Previous Chapter 3 The Forecast Database

3 The Forecast Database

Forecast fields will be stored for later retrieval by various modules, including the interactive editing tools, the product generator, and interpolation tools. During the early stages of development, this project will use an interim storage system to house the weather elements; later, they will be integrated with the AWIPS database.

The forecast database contains both general and service-specific weather elements. In addition to these forecaster-edited elements, some forecast-specific product parameters are needed by the text generators. Derivation of these product parameters from the weather elements is described in Chapter 7. All of these are part of the AWIPS database.

3.1 Database Structure

The physical structure of the forecast database will be defined by AWIPS; no details are included here. The logical structure of the database will incorporate a worksheet concept (Figure 4 on page 22). A worksheet consists of a matrix of weather elements versus time, and each element of the matrix is a two-dimensional forecast of a single element at a particular time covering the entire forecast area. Multiple matrix elements may be edited simultaneously by the forecaster. The forecast worksheet represents a complete forecast for the area and time covered.

The worksheet concept provides the forecaster great flexibility by allowing the forecast to be generated with input from a wide variety of data sources. Typically, forecasters will use several reference worksheets. One can be constructed, using predefined algorithms, from the output of each numerical model. Another will contain the forecast grids manually prepared at NMC. The forecaster will be able to refer to these fields and copy them as desired into the forecast worksheet. For example, a forecaster could generate two reference worksheets, each based on a different model, then choose the features that are handled best by each model and build a composite worksheet that utilizes each model's strengths. The graphical editors are used to modify these first-guess fields. Interpolation tools will be provided to fill in missing times so the forecaster is not required to edit every matrix element.

The final worksheet containing the desired forecast solution is saved in the database as the official forecast. All text forecasts are generated from this database.

3.2 Relationship to AWIPS Database

Since AWIPS is not yet available, an interim forecast database will be created so that work may begin immediately. This forecast database will be integrated with the AWIPS database when feasible. Since performance is critical, success in integrating the database into AWIPS will depend heavily on the performance of the AWIPS database.

3.3 Weather Elements

The weather elements to be stored in the forecast database will be derived from the contents of the AWIPS database. The AWIPS database will contain forecast fields from NMC (manually prepared) and from national- and local-scale models. It will also include surface observations, upper air observations, radar data, and satellite imagery. Methods using these data sources will be developed for initializing a predefined set of weather elements, which will be available for viewing and editing by the forecaster.

General weather elements (those applying to multiple services) will be created directly from guidance. After forecaster editing, other general elements may be derived. For example, maximum relative humidity, used in agricultural and fire weather forecasts, is derived from dew point and temperature. Service-specific elements will be derived from the general elements and, in many cases, other data.

The general/specific weather element concept is aimed at reducing forecaster workload, while allowing him or her to work in a logical progression through the forecast process. The design is a trade-off between conflicting requirements to automate an inherently hands-on process.

The weather element sets may undergo revision as work on this project progresses. The lists will be reviewed by NWS management and the AFPS Forecaster Working Group (AFWG)(1) to ensure that they are adequate for all products to be generated.

3.3.1 General Weather Elements

An initial set of general weather elements is defined in Table 1. This set contains elements that

are intuitively familiar to forecasters;
can be derived from information in the AWIPS database (observations, model fields, and central guidance and forecasts);
can be used to derive other, more specialized, weather elements; and

are required for standard NWS products.

Table 1 - General Weather Elements 
---------------------------------------------------------------------------------------------
Weather Element               Descriptors                   Data Type(a)    Spatial            
                                                                            Representation(b)  
Temperature                                                 Numerical       Continuous         
Dew Point                                                   Numerical       Continuous         
Wind                          Speed, Direction. Gust        Numerical       Continuous         


                             Amount, Type, Probability(c)  Categorical     Bounded Area       
Clouds                                                                                         
                              Base, Topc                    Numerical       Bounded Area       
Weather                       Type, Intensity, Coverage     Categorical     Bounded Area       
Sky Condition                                               Categorical     Bounded Area       
Probability of Precipitation                                Step-Numerical  Continuous         
QPF                                                         Step-Numerical  Continuous         
Visibility                                                  Step-Numerical  Bounded Area       
Snow Accumulation                                           Numerical       Continuous         
Freezing Level                                              Numerical       Continuous         
Inversion                     Type, Top, Base               Categorical     Bounded Area       
                                                                                               
---------------------------------------------------------------------------------------------

(a): Numerical refers to floating-point or integer numbers, categorical to one of a set of categories (e.g., CLR, SCT, BKN, OVC), and step-numerical to quanta (e.g., PoP is in 10% increments).
(b): Data representations are further described in Section 6.2.1.
(c): Type, probability, base, and top of clouds are needed only for aviation-oriented forecasts.

