Some Weather Forecasting Inside Baseball

Courtesy: www.coolwx.comA little ‘inside baseball’ on weather forecasting. The graphs on the left (Courtesy: – click the image for a better look) show Wednesday’s weather in New Haven as forecast over time by various computer models.

The ‘ptype’ forecast has been all over the place. What’s it gonna be? The guidance has waffled between snow, sleet and freezing rain.

Even if I got it right on the last forecast, it was little consolation to those who’d watched earlier. That made me very unhappy. Sometimes there’s no choice but to change the forecast. You can’t feel married to it.

I traded tweets with a former co-worker this week who shares my angst. It really made storm nights, hell.

Most people don’t realize the most important part of the forecast is, impact. There are fewer potential impacts than storm parameters.

If the timing is right… if the road hazards/conditions are right… if the school situation is properly handled, then how much snow falls or whether it’s a freezing sleety mix don’t matter as much.

But it killed me every storm. There was never a forecast I was really happy with. Not one.

I Remember Nights Like This

I just traded quick tweets with Ryan Hanrahan at NBC30.

geofffox: @ryanhanrahan How’d you do on this one so far?

ryanhanrahan: @geofffox Not bad – all about where fgen sets up. Haves and have nots. Some towns will get hit hard others won’t.

I don’t know about Ryan, but these were the nights I dreaded! I was on-the-air 28 years in Connecticut. It didn’t take long to realize how unhappy people are when a snow forecast goes wrong.

Ouch. Some were brutal.

Even when right I’ve been blamed for the forecast on other stations and the Weather Channel and by people who just misheard!

Forecasts aren’t blown because you haven’t worked hard enough. Predictions go south when bad guidance (computer models) leads you astray. By and large computers are superior to humans in quantifiable solutions to tough atmospheric problems.. That makes it difficult to discard them in pressure situations.

Mid-storm I was like a caged animal. I’m sure that didn’t make me a dream co-worker.

Sadly, no matter what I did it was never 100% right. There was always an outlier. Frustrating.

Post-Sandy the federal government allocated significant resources to beef up our weather computing power. Implementation is excruciatingly slow. Forecasts will improve a little. The low hanging fruit has already been picked.

The Advantage Of Not Forecasting On-The-Air

I’m watching the Pats/Broncos and remembering winter. They’re not pleasant memories.

If I was still forecasting in Connecticut, I’d have been talking about Wednesday’s potential storm for days already. Fellow forecasters, I feel your pain. The forecast has vacillated like a bride-to-be on “Say Yes To The Dress.”

Even today no one knows for sure. I certainly don’t.

However, the models have begun to stabilize. The forecast solution has become more consistent run-to-run.

Wednesday looks like rain all across the East Coast. In fact, it looks like rain most of the way from Canada to Florida! Early Thursday the rain turns to snow, but by that time the storm’s moisture should be mostly spent.

In New York City the potential is there for enough wind to keep the balloons grounded Thanksgiving Day. No one wants that.

Here in SoCal it’s temps near 70&#176 and a slight chance for rain Thursday and Friday. Slight.

Kenny Asks, Who Should We Trust?

This email just came in:

Hey Geoff,

Good Luck in your move. So I have to ask…Who should we trust for weather now ?


Good question. I’ve gotten it a lot.

First, there is little difference in accuracy between the four Connecticut TV stations providing a forecast. Nearly everyone’s correct nearly every day!

Don’t stop reading here. There are differences other than the numbers.

I do have a problem with the automated numbers you see on many websites. By and large they are correct, but they’re usually not double checked by humans. I’ve seen stupid forecasts posted that humans would never allow.

To a viewer there is no difference between 66&#176 and 68&#176. A snow forecast of 5″ demands the same action as one of 8″. These differences are not very important. Don’t dwell on them.

The difference in weather forecasts is how the forecaster tells the story. Is his/her presentation understandable? Does he/she leave you with actionable information?

Watch the meteorologist you best understand, because accuracy doesn’t matter if you can’t translate their words into usable info.

I know most of the meteorologists in Connecticut. They are all honorable. None of them benefits from hyping the forecast–and they don’t.

Find someone you trust. In that regard meteorology is like everything else in life.

