The International Association of Hydrological Sciences (IAHS) has started a public discussion in order to shape its new Science Initiative for the next 10 years, to be launched in 2013. The following comment is reposted from the site dedicated to this discussion, coordinated by Alberto Montanari.
Alberto and all IAHS officers,
I congratulate you for the way you shape the new Science Initiative of the IAHS.
I wish to start with a very didactic story which was unfolded just last week. The story is about hurricane predictions and, even though it is not a pure hydrological story, I think it has some relevance to hydrology; the article entitled “Extended Range Forecast of Atlantic Seasonal Hurricane Activity and Landfall Strike Probability for 2012” by Philip J. Klotzbach and William M. Gray (available from http://hurricane.atmos.colostate.edu/Forecasts/ ) is worth reading by hydrologists. Its abstract starts: “We are discontinuing our early December quantitative hurricane forecast for the next year” and continues: “Our early December Atlantic basin seasonal hurricane forecasts of the last 20 years have not shown real-time forecast skill even though the hindcast studies on which they were based had considerable skill”.
These scientists are admirable for their continuous efforts for two decades to tame an important prediction problem, but above all for their integrity to admit that their hypothesis proved wrong. The distinction of “hindcast skill” versus “forecast skill”, or else, model fit on past data versus real predictive capacity for the unknown future, is well recognized in hydrology. Yet we often tend to get fooled by skilful fits and provide profound explanations why the past phenomena happened–but the future will tell us our limitations.
There is nothing wrong if some scientists are working on a hypothesis that eventually will prove wrong or on a direction that will prove to be a cul-de-sac. However, it is a problem if the entire community is headed for such a direction. Therefore, a Science Initiative for hydrological community is not an easy task. Over-certainties about which is the right direction, or which core questions should be addressed, may be dangerous. Perhaps a diversity of directions and core questions is the antidote for getting trapped in deadlocks.
Two decades ago, the aspiration of deterministic predictability via a reductionist approach stimulated and even excited the hydrological community. One decade ago, the direction toward a radical reduction of uncertainty of hydrological predictions, even without data, was regarded to be a realistic target. Even today, many of us believe that it is a matter of one decade more or so to eliminate uncertainty and render useless any probabilistic/statistical/stochastic approach aimed to quantify uncertainty. Many of us believe that uncertainty can be, in principle, eliminated. Fortunately, however, this will never happen–that is my strong conviction. If it could happen, a world without uncertainty would not be livable (because predictability entails controllability) and fun (I cannot imagine anything fun if everything is expected).
Two centuries ago, Sadi Carnot determined an upper limit to the efficiency of heat machines. After that time, a hypothetical machine exceeding this limit is called a perpetual motion machine of the second kind and is not part of science and engineering. This is because it violates the Second Law of Thermodynamics, which is related to the tendency of entropy to become maximal. Today, thanks to Boltzmann, Gibbs, Shannon and others, we know that entropy is uncertainty quantified. Thus, in my view, the aspiration of elimination or radical reduction of uncertainty could be compared to the quest for a perpetual motion machine of the second kind.
Nicolas Léonard Sadi Carnot (1796-1832) in the dress uniform of a student of the École Polytechnique. Source: Wikimedia Commons
Given the complexity of hydrological systems, determination of upper limits for the efficiency of hydrological predictions, is not as easy as in heat machines. But, I think, it is very useful and didactic to demonstrate the existence of limits and quantify these limits at least in some simplified systems.
This is just one problem whose study would be useful from a philosophical and practical point of view–and I believe that both philosophy and practice are very relevant in discussions about shaping a Science Initiative. More generally, I believe that the new Science Initiative of the IAHS, instead of continuing the current trend of exorcizing or ostracizing uncertainty, should embrace and feature it. Trying to re-conciliate hydrology with uncertainty would be beneficial from a scientific point of view, because it would disburden hydrology from deterministic myths, from a social point of view, as it would give a more pragmatic basis in decision making, and from an educational point of view, particularly because it would provide students with important qualifications and skills to handle difficult problems in which naive deterministic approaches fail.