Note that we have excluded pressure from the list of general elements. Although pressure is a commonly viewed element (and will be available for reference), it is not used in forecast products. Pressure could be edited and used to derive winds, but wind forecasts are available both directly from models and from statistical guidance.

As discussed in Chapter 4, considerable QPF guidance will be provided to WFO forecasters. It would be desirable to relate QPF to precipitation intensity and duration (part of the "weather" weather element); this will be difficult, at best. Further, the relationship between QPF and probability of precipitation (PoP) is complex.

QPF grids will be sent from WFOs to RFCs with three probabilities for each accumulation period. Though these have yet to be determined, suppose that they are chosen as 20%, 50%, and 80%. Consider, for example, an area specified at 80% chance of 0.50 inch accumulation. The 50% map might show 0.75 in and the 20% map 1.00 in. In addition, the 50% map might show 0.25 in and the 20% map 0.10 in. That is, there is a 50% chance that between 0.25 in and 0.75 in will occur.

PoP, of course, is the probability of any measurable precipitation. How this relates to QPF probabilities, and how all of these various precipitation elements will work together, will require much study.

Derived General Weather Elements

Table 2 includes those general weather elements that are derived from the "basic" general elements. The derivation techniques will be transparent to the forecaster. Temperature and relative humidity extremes are straightforward, given a time series. Sky condition will be determined from cloud fields.

Table 2 - Derived General Weather Elements 
------------------------------------------------------
Weather Element            Data Type    Spatial         
                                        Representation  
Maximum Temperature        Numerical    Continuous      
Minimum Temperature        Numerical    Continuous      
Maximum Relative Humidity  Numerical    Continuous      
Minimum Relative Humidity  Numerical    Continuous      
Sky Condition              Categorical  Bounded Area    
                                                        
------------------------------------------------------

3.3.2 Service-Specific Weather Elements

The product generators require many specialized parameters that are not in the above lists of general weather elements. As with the derived elements listed above, these are derived from the general weather elements, plus whatever other information may be required from the AWIPS database. For example, the agriculture forecast requires hours of sunshine, which can be derived from cloud cover and sunrise/sunset information. These elements differ from the derived elements only in that they apply to just one service. Tables 3 - 7 list these elements by service. Aviation-specific weather elements are of two types, supporting terminal and route forecasts, described Tables 3 and 4.

To-be-developed high-quality derivation techniques will minimize forecasters' need to edit these fields.

Table 3 - Aviation-Specific Weather Elements -- Terminal Forecasts 
--------------------------------------------------------------------------
Weather Element(a)  Descriptors             Data Type       Temporal(b)     
                                                            Representation  
Wind                Speed, Direction, Gust  Numerical       Continuous      


                   Type, Amount            Categorical     Stepped(c)      
Cloud                                                                       
                    Base                    Numerical       Continuous      
Weather             Type, Intensity         Categorical     Stepped         
Visibility                                  Step-Numerical  Continuous      
                                                                            
--------------------------------------------------------------------------

(a): All apply to Prevailing, Occasional, and Chance conditions.
(b): Since terminals are point forecasts, there is no spatial representation to consider.
(c): "Stepped" refers to categorical changes with time.

Table 4 - Aviation-Specific Weather Elements -- Route Forecasts 
-------------------------------------------------------------------
Weather Element(a)  Descriptors      Data Type       Spatial         
                                                     Representation  


                   Base, Top        Numerical       Continuous      
Cloud                                                                
                    Amount           Categorical     Stepped(b)      
Weather             Type, Intensity  Categorical     Stepped         
Surface Visibility                   Step-Numerical  Continuous      
Hazards                              Categorical     Stepped         
-------------------------------------------------------------------

(a): All apply to Isolated, Local, Areas, and Widespread conditions.
(b): A route forecast is essentially a line; thus, "stepped" applies as above.

