30th October lectures entitled
Is climate change science a barrier to flood
management decision making? by Jon Wicks
What is the risk of drought in the Thames basin? by Jim Hall
Both speakers had used climate change projections to estimate the effect on water flow. Both had been intensely aware of the political interest in their work.
Speaker Jon advised that modelling was useful when it considered the probability of stepping over a critical threshold. First you must know what is the critical threshold e.g. a certain depth of flood water, and this is essentially a political decision.
He added that modellers tend to focus on the things they can actually model and to underplay by their absence, the things they cannot model. For example, an administrative decision might increase the population in a low lying area thus approaching a loss of life threshold even in the absence of climate change.
He presented an estimate that by 2050 the damage bill in Shanghai would increase by a factor of 5 due to asset value increases and 5 due to climate change. Combined effect between 15 and 20 times. When asked if he knew what the error bars were he argued that error bars wouldn’t help politicians to judge the evidence.
Speaker Jim advised that:
- the direction of change in precipitation in the Thames catchment area was uncertain,
- per capita demand change was uncertain and
- population change was uncertain.
- He was quite sure that evaporative loss from the catchment area would increase.
Given this he had concluded that planning to have a safe operating margin for water supply of say 5% was the wrong question. Better to model the probability of a hose-pipe ban. This he did at length by combining natural variability in the present system with the uncertain projections for the key parameters.
In effect he was able to identify those combinations of futures that would lead to high second derivatives in the probability of exceeding the 5% chance of a hose-pipe ban. In this way, managers could focus on preventing those combinations from reaching a tipping point.
He confessed that he too was inclined to model the things he could model whether or not these were a true representation of the system at hand. [he called it epistemic uncertainty].
In effect speaker Jon had concocted a future scenario for Shanghai which gave him the answer he wanted to portray to the politicians. The effect of climate change science uncertainty was of no importance to him, so he had no interest in answering the question in the title of his talk. My proposal to him was that if the climate change loss multiplier had been times 5 but plus or minus times 6 then the politicians would not have accepted his scenario. He had no answer to this as he had no estimate of the uncertainty in his projection.
Speaker Jim had no real need for absolute risk as he was interested in rate of change. This he could model, but admitted that the model could be very different from reality.
As previously advised, my recommendation to all decision-makers is to first define the degree of change with which you can cope and the range around that which is acceptable. Then, when engaging the help of modellers, tell them how precisely you need to know the projected change. If the modellers cannot meet your precision needs, don’t hire them.
In this case speaker Jon had heard how much change would worry politicians and produced a scenario which would trigger their interest. Which corners he had cut to achieve this end was unknown even to him. I was left wondering how common this approach was and whether I minded very much – provided as a result the politicians had a better idea of what would worry them. This could of course be achieved without any help from climate change scientists.
My answer to the question in the title of the first talk is that yes it would be a barrier to flood risk management decision-making if it was presented accurately. The need to make (and influence) decisions overrides the need to be accurate.