SR/SSD 97-46

10-15-97

Technical Attachment

The Use of Forecast Rainfall Amounts in Zone Forecasts

John F. Robinson

NWSFO Little Rock, Arkansas

1. Introduction

The demise of the Agricultural Forecasts by the National Weather Service ended the availability of forecast rainfall amounts to the general public. At approximately the same time, NWS offices began providing Basin-Average Quantitative Precipitation Forecasts (BAQPF) to the River Forecast Centers (RFCs), for input into the river forecast models. The forecasts for the RFCs involved a new investment in time and effort, but virtually none of the information provided to the RFCs reach the public directly.

Phone calls received by the forecast office indicated that many people, especially those involved in outdoor work, needed forecasts of rainfall amounts to plan their workday properly. In addition, talk among forecasters often involved decisions in their own neighborhoods that revolved around how much rain was expected. A couple of examples: Even though we have some pretty big probabilities in the forecast, I don't think I'll fertilize my lawn. It just looks like the rain amounts will be too small; or You should have seen my neighbors. They just had a truckload of topsoil delivered, and that heavy rain last night washed half of it down the street.

So, it seemed that forecast rainfall amounts ("Public QPF") would be of use.

After considerable thought as to methodology, an experiment was devised to provide Public QPF. Southern Region Headquarters approved the idea, and NWSFO Little Rock conducted the initial experiment from December, 1996, to June, 1997.

This paper describes the initial ideas and rules that were devised to test the concept, lessons learned from the experiment, and the recent resumption of the test.

2. Initial Methodology

From the start, it was clear that Public QPF would have to be somewhat different from the BAQPFs provided to the RFCs. For one thing, the public would have a very difficult time understanding BAQPF, since this method averages the zero amounts with the amounts where rain does occur. Thus, it was decided that Public QPF would need to use the concept of giving average rain amounts only for places where rain occurred.

The next step was to determine how to get the rainfall forecasts to the public. A separate product was considered, as was adding the QPF to the State Forecast. However, it was decided that inclusion of the amounts in the Zone Forecasts would reach the widest audience.

Next, the forecast "rules" were devised. The QPF would normally be included only in the first 24 hours of the forecast (the first and second periods), in order to keep the Public QPF time frame in line with the one used for the RFCs. However, if the forecaster was confident enough to mention heavy rain in the third or fourth periods of the Zones, QPF could be included there, also. The QPF forecasts would match the discrete period-by-period format of the Zones; thus, there would not be a "storm total" as is used for winter storms.

QPF would not be included for winter weather events, since this would involve using water equivalents, which are not well-understood by most of the public. QPF in such cases would also detract from the emphasis on snow or ice amounts, which are more important in such situations.

Past experiences with rainfall forecasting (including the basin-average QPFs), indicated that some forecasters might use odd amounts, at times. Thus, standard forecast values were to be used: less than 0.10 inch, 0.10 inch, 0.25, 0.50, 0.75, 1, 1.50, 2, 3, 4 and on up in whole inches. Decimal numbers were specified since they routinely appear in river summaries, climatological summaries, and other products. Forecasters were urged to use "around" a certain forecast value, when possible. If a range of values was necessary, the range was to be kept as small as possible. A range and a phrase such as "with local amounts to [a particular value]" could be used when absolutely necessary, such as embedded convection.

As an educational tool, amounts were also to be used for the "trace events", such as drizzle and sprinkles. In these cases, "trace" was normally to be used for the QPF, but "less than 0.05 inch" could be used to indicate some uncertainty in such things as warm advection situations.

Forecast rainfall amounts were to be placed at the end of each forecast period, to avoid disrupting the usual sequence to which customers were accustomed. An exception was allowed for heavy rain events. In such instances, the QPF could be moved up into the "body" of the Zone, in order to emphasize its importance.

Finally, for rain probabilities up through 50 percent, phrasing such as "Where rain occurs...

amounts will average..." was to be used. This was to emphasize the necessity of considering both the probability and the amount, since most places would not receive rain in such situations. For rain chances of 60 percent or greater, no qualifying statement would be used; the forecast would simply use phrasing such as "Rainfall amounts will average..."

3. The Experiment Proceeds

The experiment proceeded smoothly, with only very minor problems. Few forecasters asked for any changes in the rules. Subjectively, the number of phone calls to the forecast office decreased. The use of the QPF on local television weather shows was quite sporadic, however. Before the experiment began, many of the TV broadcasters were surveyed. Some were enthusiastic, others thought the idea was OK, and one thought QPF was just another case of sticking our necks out. So, the lack of any regular use of the QPF on TV was rather disappointing.