Table 5 - Fire-Weather-Specific Weather Elements 
------------------------------------------------------------------------------------
Weather Element           Descriptors                 Data Type       Spatial         
                                                                      Representation  


                         Transport Speed/Direction

  

               

               
Wind                      Free Air Speed/Direction    Numerical       Continuous      
10-Hour Fuel Moisture                                 Numerical       Continuous      
Precipitation Duration                                Step-Numerical  Continuous      
Haines Index                                          Step-Numerical  Continuous      
Mixing Depth                                          Numerical       Continuous      
Stability                                             Categorical     Bounded Area    
Lightning Activity Level                              Categorical     Bounded Area    
Chance of Wetting Rain                                Step-Numerical  Continuous      
                                                                                      
------------------------------------------------------------------------------------

Table 6 - Agriculture-Specific Weather Elements 
-------------------------------------------------
Weather Element    Data Type       Spatial         
                                   Representation  
Dew Intensity      Categorical     Bounded Area    
Dew Dry-off Time   Step-Numerical  Continuous      
Drying Conditions  Categorical     Bounded Area    
Hours of Sunshine  Numerical       Continuous      
Minimum Dew Point  Numerical       Continuous      
                                                   
-------------------------------------------------

Table 7 - Marine-Specific Weather Elements 
-----------------------------------------------------------------------
Weather Element       Descriptors           Data Type    Spatial         
                                                         Representation  
Waves                 Height                Numerical    Continuous      
Swells                Height and Direction  Numerical    Continuous      
Superstructure Icing                        Categorical  Bounded Area    
                                                                         
-----------------------------------------------------------------------

3.4 Domain and Resolution

The domain and resolution in both space and time must be sufficient to accurately detect any approaching phenomena that could influence the forecast.

3.4.1 Spatial

Although forecasts will be issued for just the WFO's area of responsibility, the domain must be large enough to view upstream phenomena that could affect the forecast. We also must accommodate aviation route forecasts (Transcribed Weather Broadcasts, or TWEBs), some of which extend beyond WFO boundaries. The forecast database will store data on the AWIPS Local scale. Resolution must be fine enough to define local effects sufficiently. Our initial plans specify a 10-km horizontal resolution; whether this is appropriate will be determined through investigation.

Occasionally, forecasters will need to define forecasts in higher spatial resolution than supported by the standard forecast grid. To support this requirement, the forecaster will be able to define, store, and reuse special local-effects areas.

There will also be a terminal forecast tool. Suppose a forecaster has forecast responsibility for two airports that are separated by only several kilometers. One lies at the bottom of a valley, and the other is much higher in elevation. Fog has enveloped the entire area, and it is the forecaster's job to predict when the fog will lift, allowing for normal aircraft operations. Expecting the fog to burn off early in the day at the higher airport, while the valley airport remains in fog most of the day, the forecaster creates a general forecast that shows the fog slowly dissipating with time, then uses the terminal forecast tool to define a specific forecast for the valley site that shows fog and low visibility remaining throughout the day.

3.4.2 Temporal

The time domain of the forecast database will generally be 0 to 6 days, reflecting current forecast requirements. Aviation-specific elements are limited to 24 hours (terminal forecasts) and 18 hours (route forecasts). Time resolution for most elements is one hour, though several require less (e.g., fire weather winds are all forecast with 12-hour resolution). Further, time resolution will generally be finer for shorter forecast projections and somewhat more coarse for longer projections. This approach will give us flexibility for expansion and customization and will provide an added benefit of saving storage space and time for performing interpolations.

The choice of resolution will affect the wording in the forecast. For example, if we were to limit ourselves to maximum and minimum temperature, we would not be able to produce rush-hour forecasts or create phrases such as "sharply colder in the afternoon." On the other hand, retaining a 1-hour resolution in later periods may be inappropriate since the guidance will not have that resolution and the forecaster cannot accurately distinguish between one hour and the next after several days, e.g., between hours 95 and 96.

Time in the database will be reckoned from the current hour, rolling forward each hour. Users will work in terms of valid time, so this should not result in any confusion.

In the future, when W/W/A functions are integrated with AFPS, it will be necessary to increase time resolution to 1 minute.

Footnotes

(1): The AFWG met for the first time in February 1993. This group, comprising representatives of each NWS Region, the Office of Meteorology (OM), OH, NMC, the AWIPS Program Office, and FSL and TDL, will help ensure that NWS operational needs are properly considered during development.

Table of Contents

Next Chapter