Hurricane Sandy: NWS Assesses Itself os assessments pdfs Sandy13.pdf-1

The Weather Service just released its Hurricane Sandy “Service Assessment.” Publications like this aren’t unusual. Every named or numbered storm gets some sort of after-the-fact scrutiny. Of course, Sandy is a special case, having affected so many people and so much property. This is a beefy report touching lots of bases.

The Weather Service and Hurricane Center did a good job–not good enough. There were weak points. That’s me speaking, though the report acknowledged them too.

I was on-the-air at FoxCT for Sandy. We used lots of NWS/NHC raw data and forecast products. A huge part of my job was assimilating the immense treasure trove of data available. Some of what we used was so esoteric, co-workers didn’t know it existed!

If there’s ever been a time my years of experience and nerdy curiosity came in handy, it was during Sandy.

Once Sandy moved north of Cape Hatteras the National Hurricane Center passed off much of its responsibilities to local forecast offices. That was a big mistake which served to confuse more than inform. os assessments pdfs Sandy13.pdfI said it then. Even worse, I’d said it before, having complained loudly and traded emails with the Hurricane Center’s director Ed Rappaport after Hurricane Noel received the same pass-off in 2007.

This Hurricane Center policy will be changed going forward. It’s about time!

For future storms like Sandy, NHC should be the principal point of contact responsible for the event, including delivery of a consistent suite of products and a unified communications protocol within NOAA, to key NOAA federal partners, and the media. NOAA/NWS websites should consistently reflect all watch/warning/advisories on websites, regardless of organizational structure or office/center responsibility. Web page design should ensure the most important message is quickly evident.

The are other recommendations, including a some having to do with coastal flooding and the current lack of definitive storm impacts. Giving a tidal flooding range in feet is worthless to most people. More important would be to say, “Lower Manhattan will be under water,” or similar specifics.

The truth is most non-professionals need a trusted voice. There’s too much for you to wade through.

I hope I was your trusted voice, leading you in the right direction. If you were watching us on FoxCT you weathered the storm without any big surprises. It goes without saying I will miss being that voice for you in the future.

The NWS assessment and its findings and recommendations should help all of us do better next time. There will be a next time.

How The Season’s Change Affects The Forecast

Because of its tilted axis the Earth is heated unevenly. Its surface is also rough and irregular. All of this creates temperature and pressure differences which cause winds–especially upper air winds.

My guess is you don’t pay attention to the weather as much as I do. We’re about to enter a transition period. Meteorology works a little differently during the warm months.

Because of its tilted axis the Earth is heated unevenly. Its surface is also rough and irregular. All of this creates temperature and pressure differences which cause winds–especially upper air winds. Temperature contrasts are stronger in the cold months and that drives everything!

When the upper winds blow strong weather systems move quickly. That’s what we’re leaving.

From now until the fall weather systems will move slower. Fronts and air masses will be a little more spongily defined.

There are certainly forecasting differences. Timing becomes tougher. On convective days very small, but nearby areas will sometimes have radically different weather.

Of course you’re less critical because the forecast usually has less impact this time of year. I appreciate that.

The Forecaster’s Admission

Tomorrow morning’s piece is about accuracy in forecasting. Not how accurate I am, but when is the toughest time to forecast.

I write a little blurb in the Hartford Courant most mornings. It’s 500 characters, around 100 words. There is a finite space allotted and I can’t color outside the lines! I have to write about the weather, but beyond that they’re pretty flexible.

Tomorrow morning’s piece is about accuracy in forecasting. Not how accurate I am, but when is the toughest time to forecast.

The simple answer is right now!

We are most accurate when systems are flying by. When they move s-l-o-w-l-y we start to sweat!

Should I lie? This is how it is.

I’m not totally sure why this is so, but the most likely answer is systems move slowly when upper air influence is low. A gentle nudge from a lesser force won’t mean much to a low pressure system doing 35 mph, but when that same system is progressing at walking speed the nudge is a more significant portion of the puzzle.

We see this a lot in tropical systems. Most of the time they’re in very light upper air and creeping along. That’s when we complain how tough they are to predict.

As we head to winter the jet stream will pick up and these lazy lows won’t appear as often. We won’t have 100% accuracy, but we’ll do better than we did this weekend. Like I didn’t know you noticed.