4. The Experiment Comes to a Close

For a couple of weeks surrounding the end of the test, periodic messages soliciting comments were sent on NOAA Weather Wire and broadcast on NOAA Weather Radio. E-mail comments were solicited on the office's home page.

Altogether, 53 responses were received, and 52 of these were positive. Of the 53 respondents, 44 people stated the area in which they lived. Only 11 of the replies came from metropolitan Little Rock; all the others were from rural areas or small towns. Seven alone were from Stuttgart, in southeast Arkansas. Most of the out-of-town comments were received via telephone, meaning that these people either had to spend money on long-distance calls or use some of their cellular airtime to voice their opinions. The responses indicated that the majority of the people were getting the forecasts from NOAA Weather Radio or the Internet.

Activities or occupations that people said were affected by the QPF forecasts included: Gardening/yardwork (18), farming (15), outdoor activities/recreation (11), travel (9), camping (5), ranching (2), trucking (2), building construction (2), and other outdoor work (2). The following were each mentioned once: radio news director, a TV meteorologist in Springfield, Missouri (the station's coverage area includes quite a few counties in northern Arkansas), letter carrier, waste water utility, roofing, golf course maintenance, and household purposes. (The total adds up to more than 53, since a number of people mentioned more than one activity.) Clearly, some of the occupations and activities were expected, but others were surprises.

Material for forecast verification was collected. The results have not yet been tabulated. Subjectively, however, forecasters tended to overforecast small amounts and underforecast large rainfalls. Interestingly, three of the public responses indicated that the people liked the QPF because it was so accurate.

The one negative reply noted that the particular town was served by an older cable TV system which could not display all the characters in the lengthened forecasts. Several forecasters and hydrometeorological technicians at the NWSFO noted that, even on The Weather Channel, the third period of the forecast was sometimes truncated when the forecast was especially lengthy.

5. The Next Step

Given the results indicating that customers were well-served by the Public QPF, Southern Region Headquarters indicated that QPF in NWSFO Little Rock's Zone Forecasts should resume. The resumption took place on October 1, 1997.

To allay some of the concern about lengthy forecasts, several rules were changed.

The phrasing of the QPF statement was shortened to "Average rainfall..." (followed by the specific value or range), no matter how large or small the rain chance is. Periodic statements on the Weather Wire and NWR will be used to remind customers that the probability and the QPF must be considered together.

Forecasters have also been encouraged to include the probability in the body of the forecast, rather than as a separate statement. For example, "Mostly cloudy with a 30 percent chance of showers," rather than "Mostly cloudy with a chance of showers ... chance of rain 30 percent." This method also reduces the number of characters necessary. This wording was not made mandatory, however, since separating the "rain wording" and the probability can sometimes emphasize a certain portion of the forecast very effectively.

QPF for "trace events" was dropped.

Fractions are now being used, rather than decimal numbers. While the latter still seem more logical, they do not display properly on The Weather Channel, sometimes causing misleading numbers. Apparently TWC's software always places a space after a period, whether a space is needed or not. This caused a number such as 0.50 inch to come out as 0. 50 inch. In the worst cases, the "0." appeared on one page of display and the "50 inch" on the next.

Because of a request by several forecasters, 1 1/4 inches was added as a forecastable value.

An Example Forecast, following the revised rules:

.Today...Mostly cloudy with a 40 percent chance of rain. High in the mid 60s. South wind 5 to 10 mph. Average rainfall 1/10 inch.

.Tonight...A 70 percent chance of thunderstorms. Low near 50. Southwest wind around 5 mph. Average rainfall 1/2 to 1 inch.

.Wednesday...A 40 percent chance of rain in the morning...then becoming partly cloudy. Cooler with a high in the upper 50s. Wind becoming northwest 10 to 15 mph. [No QPF is used here, since this is the third forecast period.]

5. Conclusions

The initial experiment clearly indicated that Public QPF in the Zone Forecasts can be useful to many customers. Additional educational efforts will be necessary in order to gain more widespread use by broadcasters, and thus put the QPF before a greater percentage of the public. The latest set of "forecast rules" is by no means carved in stone, and further adjustments may be necessary based on public and forecaster input. Still, the drawbacks seem fairly minor in comparison with the benefits indicated by our users.