Feel free to check out our graphical version of the Hurricane Local Statement at the following address: http://WWW.SRH.NOAA.GOV/MLB/hlsfiles/hlsmain.html .

Constructive comments are always welcomed...

1. INTRODUCTION

The National Weather Service (NWS) office in Melbourne, Florida has piloted an experimental project that uniquely presents the daily "Hazardous Weather Outlook" (HWO) product in graphical format. The purpose is to complement the official text version of the HWO, exploiting both the communication and display capabilities of the Internet (Zaff, 1999). Due to the diversity of threats, as well as the chance for multiple coincident threats, the text version (alone) is often insufficient to properly express all weather concerns. As a result, the text HWO may become overwhelmingly large in order to accommodate detail, or overgeneralized to accommodate product length manageability. Adding a graphical component to the HWO greatly improves its usefulness and may revolutionize the way the NWS provides hazardous weather information to its users.

2. GRAPHICAL PRODUCT DEPICTION
 

Fig. 1. Main HWO control panel

The experimental graphical HWO (gHWO) is composed everyday coincident with the textual HWO and is issued between 6 to 7 AM. Users consist of Emergency Managers, Law Enforcement, Water Management Districts, Media, other government agencies, and the public, with a subset of each serving as the customer feedback group. The intent is to provide decision makers with the necessary hazardous weather information for planning purposes early each day. Updates and amendments are made whenever necessary, but a routine update is provided at 11 AM each day during the central Florida warm/wet season to better depict the forthcoming afternoon convection. The gHWO consists of a combination of interactive text and web graphic products that individually address expected weather threats for the next 12 (up to 24) hours. Upon entering the main gHWO web page, the user is shown a control panel (Figure1) which indicates hazard categories relevant to east central Florida. The hazard categories include THUNDERSTORMS (severe and strong), FLOODING, SEASONAL, MARINE, and WILD FIRE. The control panel acts as the center for navigation where the user can quickly view each threat category in separate web frames.

When selecting a highlighted hazard category (highlighted by the color-filled circle) on the control panel, information about that threat is made available. A hazard category may contain several specific hazards. For example, the FLOODING category addresses (Flash) Flooding, River Flooding, and Coastal Flooding. The SEASONAL category has two versions; one for the warm season and one for the cool season and addresses such hazards as Heat Stress, Wind Chill, Rip Currents, Fog/Smoke, and Sustained Winds. Each day, current hazard information is depicted through a combination of Degree of Threat (DoT) bar charts and Threat Area Maps (TAM). Ready access to supporting information such as preparedness, safety rules, fact sheets, and frequently asked questions is also available.
 
 
 

Fig. 2. Degree of Threat bar chart showing Wild Fire threat definitions for each corresponding threat level. Each hazard has its own unique set of threat definitions.

The DoT bar chart (Figure 2) concept was adopted in order to express the level of threat for a particular hazard while offering some reasonable form of day-to-day (forecaster-to-forecaster) consistency during product composition. Each DoT has five threat levels with Level 5 representing the highest threat. Importantly, if Level 1 is achieved then the user must recognize that at least some discernible degree of threat exists for that hazard. If a bar chart does not have any levels highlighted, then the corresponding hazard is not expected within east central Florida. It would be ideal to provide a DoT for every hazard for each individual county. However, at this time it is not practical and, at times, is beyond current forecast skill. Therefore, for now, the DoT will depict the highest level of threat existing anywhere within east central Florida. Great care is taken not to inadvertently minimize any hazard threat.
 
 

Fig. 3. Example of the Threat Area Map used to depict locations at greater risk or the timing of hazards.

A separate TAM (Figure 3) is linked to each DoT bar chart through a navigation tab at the top of the page. The TAM can be used to highlight locations at greater risk or to delineate the timing (arrival/departure) of a particular hazard. The TAM is color coded to correspond to the current DoT levels (colors) and gives forecasters the opportunity to express geographical dependencies using a combination of graphics and superimposed text boxes. It is imperative that the user reference both the DoT and TAM in order to fully comprehend the hazardous threat; they are meant to be used in tandem.
 
 

Fig. 4. Example of expected Spotter Activation.

