Weather Service aims to improve winter forecasts
Despite major advances in scientific understanding and the development of computer-based weather models, there remains much room for improvement. Anyone who has followed this season's winter storm predictions has, no doubt, observed this. While there never will be perfect forecasts due to chaos (i.e. the butterfly effect), there is considerable opportunity for more accurate predictions and better estimates of associated levels of confidence in hazardous winter weather systems.
To address these issues, the National Center for Environmental Prediction's (NCEP) Hydrometeorological Prediction Center (HPC) - part of the National Weather Service (NWS) - recently embarked on what it refers to as the Winter Weather Experiment (WWE). The WWE, which will extend through February, is designed to identify promising techniques and tools most useful in addressing three especially challenging problems in Day 1 and 2 predictions:
1) Precipitation type and amount
2) Mesoscale (i.e. small scale) snowbands
3) Dealing with the ever present degrees of uncertainty in forecasts
Keep reading for details on this experiment...
The common challenge of items one and two is resolving the smaller-scale details ("mesoscale", horizontal dimensions from about 10 to several hundred miles) embedded in the larger-scale ("synoptic scale", extending from hundreds to around 3, 000 miles in horizontal extent) storm system which you'd typically see in routine analysis of surface and upper air weather maps.
As readily apparent from the storm earlier this week, the transition zone from snow to freezing rain can be very narrow and is dependent upon details in the vertical profile of temperature and moisture. Mesoscale snowbands can be as narrow as 5-10 miles in width and up to a few hundred miles in length. They reflect instabilities in the airflow over the sloping surfaces ("slantwise convection") of frontal systems aloft tied to the detailed structure of mesoscale disturbances. As in the recent experiences of the blizzards affecting the northeast (but not D.C., ugh!), their occurrence can significantly enhance snowfall rates and areal distribution of total amounts.
Enhancements in the predictability of precipitation type/amount and snowbands, of course, require accurate forecasts in the timing, track and intensity of the developing synoptic-scale (larger scale) storm. And, not incidentally, this larger-scale background can be affected by complex interactions with the mesoscale features in a feedback loop which tends to decrease the overall predictability on all scales. Did I say we're dealing with a formidable challenge?
A principal approach in the WWE is to further explore the use and value of the current modeling systems and procedures along with experimental high-resolution (4 kilometer) models and ensemble systems. Crucially important is forecaster interpretation of data sets and products derived from models in producing new products and displays.
Emphasis is upon quantifying and characterizing information on case dependent levels of uncertainty/confidence, which addresses a publically stated high-priority strategic goal of NOAA/NWS and NCEP: emphasizing improved decision support services.
The WWE is being executed within the context of the HPC Hydrometeorological Test Bed (HMT-HPC). A key aspect of the HMT is facilitating interactions among researchers and operational forecasters, including rotation of experts from within NCEP and NWS Weather Forecast Offices (WFOs).
Experimental WWE forecasts are generated once per day for Day 1 (12-36 hours) and Day 2 (36-60 hours) from 00Z (7 p.m.) and focus on a single storm system selected in advance. These forecasts are produced essentially independent of HPC's, so comparisons can be made between the corresponding operational and experimental predictions.
The potential benefits that might emerge from the WWE include determining the value and operational viability of experimental guidance that could be integrated into HPC operational procedures. Such complements to current operations might include more detailed and generally higher confident forecasts of hazardous winter weather.
But, success is certainly not guaranteed. Still, it is expected that knowledge and understanding of model capabilities/limitations and lessons learned from forecaster experience will lay a firm foundation for the design and conduct of prospective WWEs in years ahead. I'll be doing a follow-up post when an evaluation of the current WWE becomes available.
In view of the high impact of hazardous winter weather, NCEP has reallocated resources to spin up the Winter Weather Experiment this year and will continue to find ways to continue this work even in the face of the challenging budget environment.
NCEP does face limits in its capability to implement the new higher resolution models and ensembles being assessed in the WWE on the current operational computer. Current plans for acquiring the next generation computer are expected to enable running the higher-resolution models operationally by 2015. In the meantime, experimental runs will continue as NCEP partners with other research organizations to continue evaluating higher resolution modeling systems in prediction of extreme and high impact weather of all types.