportunities Bernard N. Meisner Scientific Services Division NWS Southern Region Fort Worth, Texas.
Impact of Targeted Dropsonde Data on Mid-latitude Numerical Weather Forecasts during the 2011 Winter Storms Reconnaissance Program Presented by Tom Hamill.
The Development of a Wave Packet Tracking Algorithm: Preliminary Climatological and Model Verification Results Matthew Souders, Brian Colle, Edmund Chang.
A Comparison of the Northern American Regional Reanalysis (NARR) to an Ensemble of Analyses Including CFSR Wesley Ebisuzaki 1, Fedor Mesinger 2, Li Zhang.
Forecast Pressure. Pressure Observations ASOS is the best…the gold standard Ships generally the worst.
Alaska Coastal Climatologies Wind/Wave PRIDE Alaska Coastal Climatologies Wind/Wave Workshop Anchorage, Alaska August 2, 2005 Storm Track Capabilities.
Polar Prediction The Scientific Challenges - Antarctica John Turner British Antarctic Survey Cambridge, UK.
1 Rolf Langland Naval Research Laboratory – Monterey, CA Uncertainty in Operational Atmospheric Analyses.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 30 December 2013 For more information,
1 Results from Winter Storm Reconnaissance Program 2008 Yucheng SongIMSG/EMC/NCEP Zoltan TothEMC/NCEP/NWS Sharan MajumdarUniv. of Miami Mark ShirleyNCO/NCEP/NWS.
1 Results from Winter Storm Reconnaissance Program 2007 Yucheng SongIMSG/EMC/NCEP Zoltan TothEMC/NCEP/NWS Sharan MajumdarUniv. of Miami Mark ShirleyNCO/NCEP/NWS.
1 Rolf Langland NRL-Monterey Plans for Evaluation of Lidar Wind Observations at NRL-Monterey Working Group on Space-Based Lidar Winds 05 Feb 2008.
1 The Asian-Australian Monsoon System: Recent Evolution, Current Status and Prediction Update prepared by Climate Prediction Center / NCEP August 9, 2010.
The Operational Impact of QuikSCAT Winds at the NOAA Ocean Prediction Center Joe Sienkiewicz – NOAA Ocean Prediction Center Joan Von Ahn – STG/NESDIS ORA.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 4 January 2010 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 9 February 2015 For more information,
Interannual Time Scales: ENSO Decadal Time Scales: Basin Wide Variability (e.g. Pacific Decadal Oscillation, North Atlantic Oscillation) Longer Time Scales:
1 Examples: Predicting the Extreme Events. 2 Recent Forecast Examples: Extreme Events West Coast – November 1981  3-5 January 2008 Central U.S.
Madden/Julian Oscillation: Recent Evolution, Current Status and Forecasts Update prepared by Climate Prediction Center / NCEP January 29, 2007.
Madden/Julian Oscillation: Recent Evolution, Current Status and Forecasts Update prepared by Climate Prediction Center / NCEP March 12, 2007.
Impacts of Rossby Wave Packets on Forecast Uncertainties and Errors
Exploring Multi-Model Ensemble Performance in Extratropical Cyclones over Eastern North America and the Western Atlantic Ocean Nathan Korfe and Brian A.
Consolidated Seasonal Rainfall Guidance for Global Tropics, December 2015 Initial Conditions Issued 14 December 2015 Forecast Background – ENSO update.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 6 February 2012 For more information,
1 The Asian-Australian Monsoon System: Recent Evolution, Current Status and Prediction Update prepared by Climate Prediction Center / NCEP 14 July 2008.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 20 December 2010 For more information,
Madden/Julian Oscillation: Recent Evolution, Current Status and Forecasts Update prepared by Climate Prediction Center / NCEP November 6, 2006.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 7 November 2011 For more information,
Forecast Pressure. Pressure Observations ASOS is the best…the gold standard Ships generally the worst.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 18 April 2011 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 18 November 2012 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 18 May 2015 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 11 October 2009 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 23 March 2015 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 4 June 2012 For more information,
Madden/Julian Oscillation: Recent Evolution, Current Status and Forecasts Update prepared by Climate Prediction Center / NCEP July 31, 2006.
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 16 February 2015 For more information,
The South American Monsoon System: Recent Evolution and Current Status Update prepared by Climate Prediction Center / NCEP 30 March 2015 For more information,
Weather Regimes and Forecast Errors in the Pacific Northwest Lynn McMurdie (co-author Joe Casola) University of Washington.
Hemispheric Evolution and Origin of Extra-tropical Cyclone Errors Within NCEP Operational Models Brian A. Colle 1 and Michael Charles 1,2 1 School of Marine.
WRF-EnKF Lightning Assimilation Real-Observation Experiments Overview
The South American Monsoon System: Recent Evolution and Current Status
The El Niño/ Southern Oscillation (ENSO) Cycle Lab
A Review of the CSTAR Ensemble Tools Available for Operations
The Stone Age Prior to approximately 1960, forecasting was basically a subjective art, and not very skillful. Observations were sparse, with only a few.
Forecast Pressure.
Forecast Pressure.
Forecast Pressure.
University of Washington Center for Science in the Earth System
Forecast Pressure.
14th Cyclone Workshop Brian Ancell The University of Washington
Impact of aircraft data in the MSC forecast systems
Presentation transcript:

Forecast Skill and Major Forecast Failures over the Northeastern Pacific and Western North America Lynn McMurdie and Cliff Mass University of Washington

Why the North Pacific Ocean? The Pacific is one of the largest areas of sparse insitu observations in the world Uncertainty over the Pacific has a large impact on predictability over the North American continent and beyond There are often large initialization errors and short-term forecasts over the northern Pacific ocean. One symptom of such problems is that short-term forecast skill on the western side of North America is worse than in other areas.

An example of a short-term forecast error Eta 24-h 03 March 00UT 1999 Eta 48-h 03 March 00UT 1999

And the public and media have noticed these failures….

Seattle Times Eugene Register Guard

Some applicable research in the literature Langland et al. (2002) – poor forecast of Jan 2000 storm on East Coast due in part to sensitivity over the Pacific. Bosart et al. (2002) – lack of convection over the midwest not represented in forecasts due to poorly initialized trough along west coast Schultz et al. (2005) – 70% of troughs arriving on the West Coast were underforecast, a portion of which continued to effect short-term forecasts across the North American continent. McMurdie and Mass (2004) – documented forecast failures over the eastern Pacific.

This talk will … Demonstrate that large initial condition and short-term forecast errors still occur over the eastern Pacific and downstream Present a feature-based approach to monitoring errors in this region Discuss implications for THORPEX

How frequent are large numerical forecast errors? Approach: compare buoys/coastal observations of sea level pressure (SLP) to NCEP’s Eta and GFS 00, 24, 36 and 48 hr forecasts. Error = Forecast SLP – Observed SLP At each station, calculated average and absolute error and the standard deviation using winter (Oct – Mar) data. Large Error = |Error| > (average error + 2 * SD)

Station Locations Tatoosh Is. Cape Arago

24 h Large Errors Tatoosh Is., WA Cape Arago, OR Inter-annual variability

48-h Errors 48h errors much larger and more frequent than 24-h errors

GFS vs. Eta 24-h errors NCEP GFS better than Eta on average

48-h errors GFS over forecasts Eta under forecasts

GFS has more accurate SLP initializations and forecasts than Eta over the Northeast Pacific For 00-h forecasts (initial conditions), GFS has smaller mean absolute error (MAE) and standard deviation (SD) than Eta at all 17 stations For 24- and 36-h forecasts, GFS has smaller MAE and SD than Eta at 13/17 buoy and coastal stations For 48-h forecasts, GFS has smaller MAE and SD than Eta at 12/17 stations.