Accessible through the main control panel is the AStatus of Spotter Activation@ and the "Situation Room". Each day, a county-by-county depiction of expected spotter activation is given (Figure 4) so that storm spotters can also have the necessary information required to plan their day in support of the SKYWARN program. Activation levels include Self Activation, Formal Activation, and WX4MLB Activation. Finally, a ASituation Room@ is provided for when hazardous weather begins to break. At such time, short-fused products and real-time data become more relevant. NWS watches, warnings, advisories, and forecasts can be obtained along with radar, satellite, and observational information. It is important to note that the NWS can not guarantee the timely availability of warning information across the Internet. However, the Internet has provided a wonderful laboratory to experiment with graphical products and is a means of information dissemination that should not be ignored
 

3. THE DEGREE OF THREAT

For each hazard, assessed threat is separated into five ascending threat levels. For longer-fused hazard situations (i.e., Wild Fire, Coastal Flooding, Heat Stress, etc.), threshold definitions can be formulated based on historical knowledge, local experience, and the current tiered watch, warning, advisory infrastructure. For example, Wild Fire threat levels (see Figure 2) can be reasonably defined as Level 1 - AFire Danger@ (forecast Fire Weather Watch conditions), Level 2 -AHigh Fire Danger@ (forecast Red Flag conditions with winds less than 15 mph and no lightning, Level 3 - AVery High Fire Danger@ (forecast Red Flag conditions with winds less than 15 mph but with lightning, Level 4 - AExtreme Fire Danger@ (forecast Red Flag conditions with winds 15 mph or greater but no lightning, Level 5 - AExtreme Fire Danger@ (forecast Red Flag conditions with winds 15 mph or greater and lightning). Tuning the definitions to the local area is important and minor definition adjustments are still in progress.

Conversely, for short-fused hazards (mainly convective hazards), representation of local threat becomes more difficult. For example, with respect to Severe Thunderstorm hazards (tornadoes, damaging wind, and large hail), concerns are governed by the confidence that each independent hazard will occur within the county warning area and its subsequent coverage. More so, each presents a differing (greater/lesser) public risk, depending on the particular hazard and its potential severity. Therefore, a function was developed to help forecasters make a reasonable determination of the appropriate DoT. The DoT function is expressed as C(AxB); where AA@ is the percent Confidence (of hazard occurrence), AB@ is the percent Coverage, and AC@ is an Impact Coefficient (Figure 6). The product of Confidence x Coverage yields a range of values between 0 and 1.0. The Impact Coefficient is applied to heighten the threat level for increasing hazard magnitudes as they relate to potential impacts to life and property. Empirical values of 1.0, 1.25, and 1.5 are currently used and are being evaluated.

If properly scaled, threat levels for Severe Thunderstorm hazards, for example, can be compared against each other to Aclassify@ an anticipated event as being either a high wind event, or a large hail event, or a tornado event. In this way, the primary hazard(s) can be highlighted without minimizing the threat of the other associated hazards. Otherwise, due to the inherent nature, damaging wind and large hail would always have a greater threat than tornadoes (if based on confidence and coverage alone) which may not adequately depict the proper degree of threat to the public. The Storm Prediction Center employs a variation of this concept by reserving the AHigh Risk@ terminology for tornado outbreaks within their Convective Outlooks. Figure 5 depicts the DoT for an event similar to the 25 March 1992 central Florida Hailstorm when using this scheme.
 

Confidence Factor	Event Occurrence	A - Values
Minimal Chance	10%	.1
Slight Chance	20%	.2
Chance	30, 40, 50%	.3, or .4, or .5
Likely	60, 70%	.6, or .7
Categorical	80, 90, 100%	.8, or .9, or 1.0
Coverage  Factor	Counties Affected	B - Value
1 County (isolated)	10%	.1
2 - 5 Counties (scattered)	20 - 50%	.2, or .3, or  .4, or .5
6 - 10 Counties (numerous)	60 - 100%	.6, or .7, or .8, or, .9, or 1.0
Impact Coefficient	Event Thresholds	C - Value
Some potential impact to life & property	* wind gust 50 - 54 kt * hail 3/4 - 1 in (-> quarter size)	1.0
Greater potential impact to life and property	* One to a few F0/F1 tornado * wind gust 55 - 65 kt * hail 1 - 1 3/4 in (-> golf ball size)	1.25
Greatest potential impact to life and property	* one or more F2+ tornado * outbreak of F0/F1 tornado * wind gust 66+ kt * hail 2+ in (baseball +)	1.5

Fig. 6. Table of Values for determining the Degree of Threat for short-fused (convective) hazards.
 
 

Fig. 5. Depiction of the Degree of Threat for an event similar to the 25 March 1992 central Florida Hailstorm (used as an example). Severe weather was expected in the form of bowing segments within a broken squall line of thunderstorms. Large hail and damaging winds were the primary threats. A notable threat of tornadoes exists, but to a slightly lesser degree.

