A Mid-Hurricane Season Check-in and Discussion with Phil Klotzbach

Earlier in the summer, Colorado State University and other agencies that give seasonal outlooks for tropical cyclone activity predicted a very active Atlantic hurricane season. But it’s mid-season now and the Atlantic basin is pretty quiet. What’s going on? We sat down with Phil Klotzbach, senior research scientist at Colorado State and expert in seasonal hurricane forecasting for the Atlantic basin, to discuss the lull in activity. Full video of interview and a summarized version of the discussion — with links + additional context in brackets — provided below.

Forecasters called for an extremely active season, and all signals seemed to point that way. But leading up to the peak of the season we saw very little activity. Why?

We use several different methods to forecast seasonal hurricane activity. Conditions leading up to the beginning of the season suggested a “bananas” season. We use statistical-dynamical models which use ECMWF or UKMETs forecast of wind shear and SSTs during the peak of the hurricane season. Those also forecasted an active season. And the dynamical models, just by counting the number of storms in these long-range models, also suggested an active season. Hurricane Beryl broke records for being the earliest category 5, which also indicated an active season.

The fact that one storm has formed since Aug 12 is ridiculous. It’s been too dry. It’s hot but you can’t sustain convection.

In August shear was low, most easterly shear on record. [Climatologically the shear is strong and westerly.] Normally moisture is not a good predictor of seasonal activity. But this years it’s too dry.

It’s been a weird season. The beginning was so active but we’ve seen so little activity recently that we are currently standing at normal. My confidence in forecasting now is low but we expect activity to increase at the end of September.

So this is well outside the range of activity that you predicted?

Yes. If we doubled that ACE [Accumulated Cyclone Energy, common metric for overall tropical cyclone activity] from now to the end of the season, which is fairly common, then our forecast would have given the odds of that amount of activity at 1%.

Different things have been responsible for the lack of activity. Mid-August we had no tropical waves [the seeds for many hurricanes], they came off too far north. Normally when the monsoon trough [indicates the path of the African Easterly waves] is further north we see more activity but this year it is so far north it dumps the waves into cold water. Since then we’ve had more waves, we have three currently in peak season but these waves aren’t really developing which is unusual.

Bill Gray used to ring a bell on Aug 20 to signal the active part of the season. This year we rang the bell and had one storm.

We’ve also had no high latitude storms this year. It’s strange to not get anything.


The Atlantic is still hot, ENSO is trending towards La Niña, a favorable MJO phase is coming, models are picking up activity in 10+ days. The season should become active again in Late September and October. The monsoon being too far north should not be a problem since it starts to move south this time of year. But at this point nothing is close enough that I have any confidence.

The monsoon trough position seems to have played a crucial role this season, but it hasn’t been discussed much in research. Is there any predictability or subseasonal variability associated with the monsoon?

Bill Gray did have work showing that a further north monsoon does lead to more activity but that is assuming the waves reach water around Senegal. There isn’t much research on what happens when waves come off too far north. Occasionally a storm will form right off the coast and go north. That will be a focus in our future research.

We like to test if SSTs can give us the general pattern we see in the winds. What we plan to do is run a model with this year’s observed SSTs in the Atlantic (and average SSTs over the rest of the globe) and see if it will capture the position of the monsoon. My hypothesis is that because the North Atlantic was warm and the equatorial Atlantic was cool, we will see the monsoon shift north.

So because you hypothesize the monsoon position is tied to SSTs, that would suggest it varies mostly on seasonal time scales. Right?

Correct. Although equatorial SSTs have warmed a lot recently and the monsoon has come back south.

There is also some subseasonal variability that is likely causing the dry air, maybe it’s the NAO but we need to run some tests. We have a good team looking at stuff from a bunch of different angles.

What went wrong? Can we forecast it?

The global models don’t give me a lot of hope, the first sign that there would be a lull in the season was an early August ECMWF forecast for below average activity in two weeks. We didn’t believe the models then but we can run diagnostics on the model to see what factors were important and led to the lack of activity.

Now the models are showing some storms in the long range [forecasts out 10+ days].

Francine formed recently in the Gulf of Mexico and hit Louisiana as a category 2. If Francine had formed earlier and gotten stronger like some models were saying I wouldn’t have bothered putting an update on why the season was so inactive recently. But Francine took a while to form and we did a ton of interviews about the lack of activity last week.

