Expansion of Climate Prediction Center Products Wanqiu Wang (CPC) Stephen Baxter (CPC) Rick Thoman (NWS) VAWS webinar, January 17, 2017
CPC Sea Ice Predictions Wanqiu Wang Thomas Collow Yanyun Liu Arun Kumar David DeWitt Climate Prediction Center (CPC) NOAA/NWS
Outline 1. Sea ice prediction projects 2. Current status 3. Ongoing development 4. Future work
1. Sea Ice Prediction Projects I. Goals Advance sea ice prediction capability with improved sea ice initialization and improved model configuration Provide improved seasonal and week 3-4 predictions to stakeholders II. Projects Seasonal prediction Forecast frequency Target Hindcast Monthly (March to October) 9 months Most recent 10 years Week 3-4 prediction Forecast frequency Target Hindcast Weekly 6 weeks Most recent 6 years
2. Current Status (1) Seasonal prediction Model: NCEP CFSv2 + physics modifications Initialization: NCEP CFSR and PIOMAS initial sea ice Forecasts : 2015, 2016, 2017 (March October) (Forecasts are sent to NWS Alaska Region) Hindcasts: 2005-2014, 2006-2015, 2007-2016 (2) Week 3-4 prediction Model: CFSv2 + physics modifications + MOM5 Initialization: CFSR and PIOMAS initial sea ice Hindcasts : 2012-2016 CFSv2: Climate Forecast System version 2 CFSR: Climate Forecast System Reanalysis PIOMAS: University of Washington Pan-arctic Ice/Ocean Modeling and Assimilation System
Strategic Product Development Leveraging Partnerships: Experimental Seasonal Arctic Sea Ice Outlooks CPC started producing Experimental Arctic Sea Ice Outlooks for the NWS Alaskan Region in March 2015. - Outlooks are issued once per month for March to October Ics and extend for 9 months. - 3 improvements over baseline CFSv2 system: Improved sea-ice initial conditions (PIOMAS from U. Washington) Modified atmospheric physics (stratus clouds) Modified ocean physics (heat flux constraint) Exp. CFSv2.
CPC Seasonal Sea Ice Forecast Products Monthly mean ice extent Total area of grid boxes where monthly mean sea ice concentration is greater than 15% Ensemble mean and individual members Monthly mean sea ice concentration Ensemble mean and spread Probability of monthly mean sea ice Concentration greater than 15% First ice melt day (IMD) and ice freeze day (IFD) Ensemble mean and spread
CPC Seasonal Sea Ice Forecast Products Monthly mean ice extent Monthly mean sea ice concentration Digital figure files and data files are provided Probability of monthly mean sea ice First ice melt day (IMD)
2. Current Status (1) Seasonal prediction Model: NCEP CFSv2 + physics modifications Initialization: NCEP CFSR and PIOMAS initial sea ice Forecasts : 2015, 2016, 2017 (March October) (Forecasts are sent to NWS Alaska Region) Hindcasts: 2005-2014, 2006-2015, 2007-2016 (2) Week 3-4 prediction Model: CFSv2 + physics modifications + MOM5 Initialization: CFSR and PIOMAS initial sea ice Hindcasts : 2012-2016
June 2012 SIC (Week 1) Comparisons NASA Team CFSv2 Operational CPC Experimental Too much sea ice in eastern Beaufort Sea and Kara Sea in CFSv2 during June 2012.
3. Ongoing Development 1) Development of a CPC Sea ice initialization system (CSIS) 2) Test week 3-4 prediction with CSIS 3) Test seasonal prediction with CSIS
Motivation for Developing a New Sea Ice Initialization Sea ice from NCEP operational CFSR is erroneous Use of PIOMAS causes delay in real time forecast
Sea Ice From NCEP Operational CFSR is Erroneous February-March average sea ice thickness (m) for 2004-2008 Sea ice extent difference CFSR - NASA Team
Development of a CPC Sea Ice Initialization System (CSIS) Model: MOM5 Atmospheric forcing: CFSR or MERRA-2 Variables to be assimilated: - SST: NCEI or OISST - Ice concentration: NASA Team, VIIRS - Ice thickness: ICESat, CrySat-2, SMOS, VIIRS, IceBridge
4. Future work 1) Evaluate CPC sea ice initialization (CSIS) 2) Perform and evaluate seasonal hindcasts using CSIS 3) Perform and evaluate week 3-4 hindcasts CSIS 4) Develop new real-time seasonal and week 3-4 forecasts
Development of CPC s Week 3-4 Forecast Products Background, Methods, and Verification Stephen Baxter (CPC)
Background CPC issues forecast products ranging from Week-2 to monthly and seasonal time scales. Given the various statistical and dynamical forecast developments over many years, we embarked upon the effort to produce Week 3-4 forecasts. These Week 3-4 products fill in a gap between short-term climate prediction that heavily utilizes initial conditions and seasonal climate prediction that emphasizes low-frequency boundary conditions (e.g. sea surface temperatures, sea ice, soil moisture). The temperature and precipitation forecasts launched in September 2015. The temperature forecast is now considered operational.
