Probabilistic weather hazard forecast guidance for transoceanic flights based on merged global ensemble forecasts Matthias Steiner National Center for Atmospheric Research, Boulder, Colorado, USA msteiner@ucar.edu With contributions from Ken Stone, & James Pinto, National Center for Atmospheric Research, Boulder, CO Matt Strahan, NWS Aviation Weather Center, Kansas City, MO Randy Bass, FAA Aviation Weather Division, Washington, DC WMO Aeronautical Meteorology Scientific Conference 6 10 November 2017 in Toulouse, France 1
Motivation Background - transoceanic flights require long weather outlooks - current weather guidance products do not satisfy ICAO/WMO ASBU weather requirements to support aviation in future ASBU weather roadmap - Block 1 (2018 2023) specifies calibrated probabilistic weather guidance for convective storms, turbulence, & icing hazards - grids with higher granularity in space & time Numerical weather prediction - ensemble approaches embraced by modeling community to capture forecast uncertainty & probabilistic prediction Goal - globally harmonized weather hazard guidance with two-day outlook in support of WAFS - fusion of multiple ensemble weather forecasts Longest Distance Flight: Qatar ~17h 30m from Auckland to Hamad 2
Guiding principles Today s fusion approach Harmonization Methodology - modular & transparent approach applicable to all weather hazards (convective storms, turbulence, icing) - accommodate changes in underlying ensemble forecast configuration (e.g., members) - scalable to allow for inclusion of additional weather prediction centers - calibration requires standardized hazard definition & truth (observations) - simple merger of WAFC London & Washington products New fusion concept - globally harmonized probabilistic prediction (with options) Additional NWP Centers Option 1 Option 2 NWP Center Super-Ensemble NWP Model Output Wx Hazard Diagnosis Post-Processing Forecast Product 3
Weather guidance - likelihood of thunderstorm occurrence - likelihood of convective cloud tops exceeding 30, 35 & 40 kft - forecast lead times from 12 48 hours in 6 hour increments - refresh 00, 06, 12, 18 UTC - 0E 360E, 60S 60N; 1 degree resolution - output in Grib2 format Real-time Prototype Methodology - GEFS & CMCE global ensemble combination - assessing ensemble members jointly exceeding specified thresholds for fields of interest (APCP, CAPE; and OLR) - model-specific thresholds chosen based on local, historical comparisons with CMORPH & CTH - adaptive scale, regional bias correction, seasonally adjusting - internal label Ensemble Prediction of Oceanic Convective Hazards (EPOCH) version 3.0 4
Sensitivity Analyses Data & domains - exploring combinations of five global ensemble forecasts - sample seasons: March, April, May (MAM) & June, July, August (JJA) of 2013 - sample domains as shown in graphic Assessment - focus on significant precipitation - comparison to CMORPH precipitation - relative performance (BSS) in comparison to calibrated GEFS Country (Model) European Centre for Medium-Range Weather Forecasts (ECMF) United Kingdom Met Office (UKMO) Meteorological Service of Canada (CMCE) United States National Centers for Environmental Prediction (GEFS) China Meteorological Administration (BABJ) Members Runs [UTC] 51 00, 12 25 00, 12 21 00, 12 21 00, 12 15 00, 12 N 1 BrierScore = å(p N i= 1 i - o ) i 2 EPac Car NAtl WPac 1 K K 2 1 = å n ( p o ) n ( o o) 2 k k - k - å k k - + o(1 - o) N N k = 1 k = 1 Reliability Resolution Uncertainty SA InOc BSS = BS - BS BS REF REF Map from: satsig.net 5
Brier Score Resolution Brier Scores for 24-hour forecast in comparison to CMORPH 2mm APCP6hr Pacific MAM All models combined Fusion impacts Calibration impacts per model Brier Score Reliability 6
BSS relative performance to GEFS (MAM 2013) Caribbean Best two models (CMCE + ECMWF) provide very good performance; nearly as good as best three, four, and five combined. South America Indian Ocean North American ensemble (freely available) tends to have higher skill than best single ensemble model Pacific North Atlantic 7
Key Points Numerical weather prediction - using ensembles to capture forecast uncertainty & create probabilistic predictions ICAO/WMO ASBU weather roadmap - targeting gridded, probabilistic weather hazard guidance for Block 1 (2018 2023) Methodology for global harmonization - harvesting global ensemble forecasts from multiple prediction centers to generate globally harmonized, gridded, probabilistic weather hazard guidance - methodology applicable for all weather hazards - requires agreed upon hazard definition & truth field > relevant observations for convective storms, turbulence, icing, etc.? Exercising prototype - combining ensemble forecasts from the United States (GEFS) & Canada (CMCE) for WAFC Washington implementation of convective storm hazards > providing real-time feed to NWS Aviation Weather Center for user evaluation - WAFC London exploring MetOffice & ECMWF combination Sensitivity analyses - exploring various calibration & fusion options, & tradeoffs among them - exploring benefits of including ensembles from several prediction centers 8
Thank you This research is in response to requirements and funding by the Federal Aviation Administration (FAA). The views expressed are those of the authors and do not necessarily represent the official policy or position of the FAA. 9