Guest commentary

Sure, the above title is a bit tongue-in-cheek as I think back on one of my most memorable forecasts in this Veterans Day week:  a forecast I made 26 years ago for Veterans Day, November 11, 1987.

One the eve of D.C.’s biggest November snowfall on record, I forecast cold rain and “possibly even some wet snow flakes”.  I thought I was a bit out on a limb mentioning wet snow flakes in early November.

It was a huge busted forecast.

Veterans’ Day services in the snow at Arlington, Virginia, November 11, 1987. The afternoon forecast was for partly cloudy skies with temperatures in the 40s but 11.5” of snow fell at National Airport, with some areas picking up more than a foot of snow. Source: Washington Weather

That cold rain turned into 11.5 inches of snow and still ranks as one of the top five surprise snowstorms in Washington.  I do remember looking at the various weather charts and my forecasting tools on November 10, 1987 and thinking to myself, “Boy if this was December, January or February, this would be a big snow” . . . but not November.  It was a true black swan event in my forecasting career.

I mention the Veterans Day storm of 26 years ago, because about 26 years before that date (the early-mid 60s), the science of weather forecasting had begun the historic move into numerical weather prediction (NWP).  NWP is heart of modern weather forecasting, the process that uses computers to solve the fundamental mathematical equations of the physical laws that govern our current and future weather.

This digital revolution in weather forecasting in the last 60 years has seen incredible progress – certainly a more accurate Veterans Day forecast today than 26 years ago.  Yet the effective communication and corresponding wording and graphics of today’s weather forecasts have not kept pace with the great advances in the science.

Words have meaning. And yet many of the words we still use confuse more than help the public make the best weather-related decision.

All the employees I know at the National Weather Service (NWS) are dedicated (and many overworked) professionals who are passionate about what they do, from the newest intern to senior forecasters to our science’s leaders. Yet its Sunday morning forecast for Tuesday in Washington was vaguely-worded: “ . . .a chance of rain or snow showers,” it said.

Forecasting tools such as ensemble numerical guidance (which Capital Weather Gang’s Steve Tracton has often written about) sure give us more to communicate than “chance of rain or snow.”

How about Sunday morning’s local digital forecast graph from the NWS?

That’s what generated the “chance rain or snow” for Tuesday.  Yet at the same time, this graphical ensemble forecast tool was available:

This graphic clearly indicated the chance of rain was much higher than any chance of snow.  Additionally, (even after Veterans Day 1987), the climatological probability of accumulating snow on November 11 is less than 10%.  Nevertheless, we are still stuck with digitally-generated worded forecasts that don’t reflect the current skill of our forecast systems.

For sure, a simple icon like this…

…is almost useless for making weather related decisions.

So many forecasts from TV to the NWS do not effectively communicate the uncertainty we know exists in a weather forecast.

Would you trust this forecast 10 days from now?

A high of 33 on day 10, not 32 or 34, but 33?  That singular number overstates the precision at which we can forecast that far into the future.

While numbers can mislead, words can confuse.

Look at the range of interpretations of the meaning of so many weather words we hear or read.  This results shown in the figure below are from a study done almost 20 years ago:

Little work has been done in the critical field of how we can more effectively communicate ever more accurate weather forecasts.  “Chance of rain or snow” was about what we could do 60 years ago. We all could do much better today.

The Capital Weather Gang regularly gives a “Forecast Confidence” indicator.  Years ago, in an on-line survey of TV viewers, I found that almost 80% of viewers of local TV weather forecasts would like a measure of “forecast confidence” included with the forecast.  Here was the question, “ Suppose a forecaster rates his or her confidence in the forecast for each day on a scale of 1 to 10, with 1 being little or no confidence, and 10 being the highest confidence. How useful would this information be to you?” The results are here…

I think such confidence information would help weather related decisions.  Why don’t more of us and our local NWS forecast offices give a confidence measure with their forecast?  Not every forecast is a “guarantee” but very few now have low confidence.

Are forecast graphics like this more helpful than a measure of forecast confidence and numerical probabilities of weather events?

What do words in the above graphic “possible”, “expected”, “higher”, “heavy” mean to the average person looking at this graphic?  How many know where they live?

Alternatively, could a 3-D view of your local city, town, neighborhood or even house with an animation of the rising streams be useful if flooding is predicted?

Or would a snow measurement with time, or winds knocking down trees, or the tornado moving through your area,  better help you make the best weather related decision?  Shouldn’t we find out?

Technically our weather forecasts are more accurate every day and more great advances in our forecasting abilities is ahead.  But, our communication skills and methods need to improve as much as our forecasting skills.  In some cases communications methods still in use from decades ago have failed as Mike Smith wrote in “When the Sirens Were Silent”.  We forecasters can do better in communicating what we know and don’t know.

What does “chance rain or snow” mean to you?  What do you do?

Miscommunication and poor understanding of how we make weather related decisions sometimes can lead to tragic results as the loss of life as we learn too often in tornado outbreaks, hurricanes and winter storms.  Today’s weather forecasting “enterprise” is an end-to-end process from observation to forecast to communication to decision by the public or user.  A perfect forecast mis-communicated resulting in a poor or tragic decision is almost a useless forecast.  A forecast of “Tuesday chance of rain or snow” can easily be better communicated, even with our digital constraints as: “Tuesday 40% chance of rain, less than 10% chance of snow with no accumulation”.

Late this Veterans Day morning, the NWS has updated its forecast to say (bold indicates added emphasis): “Cloudy with a chance of flurries in the morning…then partly sunny in the afternoon.”

Ah that word “chance”

No “Ryan guarantee” for Tuesday, but I am very confident there will NOT be any accumulating snow in Washington and school will be open unlike the day after Veteran’s Day 27 years ago.  So far my friends at The Gang agree.  Do your homework.  : >)

The author, Bob Ryan, was the  chief meteorologist at  NBC4 from 1980-2010 and lead meteorologist at ABC7 from 2010-2013, prior to his retirement this past summer.