|
National Weather Service |
Apply numerical model guidance to produce more accurate forecast products.
To produce more accurate forecast products, forecasters must be able to apply numerical guidance efficiently and effectively, particularly under operational workload and time constraints. Underlying this is critical thinking, including knowing the right questions to ask and how to figure out the answers or recognize the extent to which the answers cannot be determined using all available information. Forecasters need to utilize observations, such as with techniques developed in the IST PDS, in assessing the past and current states of the atmosphere for comparison with the model guidance and in developing a framework for considering future evolution in the forecast period. Forecasters need to be aware of model characteristics and how they result in different analysis and forecast errors under different forecast scenarios. Critical thinking needs to be applied to all aspects of the forecast process.
1. Use observations to assess model initial conditions and evaluate previous model runs.
2. Understand how the process of data assimilation impacts the extent to which different types of data are incorporated into the initial analysis, the scales at which new data are represented in the analysis, and the shortcomings this can leave in the analysis.
3. Apply knowledge of model physics to recognize probable errors in the model solutions.
4. Integrate observed environmental conditions, model characteristics, and known limitations of NWP forecasts in the human forecast process.
5. Recognize the characteristics and associated forecast situations of particular types of spurious model forecast features.
6. Evaluate the range of possible forecast outcomes and degree of forecast uncertainty.
7. Interpret changes in model behavior after a new model implementation.
8. Interpret appropriately new forms of model guidance, including ensemble fields.
9. Recognize the limitations of various forms of post-processed model output and utilize that output accordingly.
10. Utilize all of the above skills and abilities to assess which model(s), if any, have a reasonable grasp on different aspects of the forecast and what adjustments are needed.
Instructional components for PCU 3 will include a series of Web-based cased studies and teletraining sessions developed to address a variety of operational skills. Each case may combine several of the items listed above. These items are broken into more specific detail or examples below:
A.: Using observations to evaluate model analyses and forecasts
A.1 Scale-appropriate comparison between remote-sensing observations and model analyses/forecasts
A.2 Scale-appropriate comparison between in-situ (surface, radiosonde, aircraft, etc) observations and model analyses/forecasts
B.: Recognizing the parts of a model solution likely to be accurate and the parts likely to be inaccurate (e.g. due to physics limitations, data assimilation limitations, resolution, model errors in previous similar forecast scenarios, etc.).
B.1 Not blindly accept or reject the entire model solution
B.2 Utilize observations to identify misrepresented features in the initial condition and in the short-range model forecast
B.3 Utilize observations to identify existing features and conditions likely to give rise to features that the model is incapable of including
B.4 Adapt the forecast to include features that the model is incapable of including or making a decent prediction of
C.: Assessing the range of possible forecast outcomes and degree of forecast uncertainty through examination of all situation-appropriate data and products, including ensemble output. Generating forecasts reflecting the certainty or range of possibilities.
C.1 Apply probabilistic thinking throughout the forecast process, from consideration of possible forecast scenarios to the issuance of products
C.2 Account for the predictability of forecast features in different regimes
C.3 Perform intermodel and run-to-run comparisons
C.4 Appropriately interpret a variety of ensemble products
D.: Recognizing the characteristics and associated forecast situations of particular types of spurious model forecast features.
D.1 Identify such features in the forecast
D.2 Ignore these features in the forecast
D.3 Assess the impact of these features on the model forecast in surrounding regions
E.: Interpreting changes in model behavior after a new model implementation.
E.1 Adapt forecast to new model characteristics
E.2 Utilize (including proper interpretation of) new types of NWP products
F.: Recognizing the limitations of various forms of post-processed model output and utilize that output accordingly.
F.1 Account for spatial and temporal resolution limitations of AWIPS grids
F.2 Utilize BUFR data and sources for information not available through AWIPS
F.3 Account for model limitations - just because a 12-km AWIPS QPF grid is available doesn't mean that features of 12-km scale can actually be predicted
F.4 Distinguish between derived output fields (visibility, 2-meter temperature, diagnostic precipitation type, etc) and what is directly predicted inside the model (skin temperature, precipitation type from the microphysics routine, mean winds in a model 3-d grid box, etc.)
F.5 Do not use MOS in extreme situations and situations for which model performance is erratic
F.6 Account for the limitations of MOS, especially when there has been a model change, and also for situations poorly sampled during the MOS development period
Components Currently Available
3.1 Application
of NWP Concepts: (Web) This growing collection of cases
is being created to demonstrate to field forecasters how and when to
use, not use, or modify NWP forecasts in the forecast process in various
scenarios.
3.2 Using AWIPS to Evaluate Model Initializations: (Teletraining, Archived) The material in this teletraining session is designed to increase your understanding of model initializations and how observations can be used to improve model evaluation/forecasting skills. The characteristics of the model initializations, model forecasts, and satellite (and other observations) are examined.
Training Measures
To be determined.
Comments
Send comments to Bill
Bua or Stephen Jascourt.
Return to Numerical Weather Prediction PDS page