Five Questions w/ Chuck Watson

Chuck Watson in the TSU-18 Centrifuge, Cosmonaut Training Centre, “Star City,” outside Moscow

I sent an e-mail recently to Chuck Watson, Director of Research and Development for Kinetic Analysis Corporation which develops and runs hurricane damage models. Watson was a source for this story ran on Nola.com that I blogged about several weeks back. Below Watson comments on how his team makes their predictions, the results of botched forecasts by agencies in 2006, the relationship between forecasters, the media and emergency managers, and other things.
__________________________________

The Chicory: You said in your presentation that you weren’t concerned with how many hurricanes would occur in a given year, but only about the likelihood of them hitting land. In everyday speak, what do you take into account when making your predictions?

Chuck Watson: It is important to realize that even if you have perfect knowledge about the number of storms, it doesn’t tell you anything about the number of landfalls or how bad they will be – in scientific terms, there is a very poor correlation between number of storms and landfalls. Example: by any measure, 1992 was a quiet year. Few storms, only one strong one – but that one storm was Andrew! Likewise, there have been years with 20 or more storms and not a single US landfall. Our methodology computes the probability of exceeding a given wind speed (not just hurricane force winds, but any arbitrary wind speed) at any given location (not just over land, but in any 4km square grid cell). Basically, we start with a climate model spun up using the previous year’s data, then run 100 simulations starting with 1 Jan through May 1st or so, allowing the model to vary within the observational uncertainty of the data from that period, then running to the end of the year. That gives us 100 possible scenarios for the upcoming year. We analyze those 100 scenarios using the same statistical tools that we use to study the 135 years of fairly good hurricane history (1871 – present). That gives us the probability of winds for this year, as compared to history.

———-

The Chicory: What happened with the predictions in 2006 and how have forecasters adapted their predictions for this and following years?

Chuck Watson: We knew 2006 was going to be a transitional year from “neutral” to “El Nino”, but not when or how dramatically. El Nino’s kill hurricane activity by causing unfavorable winds over the Atlantic; La Nina’s have the opposite effect. The year transitioned in August, stomping on the heart of hurricane season. Pretty much everybody missed this.

When analyzing the results from 2006, we noticed that our models did a fairly good job forecasting the transition, as well as activity levels during the season on a week by week basis. In fact much better than it did when the results were rolled up to make an annual prediction. We checked this against history, and found we could significantly improve our models by using a moving +/- 15 day window then integrating the results over the season, rather than trying to treat the season as a monolith. I don’t know how the other groups changed things.

———-

They prefer that we, in the words of one irate caller from a couple years ago, ‘either tell people they might get hit by a storm this year or shut the f*** up, ’cause people won’t prepare otherwise’.

The Chicory: You said in your comment that emergency managers “always want people scared” and the media doesn’t always publish below average predictions. Does this mean authorities pressure you to deliver sensational numbers? Does it effect the manner in which researchers study and predict storms?

Chuck Watson: Huge question. I don’t think pressure from emergency managers and other sources directly impacts the research itself (although it does impact who gets funding to an extent), but it has a big, big, impact on forecasts and the way they are reported, both seasonally and operationally.

Emergency managers have a tough job. They are always pushing hurricane awareness, especially at the beginning of the season. Calling for a quiet season means less press coverage, and less scary coverage, at exactly the same time they are trying to get people to start thinking about evacuation plans and preparedness. So they get touchy when we say a below normal season for their jurisdiction because they perceive it as making their job harder. They prefer that we, in the words of one irate caller from a couple years ago, “either tell people they might get hit by a storm this year or shut the f*** up, ’cause people won’t prepare otherwise”. I (obviously) disagree with that attitude – if you treat people like idiots, generally they don’t disappoint you. I think if you explain the risks and the benefits of mitigation and preparedness, without the scare tactics, most people will react accordingly.

They prefer that we, in the words of one irate caller from a couple years ago, ‘either tell people they might get hit by a storm this year or shut the f*** up, ’cause people won’t prepare otherwise’.

I always tell people that it only takes one storm to ruin your day, and even if our odds for a hurricane in your county are half of normal, say 1 in 100, that’s still pretty big odds you will lose your roof. Sometimes that message gets lost in the technical discussions about the forecasts, but it’s not because I’m not saying it.

