Thursday, October 27, 2011

Chance of Bangkok flooding 0 to 50%: On the importance of confident language

When we checked in at hotel reception, a printed summary of the news was laid out, titled "Inner Bangkok 100% safe" from floods. It seemed odd, considering that we just stepped over a line of sandbags in the lobby.

The welcome sign at the hotel. They doth protest too much, methinks.
Sandbags, covered in tarps held down with bricks in the lobby.

Language in river forecasts is very important and there is a constant tension between sounding an alarm that is confident enough to inspire action, while considering all the possibilities of what could happen.

The above quote came from headlines about a week ago, but buried deeper in the newspapers were caveats about how it was for the part of the inner city protected by flood walls. Also, the only way an inner city flood could happen is if there was heavy rainfall coming on top of the floodwave already headed downstream. Naturally one would have to ask, what's the chance of that heavy rainfall happening? 1%? 5%? 25%? Unless it's zero, then I think this means the city is not 100% safe.

Of course the possibility of flood depends not only on nature, but also what people do, adding another layer of uncertainty “I want to beg people, please don’t destroy the levies,” [the Prime Minister] told reporters today. “If you do, it will create a bigger impact and be harder to control.”

A few days later, people weren't as confident about the flood situation..."Prime Minister warned yesterday that there was a 50/50 chance that much of Bangkok would be inundated in the coming days but insisted that Suvarnabhumi Airport would not be flooded. The prime minister said new factors were compounding the seriousness of the situation and it would be hard to fend off all the flood water from Bangkok....Yingluck pleaded for understanding from all parties, saying that nobody had predicted the mass of water would be this big and that it would travel so fast." As an aside, I can say that the Forecaster's Curse is that there is always somebody somewhere willing to say that they did predict things would be as bad as they were, it just may take a while to find that person after the fact. 

Another source says "Brace yourselves for a possible month-long flood as the northern and central region floodwater run-off was draining into the Gulf Of Thailand, via the capital... [The prime minister] noted the flood was inevitable despite government attempts to divert the floodwater run-off to eastern Bangkok."

Most recently, the message is "Floods will hit every area of Bangkok but each area will see different levels of water," director of the Flood Relief Operations Centre (FROC) Pol General Pracha Promnok said yesterday. Pracha... urged Bangkok residents to adjust to the situation and accept what was going to happen.... An informed source at the BMA said the flood-water level would range between 30 centimetres and 1.5 metres across 50 districts of Bangkok.... "The inundation will last for about one month," 

To put a finer point on it, one source translates a story in Thai as "[Prime minister] admitted that the whole of Bangkok will be hard hit by a severe flood." 

Many of the quotes above are from the media and not necessarily directly from the government officials. Indeed, the language of government officials may not be the same as what the river forecasters originally told them. Therefore, I can't say anything about how well the communication process is currently working in Bangkok. Indeed, Thailand has a strong reputation as one of the best and most sophisticated river forecasting systems in Asia.

I will say that among forecasters I knew in the US, there was often a fear of being perceived as a waffler or a flip-flopper. "First you said it would flood, then you said it wouldn't...Which is it? Where is your credibility?" In politics especially, changing one's position can have a surprisingly devastating effect on one's career. It seems to show a lack of conviction.

During recent US presidential elections, protesters would carry flip flop shoes as a sign that they didn't want to elect someone that changed their position on issues. (image source)
However, in an excellent essay on political vs scientific flip-flopping Lewis Eigen says

"To the scientist, failure to flip-flop in the face of contradictory evidence is irrational and dangerous behavior.  And scientists will often flip-flop at almost light speed.  Often it takes only a single observation to flip-flop thousands of scientifically learned people."

The essay quotes John Maynard Keynes as responding to a critic: "If the facts change, I'll change my opinion. What do you do, Sir?"

I leaned more towards the scientist's viewpoint, that it is a forecaster's imperative and duty to update information if the situation changes. However, it was hard for me to back out of situations where I started off with words like "100% chance", or "inevitable". Even conditional statements like "Drought will occur if the spring rains are low" ended up being a dare for mother nature to embarrass me. In some cases, the spring rains were low and drought didn't happen, as well as vice versus. I always tried to think of a set of forecasts over time as being shaped like wavy wizards hat..starting off with wide bounds (i.e. high uncertainty) at the beginning and narrowing down to a point (i.e. what actually happened), even though it may sway back and forth between the brim and the tip.
Wizard's hat (image source)

Change of river forecast vs time. The longest leadtime forecasts with the most uncertainty are on the left (e.g. 80% chance of between 70 and 400), and the more certain shorter leadtime forecasts are on the right (e.g. 80% chance of between 50 and 140). Learn more about these forecasts here.

The problem for hydrologists is that forecasting is a mix of science and politics, where the preparations have real costs and the outcomes can have serious consequences. Forecasters have struggled with this for a long time and there's no easy answers to the challenge of having calm assertive confidence and leaving room for the inevitably unexpected.

No comments:

Post a Comment