By Tracy Hansen
Many questions arise when trying to understand the forecast methodology, one charter of the Rapid Prototype Project (RPP). The issues center around such questions as:
There is no one process – the steps to build a gridded forecast will vary with each site, forecaster, and situation. Thus our goal in understanding forecast methodology has been to describe specific representative scenarios. This idea has been incorporated into our training sessions, and equally as important, these scenarios have helped us, as the developers, to build software that better meets our customers' needs.
Here we discuss two different forecast scenario formats provided by the Unified Modeling Language (UML) Activity Diagrams and Job Sheets, which have been very useful in training forecasters how to use the GFESuite.
Activity Diagrams – In some cases, we have used a visual format, UML Activity Diagrams, to express forecast scenarios. The diagram in Figure 1 shows a summer convective scenario that describes the relative position of a front of air masses. Figure 2 is an "Activity Diagram" describing how the forecaster creates grids depicting the probability of precipitation (PoP) and weather (Wx): rain, snow, etc. The left-hand side of the diagram shows the step the forecaster is taking and the background meteorology. The arrows and right-hand columns describe how that step can be accomplished within the GFE. Each column corresponds to a type of GFE functionality, sometimes a particular menu item.
The diagram in Figure 3 shows a Winter Scenario in which the forecaster is creating the Quantitative Precipitation Forecast (QPF), Wind, Freezing Level (FzLevel), and Snow Amount (SnowAmt) grids, which are interdependent. It illustrates how to operate the system within the context of a meaningful meteorological task; that is, to produce a forecast amount of snow. During forecaster training on the GFE, these diagrams helped them with the concept of grid-based forecasting and understanding how to solve particular problems that arise on the job.
Figure 4 shows a more general process: higher level steps that forecasters might follow on their shift to produce the day's gridded products. As a result of looking at this view of forecaster methodology, we rearranged the GFE menu bar to make it more intuitive according to the process steps. This diagram was also the focus for discussing how the GFE and grid-based forecasting will operate in the field. It provided a common language to the developers and forecasters from which sprang new ideas and suggestions.
Job Sheets – Innovative staff at the NWS Forecast Office in Tucson, Arizona, developed "Job Sheets" for GFE training. Like the Activity Diagrams, these sheets (example in Figure 5) lead the forecaster through the steps to perform specific tasks using the GFE. They have been adopted by several other forecast offices.
Benefits of Focusing on Methodology
Many benefits are derived from the use of Activity Diagrams and Job Sheets. They provide a common language between client and developer. The Activity Diagrams are a visual representation, and as such are a great focus for communication. They complement traditional requirements and use-case diagrams with procedural, concrete examples; thus the resulting requirements better meet client needs.
Both Activity Diagrams and Job Sheets are excellent training tools because they illustrate not only how the system operates, but also how to produce useful products with the system. They give the learner the big picture and provide incentive for using the tool. Another advantage is that they constitute a collection of functional test cases for the system. As the software is developed and modified, it can be verified against these standard scenarios.
We have built a Methodology Page at the Rapid Prototype Project Website, http://www-md.fsl.noaa.gov/eft/rpp, as a repository for these ideas. Ongoing discussions will continue between forecasters and developers so that the methodology of grid-based forecasting and the transition from traditional practices can progress efficiently with minimal disruption to NWS operations.
(Tracy Hansen can be reached at firstname.lastname@example.org.)