Air Force Weather Ensembles

Size: px

Start display at page:

Download "Air Force Weather Ensembles"

Meryl Sharp
5 years ago
Views:

1 16 th Weather Squadron Air Force Weather Ensembles Evan Kuchera Fine Scale Models and Ensemble 16WS/WXN

2 Overview n Air Force Weather Operational Ensembles n Probabilistic inline diagnostics n Interactive data interrogation n n n Operational products, configuration information, case studies, etc Password protected contact me for access info 2

3 AFW Operational Ensembles n Goals for operational ensembles n Reliable, sharp prediction of mission-impact phenomena n Change decisions! n Tailored products: both quick-look AND detailed n Timely! n Global Ensemble Prediction Suite (GEPS) n 62 members from NCEP, CMC, FNMOC n Mesoscale Ensemble Prediction Suite (MEPS) n 10 members of WRF-ARW with diverse initial conditions and physics n 144 hour global at ~20 km, 72 hour regional at 4 km 3

4 AFW Operational Ensembles n Physics configurations Mem Atmos IC/LBC Land IC Snow/ SST IC LU Surface LEVS LEV2 SW-Rad LW-Rad PBL DSR Microphysics Hail CCN Cumulus (20 km only) 1 UM LIS UM USGS NOAH Dudhia RRTM ACM2 GNX WDM6 1 5E+08 BMJ 2 GFS LIS GFS USGS NOAH Dudhia RRTM BouLac DRI Morrison 1 1E+08 Tiedtke 3 GEM LIS GEM USGS NOAH Goddard Goddard YSU GNX WDM6 0 1E+09 New SAS 4 GEM UM GEM USGS NOAH Goddard Goddard BouLac DRI Morrison 1 1E+09 BMJ 5 UM UM UM USGS NOAH CAM CAM YSU GNX Thompson N/A N/A Tiedtke 6 GFS LIS GFS USGS PX Dudhia RRTM ACM2 DRI WDM6 1 1E+08 Tiedtke 7 GEM UM GEM USGS PX Dudhia RRTM BouLac GNX Thompson N/A N/A New SAS 8 GFS LIS GFS USGS PX CAM CAM ACM2 DRI Morrison 0 1E+08 BMJ 9 UM UM UM USGS PX CAM CAM YSU GNX WDM6 0 5E+08 BMJ 10 GFS UM GFS USGS PX Goddard Goddard ACM2 DRI Thompson N/A N/A New SAS n No data assimilation, most recently available global model output is interpolated to the domain n vertical levels does not seem to degrade quality, saves on cost (Aligo, et. al., 2008) 4

5 4 km MEPS domains Green static; Blue relocatable (positions subject to change) Each domain runs to 72 hours once per day (CONUS and East Asia 2X/day) *(Current as of June 2014) 5

6 AFW Operational Ensembles n Product Examples 6

7 65 knots knots

8 Probabilistic Diagnostics n Purpose of diagnostics: Access data from the model for algorithmic post-processing while it is available in memory n Reduces I/O load n Enables tracking of maxima/minima n Enables variable temporal output cadence n Given model code parallelization, is generally cheap n Examples of variables n Simulated clouds/satellite and radar n Clear and rime aircraft icing n Precipitation type/snowfall accumulation n Surface visibility (fog/precip/dust) n Severe weather (lightning/hail/tornado/winds) n All diagnostics within community WRF framework in n Implementing in AFWA operational MEPS in July

9 Probabilistic Diagnostics n Algorithms n Many variables of importance to users are sub-grid n Uncertainties in occurrence/frequency require probabilistic algorithms to ensure reliable prediction n Limited ensemble members n Only having 10 members means the PDF is poorly sampled even if the ensemble is perfect n Weibull distribution n Three variables to describe: n Shape, scale, shift n Flexibility to make different distributions (Gaussian-like, exponential, etc) n Can put a curve on grid-scale data (i.e. precipitation) and on sub-grid data (i.e. lightning frequency) 9

10 Probabilistic Diagnostics Weibull for wind gusts: Shift: sustained wind speed Scale: (sustained wind speed)^0.75 Shape: 3 (gaussian-like) No weibull modifications, uses uniform ranks to determine thresholds 10

11 Future n Rolling ensemble n Frequent (~2 hrly) runs with GSI n Time-lag for probabilistic data n Global WRF for MEPS (~20 km) n Replace global-coverage MEPS domains n Goal: global coverage of tailored inline products n Interactive data exploitation n AF missions and weapons systems have unique criteria n What is the chance I can do X today? n Must make statistical assumptions for multi-hour and joint probabilities (not trivial) 11

12 Interactive data 12

13 Summary n Air Force Weather is executing operational ensemble prediction systems n Timely, storm-scale products tailored to mission needs n Implementing probabilistic diagnostics n Within model code for efficiency, quality, flexibility n Improved characterization of certainty n Interactive data interrogation n Provide users the ability to tailor probabilistic data to the mission n Improved risk assessment and $$$ saved 13

Air Force Weather Ensembles

16 th Weather Squadron Fly - Fight - Win Air Force Weather Ensembles Evan Kuchera Fine Scale and Ensemble Models 16WS/WXN Background Meeting theme: Develop better, ensemble-derived, decision support products