Meteorological Development Lab (MDL) FY15-16 Overview + Future (Direction and Challenges) Dr. Mike Farrar Director, Meteorological Development Lab (MDL) NCEP Production Suite Review Dec 7-9, 2015
MDL FY15 Highlights National Blend of Models (NBM): Fielded first experimental prototype Verification: Model Blend Viewer v1 implemented; prototype Aviation Forecast Verification Tool (AFVT) evaluated at AWT summer experiment MOS: Refreshed MOS station guidance (GFS) and EKDMOS (GEFS); implemented new ECMWF station MOS LAMP: Improved Gridded LAMP CIG, Vis, Temp, Dewpt; Prototypes for (a) improved Convection and Lightning and (b) Visibility using HRRR NDFD: Implemented NDFD map viewer into NIDS; made 2.5-km NDFD available via SOAP/XML Impacts Catalog: Migration of Iris/iNWS/Impacts Catalog prototype to IDP Storm Surge: Implemented P-Surge 2.5 and ETSS 2.0/2.1 National Smart Inits: Initiated with 4 CONUS regions
MDL Plans for FY16 NBM: Implement phase 1 for days 3-8 over the CONUS (Q1); Implement phase 2 with added elements + add OCONUS domains (Q4) Verification: Add new elements + OCONUS domains for Map Viewer MOS: Refresh MOS in concert w/EMC model upgrades; begin modernization - Weather Information Statistical Postprocessing System (WISPS) LAMP: Implement extended/improved LAMP convection and lightning grids, improved GLAMP CIG/VIS using HRRR, and LAMP wind gusts NDFD: Make hourly precip potential grids available via NDFD in CONUS; Increase frequency of day 1-3 NDFD issuances; refresh NDFD Central Servers Impacts Catalog: Implement Iris/iNWS/Impacts Catalog on IDP; conduct OT&E VLab: Migrated to IDP; CR SOO Communities; NOAA R&D Project DB National Smart Inits: Implement across 4 CONUS regions (on track!) FACETS Prob. Hazards Info (PHI): Dev/test AWIPS Hazard Services PHI prototype Rip Current: Perform experimental validation of forecast model at pilot WFOs MDLNet: Re-host MDLNet capabilities and webpages to standard systems
MDL Key Areas for FY17 Continued improvement in all capability areas; examples include: –NBM: Implement phase 3 for short range (days 0-2) using regional models –Impacts Catalog: Complete operational version; roll out to core customers –LAMP: New capabilities to support Digital Aviation Services –VLab: Build towards Centralized Development and Testing Environment (CDTE) –NDFD: Better integration within IDP and AWIPS (with new AWIPS contractor) –Storm Surge: Continued development to support NOAA Storm Surge Roadmap –Rip Current: Using NWPS to drive new Rip Current guidance Shift resources to create new capabilities and expand R2O across NWS –Optimize resources to maintain legacy capability to refocus on new capabilities –Expand role in field-driven R2O (more projects like National Smart Inits!) –Build new capabilities for Probabilistic Hazards Information (i.e., FACETS)
MDL’s biggest challenges related to our Production Suite Taking on new high-priority work while maintaining legacy products Maintaining an extensive operational product suite (GFS MOS, NAM MOS, LAMP, EKDMOS, etc.) while building new capabilities NBM. –Keeping these products in sync with ongoing changes to underlying models Aging statistical software/data infrastructure Transitioning NDFD (from AWIPS NCF and NIDS) to IDP Finding and optimizing data storage to maintain hourly 2.5km forecasts and analyses for verification queries Inconsistent IDP Dev and Ops environments can impact implementations –NCO helped us overcome with recent Impacts Catalog implementation Synchronizing NCEP production implementations with AWIPS
How does current production suite help meet challenges? For the most part EMC’s model products adequately fulfill MDL needs to generate value-added products to field users Related issues: Recent EMC commitments to retrospective runs are very welcomed –Once incorporated they will support development of value-added products based on model climatologies (akin to ECMWF Extreme Forecast Index) Some production suite implementation practices add to our challenges –Requirement for field evaluation needs to be focused on those implementations where field review adds value (minimize field efforts for minor/maint. upgrades). Consider alternate forms of feedback; broaden pool of field evaluators –IDP onboarding bottleneck (esp. in 2016 with focus on NWSTG rearchitecture) –2+ years of retrospective runs required for training new MOS equations for model upgrades; NCEP trying to build into schedule (work in progress)
Is current guidance too much, too little, or about right? From MDL perspective, guidance may be too much –Every model requires separate post-processing functions (e.g., MOS) –Do we need both NAEFS and National Blend of Models (NBM)? Related issues: Can NBM alleviate need to ship multiple model guidance data to WFOs? –Can/should NCEP Service Centers add value to automated NBM? Impacts Catalog: high growth potential for users could overtax NCEP systems and manual oversight of user access + impacts thresholds –Potential user demand for higher-res NDFD (greater demand on IDP & WFOs?)
What do we need from NCEP models/ products to meet challenges? Short-term (1-2 years) –Need longer-range HRRR forecasts (~36 hours?) to improve CIG/VIS for digital aviation services (~3-hourly; hourly not required) –Need stable, fully operational NWPS to support new Rip Current guidance –Alleviate/mitigate IDP onboarding bottleneck Longer term –Need follow-through on plans for retrospective model runs Need dedicated resources for retrospectives in HPC allocation for WCOSS –May need real-time access to historical climate data (currently on NCEI/NCDC) and all retrospective data to produce new guidance (akin to ECMWF’s EFI) –Need integrated NWS overall architecture (rather than separate HPC, AWIPS, and dissemination architectures and system boundaries)