Methods - Statistical Prediction skill at the Week 3-4 time scale comes from multiple phenomena with various characteristic frequencies. Our flagship statistical model utilizes multiple regression with three predictors: MJO indices (RMM1 and RMM2 from Wheeler and Hendon) Niño 3.4 index Linear trend over the training period
Latest MLR Guidance MJO ENSO TREND ANOM PROB
MLR Forecast Skill Past Year OFCL CPC 21.2 MLR 23.5 CFS 13.8 ECMWF 18.2 JMA 22.4
Methods - Dynamical Equal-weighted blend of CFS, ECMWF, and JMA dynamical models. Correlation-weighted probability forecast from CFS, ECMWF, and JMA. The inputs are calibrated probabilities using ensemble regression.
Model Skill Past Year OFCL CPC 21.2 MLR 23.5 CFS 13.8 ECMWF 18.2 JMA 22.4
Official Forecast Verification
Future Work Objective consolidation of the statistical and dynamical guidance is desirable. Hybrid dynamical-statistical guidance that utilizes forecast MJO index values could improve over current MLR forecast skill. Statistical guidance that captures prediction skill afforded by stratosphere-troposphere interaction is in the works.
Week Two in Alaska: Tools for Impact Based Decision Support Rick Thoman NWS Alaska Region
Alaska Outlook Needs in Week 2 Marine Based Activity: Winds, Waves and Ice Industry: Extraction, Fishing Shipping: Trans-Pacific and regional Subsistence: Ice and Water based Tourism Recreation Storm and Extreme (or seasonally expected) Weather Community Preparation Flooding Water Supply Infrastructure Supply Replenishment Planning and Resource Allocation Alaska DHS (Vulnerable people evac, RiverWatch) DOT & PF (e.g. Dalton Highway aufeis flooding) Alaska Fire Service
Week 2 Outlook Philosophy Week 2: Edging toward the weather/climate interface, especially NW North America/NE Asia Single deterministic models not appropriate Ensemble systems are first step dealing with initial condition uncertainty, e.g. NOAA s Global Ensemble Forecast System (GEFS) Multiple model ensembles are a second step dealing with initial condition uncertainty and model bias e.g. North American Ensemble Forecast System (GEFS plus Canadian GEM Ensemble System) Bias correction and calibration an additional step for single or (currently underdeveloped) As of today, Week 2 IDSS for Alaska is mostly about using available tools to subjectively evaluate meteorological patterns known to be conducive to high impact events
Week Two Toolkit As always, only some tools available for CONUS are useful for Alaska Time Aggregated: CPC Time and Spatial Aggregated: Cyclone Density (new version in the works) Time Specific Spatial EMC NAEFS Confidence EMC Relative Measure of Predictability OPC and WPC GEFS Percentages (Ensemble system percentages: NOT true probabilities) GEFS surface pressure centers (new version in the works) GEFS Cluster Analysis GEFS Atmospheric Rivers Point Specific NOMADS GEFS Percentages (NOT true probabilities) EKD MOS
CPC Probabilistic Temperature Outlook Traditional Hazards Product Temp, pcpn, wind, etc. Experimental Probabilistic Hazards
SUNY Stonybrook Cyclone Density
EMC Ensemble Situational Awareness
EMC Relative Measure of Predictability
OPC NAEFS Wind Threshold Frequency
OPC GEFS Wind Threshold Frequency
OPC GWES Sea Height Threshold Frequency
WPC GEFS MSLP Frequency
WPC GEFS Accumulated Precipitation Frequency
Individual GEFS Pressure Centers
GEFS Cluster Analysis
Note this is does not correspond to ANY model run MDL EKD MOS
MDL EKD MOS: Probabilistic Sensible Elements to 15 days
EMC Threshold Exceedance Tool
Links for Week 2 Products CPC Hazards: http://www.cpc.ncep.noaa.gov/products/predictions/threats/threats.php RMOP: http://www.emc.ncep.noaa.gov/gmb/yluo/html_pqpf/rmop.html Cyclone Density: http://smokey.somas.stonybrook.edu/cyclonetracks/alaska_tprob.html GEFS Ensemble Centers: https://www.tropicaltidbits.com/analysis/models/ WPC Ensemble: http://www.wpc.ncep.noaa.gov/exper/gefs/gefs.html Atmospheric Rivers: http://vortex.plymouth.edu/~j_cordeira/arportal/current/products.html OPC (Ensemble Probabilities in Data ): http://www.opc.ncep.noaa.gov/index.php GEFS Cluster Analysis: https://www.woweather.com/cgi-bin/expertcharts?lang=us&menu=0000000000&cont=usus EKD MOS: https://www.weather.gov/mdl/ekdmos_home EMC Point Probability Tool: https://nomads.ncdc.noaa.gov/ensprob/