They prefer that we, in the words of one irate caller from a couple years ago, ‘either tell people they might get hit by a storm this year or shut the f*** up, ’cause people won’t prepare otherwise’.

Operationally, the hype from the media and pressure from emergency managers is intense. NHC sometimes uses what they call the “forecast of least regret” (their words). For example, if the storm is forecast to brush the coast, they tend to show it making landfall, making a direct hit on a major city rather than an adjacent lower populated area, or call for the winds to be higher than either the models or unbiased forecasting would indicate. NHC has reportedly changed tracks at the behest of emergency managers to make them “scarier” and encourage people to evacuate, especially for high risk areas like the Florida Keys. I think this is a bad idea. The forecast should be the best possible rendition of where the storm is going and how strong it will be when it gets there. Fudging the tracks and, more typically, the intensities, tends to decrease the credibility of the forecasts and over time is counterproductive.

As I said, EM’s have a tough job. It requires at least some knowledge of a wide variety of fields – hazmat, natural disasters, terrorism, etc. Quite frankly, most county level EM’s (as well as many state level and even an unfortunate number of Feds) are not up to the job, and even the ones that have the knowledge often have a bad attitude. They normally come from a law enforcement background, and many do not communicate with the public well in a persuasive role. Their attitude is that they are the experts and you need to do what they tell you, no questions asked. They really aren’t that interested in the media or anyone else reporting objectively – they want to control the flow of information to reinforce their directives. But it’s up to the emergency manager to get people to take precautions based on the facts in hand, not the job of the forecaster or scientist to shade those facts to support that decision. Our job is to crunch the numbers and give an honest assessment of the risks. Otherwise, we risk our own long term credibility which, given the nature of forecasting, tough enough to maintain as it is.

———-

The Chicory: Many people suggest busier hurricane seasons are due to global warming. While I am convinced global warming is happening, I’m not so quick to blame 2004 and 2005 on it. What’s your take?

Chuck Watson: I’m one of the scientific peer reviewers for the IPCC reports, so I’ve reviewed most of the publications in this area (as well as published a couple of papers in the field – just finished a book in a chapter on climate change and disasters that Cambridge University is publishing). As you indicate, any one or two year’s doesn’t tell you much. The state of the science is this: the impact of global warming (which itself is fairly well established at this point) on hurricanes is still open, although there are some very interesting results published recently. The modeling is getting better, as are the analyses of the historical record. In the next 5-10 years we should have better answers. My personal view is that in the short run we may see storms be slightly stronger, and last slightly longer, although the total numbers may not change much from historical. In the long run, increased stability in the tropics may even push the numbers down (sort of like what happens in El Nino years), or just relocate activity north 5 or 10 degrees in lattitude. For what it’s worth, our models tend to show the slight increase and northward shift. The simple fact is that we are doing a vast experiment with the earth – changing both the composition of the atmosphere and the characteristics of the surface. It’s a subtle system, and we really don’t know how it’s going to turn out.

———-

The Chicory: I read about Project Stormfury on Wikipedia, It sounds crazy. What were it’s flaws and in what kind of regard is it held in among researchers?

Chuck Watson:It was a very interesting project, and the knowledge gained was probably worth the expense even though it apparently didn’t do what it set out to do. It was terminated before it really demonstrated any results. I
think that it wasn’t as theoretically unsound as some seem to (especially at HRD today). On the other hand, I think that Stormfury ran a real risk of making storms *worse* by either accelerating rainfall (a lot of hurricane damage is caused by rainfall) or triggering intensification via an unfavorable eye replacement cycle.

Stormfury was a great example of the 1960’s mega-engineering approach to projects – and it is sad that today we don’t think big any more. And if you do, chances are somebody will sue you over it. I started out majoring in aerospace engineering in ’79, and got out of it largely because there just wasn’t the potential to work on grand projects I grew up with in the 60’s and early ’70s like Apollo. Was a heartbreaking but probably sound decision, given what has happened to NASA and space exploration.

Given our increased knowledge and modeling ability, we could probably do some interesting weather manipulation, especially with severe weather. For example, using satellite lasers we could potentially heat up cloud tops during severe convective events and reduce their intensity. Now that I think about it, maybe we could use sharks with lasers on their heads?

On the other hand, maybe we’re not smart enough or responsible to mess with the climate system, of which hurricanes are an important part . . . see the answer to question 4!