I’ve seen the strongest signals for activity in days 11-15 for a while. The model ensembles are aggressive and the MJO should enhance convection in the western hemisphere. Maybe we’ll have a bunch of late season activity.

We could have a season like 2020 which had, I think, four major hurricanes in October and November.

We actually had five!. The previous record was two. So I hesitate to say the season is a dud because there is still plenty of time for devastating hurricanes to occur.

But the reason I think this inactive period has been so noteworthy is because we have had absolutely nothing, not even a weak tropical storm for so long and because we had such a bullish forecast. This wouldn’t be a big story if we had forecasted an average year. We had 2005 [the year with Cat 5 Hurricanes Katrina, Rita, and Wilma] in our analog list which made me nervous. We didn’t have 2013.

I was just about to bring up 2013. That was the last year where there was a significant overprediction of the hurricane season. Are there any similarities between this year and 2013?

2018 is like the anti-2013 we expected a below average season but activity was slightly above average. The thing about 2013 and 2018 is they were about in the middle with respect to shear. [I think the implication being it isn’t too surprising that the seasons swung active or inactive.] This year all the signals indicate an active season and we forecasted the signals, SST and shear, correctly, but there have been several subseasonal things that reduced activity. The monsoon position and MJO were timed poorly for TC formation. Francine also encountered this TUTT [tropical upper tropospheric trough] that sheared it in the Caribbean. Had Francine arrived 2 days later it likely would have been a major hurricane. You can always have these wild cards. The MJO is moving slow and has been in unfavorable phases during peak season. But the water is still warm and upper-troposphere temperatures should cool, so activity could pick up.

I wanted to bring up climate change. In your discussion warm upper tropospheric temperatures may have increased stability and suppressed TC activity. There has been lots of research on climate change leading to warmer upper-tropospheric temperatures. And you say that SST patterns are likely linked to the shift in the monsoon. Could climate change have been partly responsible for this lack of activity?

This year I think the upper troposphere is warm because last year we had an El Niño, there is a lagged warming effect which is expected. Normally upper- level tropospheric temperatures don’t have a strong correlation to hurricane season activity. But maybe there’s a threshold that was reached this year that lead to an inactive season. This year I think it was related to El Niño, maybe in the future it’s related to a climate change signal.

I haven’t seen much that says that the SST pattern that shifted the monsoon, of the North Atlantic being warm and the equatorial Atlantic being cool, will become more common in the future.

If we end up with a busy late season then our forecast won’t have been too wrong. We’ve already had a category 5 so it isn’t as bad as 2013. We’ve already had three hurricanes hit the U.S in that sense we are above average.

It’s a weird season, monsoon rainfall has been high over Africa. Trade winds have also been weak which usually enhances activity, but this year they are so weak that nothing is moving. It seems like the Atlantic was too hot. Normally conditions you want [for favorable hurricane activity] are there but this year they were cranked too high. But there hasn’t been any work I’ve seen that says you can turn the thermostat up too much that will cause the hurricanes to come down. But it’s hard because if the MJO was timed differently with relation to the monsoon we’d probably have 2 or 3 hurricanes.

So some of it is noise?

Yeah, that’s part of it. This forecast isn’t going to verify great. How much do you change when one year doesn’t behave the way you expect? There is still time for an above-average season to occur. If the rest of the season is below-average then we’ll worry about it. We need to understand this season better so that we can better model it. You would hope in August you can see some of this stuff coming.

So it was unusual that even in August we didn’t have a clear idea of what the season would be like.

Correct. Had ECMWF four-week forecasts been reliable, and I don’t want to dump this on them that is a very difficult problem, then we would have paid more attention and anticipated the lull in activity. But at long- lead times forecasts are often wrong.

Weeks one and two are really the limit of where global models have skill in forecasting TCs.

Yeah.

Do you expect the rest of the season to be above average? And how confident?

Well right now I’m not confident. The only chance for a hurricane until Sept 23rd is this depression (now Tropical Storm Gordon). But models are active in the long range. We’ll see. Last week in September is when I think things will get more interesting.

So maybe a busy end of September and October. Finally, what have we learned this season and how will you apply that to future forecasts?

We’ve learned a lot. Even when everything is pointing in one direction its doesn’t always pan out that way. So it gives us a dose of humility. There’s a lot of factors we try to assess for the forecast but there’s obviously some ways we haven’t looked at a season before. We need to understand what “broke” the season, and can we forecast that? Obviously again is how the rest of the season turns out. We will see what happens in the second half of hurricane season.