 
 
 

The scale which is used for shorter-fused hazards was subjectively derived and is shown in Figure 7. The scale ranges from 0.02 to 1.0+. It was decided that the minimal acknowledgment of threat would be either a 20 percent confidence of hazard occurrence over one county (0.2 x 0.1) or a 10 percent confidence over 2 counties (0.1 x 0.2) using an impact coefficient of 1.0. Therefore, 0.02 represents the low end of the scale. The high end is 100 percent confidence over all (10) counties using an impact coefficient of 1.5. Therefore, hypothetically, 1.5 is the maximum attainable value, but the high end is truncated at 1.0 since the impact coefficient was not used when determining the scale.

Level 3 represents a AVery High Hazard Threat@ with the maximum value within its range determined by using 50 percent confidence and 50 percent coverage to yield a value of 0.25. This concept was used to gauge the other levels so that the maximum value within the range for Level 1 is 0.05 (50 percent confidence and 10 percent coverage); Level 2 is 0.15 (50 percent confidence and 30 percent coverage); Level 3, again, is 0.25 (50 percent confidence and 50 percent coverage); Level 4 is 0.35 (50 percent confidence and 70 percent coverage). Finally, any value over 0.35 would fall into Level 5.
 
 

Fig. 7. The Degree of Threat Scale for Short-fused hazards.

Once again, using an event similar to the 25 March 1992 Hailstorm as an example (Figure 5), one forecaster=s assessment of the DoT for each hazard might be obtained by the following thought process below. Severe weather was expected in the form of bowing segments within a broken squall line of thunderstorms moving southeast through central Florida. This example is provided to merely show DoT assessment.

a) For Stronger Tornadoes - There was a 30 percent confidence with a 40 percent coverage (4 counties affected). Since the impact coefficient is 1.5 for F2+ tornadoes, the DoT is determined to be 0.18/Level 3 (1.5 x 0.3 x 0.4).

b) For Weaker Tornadoes - There was a 40 percent confidence with a 60 percent coverage (6 counties affected). Since the impact coefficient is 1.25 for F0/F1 tornadoes, the DoT is determined to be 0.3/Level 4 (1.25 x 0.4 x 0.6).

c) For Damaging Wind - There was a 60 percent confidence with an 80 percent coverage (8 counties affected). Since the impact coefficient for forecaster determined gusts between 55 to 65 kt is 1.25, the DoT is determined to be 0.6/Level 5 (1.25 x 0.6 x 0.8).

d) For Large Hail - The same as for damaging wind using comparable values (golf ball hail possible).
 

4. THE THREAT AREA MAPS

The Threat Area Map template is a blank map of NWS Melbourne=s county forecast and warning area. This serves as a canvas for the forecaster to depict geographical dependencies corresponding to the respective DoT. As mentioned previously, the TAM is used to depict areas at greater risk, and/or to delineate the timing of a hazard (Figure 3). To minimize confusion, forecasters are restricted to the seven colors of the DoT color palette when preparing TAM graphics. Using the DoT and TAM combination, the forecaster has a powerful means to relay an abundance of information in a very concise way.

5. CONCLUSIONS

With the ability to present the HWO graphically, NWS Melbourne can provide valuable information to decision makers in a concise format. Although the gHWO project is experimental, it does speak to the current National Weather Service=s Strategic Plan (NOAA/NWS, 1999) to support the development of graphical products in order to better serve its users. So far, feedback has been very positive. Comments from the customer feedback group have been compiled with a list of suggested improvements. Many will be incorporated within the next gHWO upgrade. Future work for the gHWO project at MLB will include the addition of lightning and tropical cyclone hazards. Also, a verification scheme is in the works.

All graphics within the gHWO are generated using Microsoft=s PAINT software. This software is very simple to use and is available on most computers.

The Graphical Hazardous Weather Outlook for east central Florida can be viewed on the National Weather Service Melbourne Internet homepage at the following address: http://www.srh.noaa.gov/mlb

6. ACKNOWLEDGMENTS

The authors would like to thank Bart Hagemeyer, Peter Blottman, and Scott Spratt for their positive comments and suggestions while developing the initial version of the gHWO. The NWS Southern Region Headquarters is recognized for hosting the NWS Melbourne web site. Finally, high regard is given to all of the forecasters at NWS Melbourne for their forward thinking and hard work in the daily preparation of these experimental graphics.

7. REFERENCES

Zaff, D. 1999: Western region ATotal Forecast@ and Graphical Watch/Warning Displays via the Internet NWS Western Region Tech Attachment 99-03.

NOAA/NWS 1999: National Weather Service Strategic Plan for Weather, Water, and Climate Services 2000-2005.

___________________________________________

Corresponding author address: John Pendergrast NWS Office, 421 Croton Rd. Melbourne, FL 32935;

e-mail: john.pendergrast@noaa.gov

Back to home