Forecast Verification: The Need for Feature-Based Evaluation Examining statistics at observing sites is not sufficient for understanding the problems. Must also track features to gain an understanding of the deficiencies. Case studies of major failures should reveal important information.

GFS What are these large forecast errors associated with?

How frequently do large forecast errors of synoptic events occur? Number of Events/Season associated with Lows/Troughs/Highs SeasonTotalLowTroughHigh 1999 – – – – (from McMurdie and Mass 2004) Event = large error at 2 or more adjacent stations for 2 or more forecasts periods Data shown for Eta model only

Of the forecast failures associated with lows, what are the central pressure and cyclone position errors? Ave SLP error = 3.4 mb SD = 8.7 mb Absolute error = 7.5 mb Ave position error = km SD = 260 km

Recent examples of major forecast errors February 2002 October 2003 February 2004 November 2004 April 2005 May 2005

An example of a recent high-impact, poorly forecast storm Power outages, large trees uprooted in Eugene, OR Powerful, rapidly developing storm with strong winds (70 kts) Very poor short- term numerical guidance L 1008 L 1004 L 996 L – 8 February 2002 Cyclone

48-hr Forecasts Valid 00 UTC 8 February 2002 AVN UKMO ETA NOGAPS

24-Hr Forecasts Valid 00 UTC 8 February 2002 AVN UKMO ETA NOGAPS

Difference between UKMO and Eta 850 mb Temperature K Valid 00 UTC 7 February 2002 Solid = UKMO, Dashed = ETA, Shades, blue = differences L 1010

20 Oct 03

Flood of 20 October 2003

00hr GFS24hr GFS48hr GFS 00hr + 48hr GFS

GFS Forecasts of 12-hr Precipitation 12h Forecast 24h Forecast 36h Forecast 48h Forecast

February 04

00UTC 16 Feb 04 GFS 00-hr Forecast

00 UTC 16 Feb hr GFS

00 UTC 16 Feb hr GFS

Apr 05

24-hr forecast GFS Position error ~ 420km

Large Short Term Forecast Errors Still Occur Number of slp errors > 10 mb continues to be 10 – 15 per winter (despite the ridge this year) Vast majority of large errors due to mispositioned or under (or over) forecast low centers (see McMurdie and Mass, 2004) For Feb 02 case, forecast errors were likely due to initial condition errors (McMurdie and Mass, 2004)

Some Unanswered Questions What are the origins of these short-term forecast errors – initial condition/data assimilation errors, model errors? Are there particular flow patterns (or regimes) where short-term (or longer term) forecasts are less accurate (e.g., E-T transitions)? How do model sensitivity structures compare for major forecast failure cases? How do they project on obvious initialization problems? How do adjoint-based and ensemble-based sensitivities compare for such cases?

Unanswered Questions continued What are the downstream implications for medium to long- range forecasts when initial condition errors are large over the Pacific? To what degree are downstream errors mitigated by greater data density over North America?

Implications for THORPEX Major forecast failures still occur, even at the short- time ranges. So there is still work to be done! It is important to monitor the quality of model initializations and forecasts to know how well we are doing and where the failures are. Both statistical and feature-based approaches are needed to gain a full understanding of model failures. Case studies can provide important insights into forecast failures

The END

Recent trends in forecast accuracy From Simmons and Hollingsworth (2002) Hemispheric r.m.s. error of SLP Increased skill of 3-5 Day forecast skill of SLP (and 500 hghts) especially last 10 years. Unable to discern forecast skill of storm systems in particular locations from these statistics

Brief Outline Show statistics of short term errors along West Coast Highlight several examples of major forecast failures Briefly discuss the effect of uncertainty over the Pacific on longer term forecasts

Adjoint Sensitivity wrt 850 mb Temperature Area of forecast error projected onto sensitivities Courtesy of Brian Ancell

Nov

6-hr forecast Eta

6-hr Forecast GFS