May, 2002Numerical Weather Prediction 1 Robert R. Gotwals, Jr. (Bob2) Computational Science Educator The Shodor Education Foundation, Inc.
May, 2002Numerical Weather Prediction 2 Session Goals Describe application, algorithm, and architecture Describe and demonstrate the various NWP programs and codes Describe appropriate and authentic classroom activities using online NWP tools
May, 2002Numerical Weather Prediction 3 Application - First Principles Definition: The use of computer models to predict the future state of the atmosphere given observations and equations that describe relevant physical processes Some givens: Weather prediction is really hard Synoptic scale calculations, but local influences Equations are nonlinear
May, 2002Numerical Weather Prediction 4 Application - Results Example plots Temperature Dewpoint Mean sea level pressures (MSLP) Winds, surface and aloft Cloud cover Precipitation and types Severe weather indices CAPE Helicity
May, 2002Numerical Weather Prediction 5 Algorithm - NWP Desks Desk seat 1: calculates east-west component of the wind Desk seat 2: calculates north-south component of the wind Desk seat 3: keeps track of the air entering or leaving the box. If more is coming in than going out, decides how much air rises or sinks Desk seat 4: calculates the effects of adding or taking away heat Desk seat 5: keeps track of water in all forms and how much is changing to or from vapor, liquid, or ice Desk seat 6: calculates the air temperature, pressure, and density
May, 2002Numerical Weather Prediction 6 Architecture - Platforms NWP requires significant computing power True supercomputing required –Gigaflops - billions of calculations (floating point operations) per second –Teraflop - trillions of calculations per second Data storage –NCAR - late 2000, 200 terabytes of data stored NCAR machine –11th most powerful supercomputing in the world –IBM SP Power 3 –1260 CPUs (processors) –Peak capabilities: 1890 Gigaflops
May, 2002Numerical Weather Prediction 7 Architecture - Codes General categories –By resolution –By scale Global (northern hemisphere) National relocatable –By outlook (time-based) Well-known codes –Nested Grid Model (NGM) –ETA –Aviation Model (AVN) –Rapid Update Cycle (RUC) –Medium Range Forecast (MRF) –Mesoscale Model 5 (MM5)
May, 2002Numerical Weather Prediction 8 Nested Grid Model (NGM) National model Short-range model (+48 hours), every 6 hour forecasts Forecast output –Temperature –Precipitation –Upper and lower trough positioning –Surface highs and lows Grid size: 80 km Operational status: being phased out
May, 2002Numerical Weather Prediction 9 ETA Name comes from eta coordinate system Short-range model Four runs daily: 0000Z, 0600Z, 1200Z, 1800Z 32 km horizontal domain, with 45 vertical layers Significantly outperforms other models in precipitation predictions
May, 2002Numerical Weather Prediction 10 Rapid Update Cycle Regional model Short-term forecasts –Up to 12 hours Focuses on mesoscale weather features 25 vertical layers, 40 km horizontal resolution New experimental version: MAPS RUC/MAPS generate significant amount of data
May, 2002Numerical Weather Prediction 11 Medium Range Forecast (MRF) Model Global model Medium to long-range predictions: 60 to 240 hours Resolution: 150 km Other global models –UKMET –ECMWF –Global Ocean Model
May, 2002Numerical Weather Prediction 12 Aviation Model Generates aviation- focused data 42 vertical layers, 100 km horizontal resolution Advantage: medium- range forecasting (up to 72 hours) One of the oldest operational models Data results available mostly in MOS (model output statistics) format
May, 2002Numerical Weather Prediction 13 MM5 Fifth generation mesoscale NWP Study types –hurricanes –cyclones –monsoons –fronts (formation, interactions) –land-sea breeze meteorology –urban heat islands –mountain-valley circulations
May, 2002Numerical Weather Prediction 14 Sample Prediction Question: assuming precipitation, what will it be? Tools: –Atmospheric sounding (weather balloon data)Atmospheric sounding Shows temperature and dewpoint temperature from surface to upper atmosphere –Flowchart: precipitation type decision tree Analysis/solution shown on next slide
May, 2002Numerical Weather Prediction 15 Sample Prediction - Solution
May, 2002Numerical Weather Prediction 16 Classroom Integration - Forecasting Rules of thumb Will it be cloudy or clear? –On the 700-mb forecast chart, the 70% relative humidity line usual encloses areas that are likely to have clouds Will it rain? –On the 700-mb forecast chart, the 90% relative humidities line often encloses areas where precipitation is likely. Will it rain or snow? –On the 850-mb forecast chart, snow is likely north of the -5 C (23 F) isotherm, rain to the sou th
May, 2002Numerical Weather Prediction 17 Classroom Integration - Weather observations Correlating low-tech weather observations –Use instant weather prediction chart –Shows various weather 24 hours out based on easily observable meteorological phenomenon –Can correlate this with model data predict.html
May, 2002Numerical Weather Prediction 18 Classroom Integration Good starting place: meteograms –Relatively easy to interpret –Contain a lot of data –Typically project out 24 to 72 hours –Relatively good resolution (normally 22 km) –Available from a variety of models
May, 2002Numerical Weather Prediction 19 Classroom Integration Harder: atmospheric soundings graphs Substantial amounts of information Graphical and text-based information –Graphical: temperature, dewpoint temperatures, wind speeds and directions –Text: key meteorological indices
May, 2002Numerical Weather Prediction 20 Questions? Chat Sessions –Monday, May 13 3:30- 4:30 PM and 6:00-7:00 PM –Wednesday, May 15 3:30- 4:30 PM –Monday, May 20 6:00- 7:00 PM –Thursday, May 23 3:30- 4:30 PM and 6:00-7:00 PM