Tuesday, July 29, 2008

Stationarity: Challenging Science with Science

A friend of mine sent me an interesting paper from a February issue of Science titled, "Stationarity Is Dead: Whither Water Management?" by Milly, P. et. al (2008) (subscription-only). It deals with the assumption of stationarity, or the "idea that natural systems fluctuate within an unchanging envelope of variability." In other words, natural occurrences, like stream flow or flood peaks, change within static extremes that can be estimated using some probability density gleaned from observation records.

Policy decisions regarding water management (i.e. water infrastructure, channel modification, and drainage works) have been impeded by this assumption because anthropogenic forces, such as land-use changes and climate change have added additional variability to the dynamics of the water cycle. Modeling efforts have historically assumed that these variability are small enough to assume stationarity, leading to ever less accurate results.

To address this issue, the paper states that stationarity must be replaced by non-probabilistic models that incorporate operations research and welfare economics. Yet, while this is a normal "scientific" response to a problem, is it enough?

It is assumed that better models produce better results, but it ultimately comes down to how those results are used. The paper hints at this by stating, "a stable institutional platform for climate-information delivery may help." It can be argued that many of the issues facing the US could be better modeled - and they can and it will help - but what of what happens after? If stationarity is a baseline assumption that is impeding researchers and water management experts, what is impeding emergency managers and policy makers?

What other complimentary efforts should be implemented along with better models? Are scientists suited to expand outside of producing results, into utilizing (policy making, for instance) those results? Does science have other avenues to create more sound, rational policy decisions?

Image: 2008 Iowa floods, picture taken from BBC website.


Chris McClure aka Panhandle Poet said...

"If stationarity is a baseline assumption that is impeding researchers and water management experts, what is impeding emergency managers and policy makers?"

I think the answer to this question is money. Within the beheamoth that is our government and all of the various agencies, departments, etc. there is competition for every dollar spent. We ask politicians to prioritize those expenditures through the budget bills that come before them. Long-term infrastructure issues tend to get short-shrift in the process because of an emphasis on the urgent vs. the important (classic Franklin Time Management principles -- Stephen Covey). That's why levies deteriorate, bridges deteriorate, highways deterioriate, etc.

Sue said...

I agree with Chris about the issue of budgets as an impediment to the "emergency managers and policy makers."

I had never heard the term "stationarity" before, although the concept that it refers to is familiar, if I understand your post correctly. Would I be correct in assuming that the idea of "stationarity" is behind the notion of "100 year floods" -- the idea that certain levels of event only occur at certain intervals? And that the demise of "stationarity" is due to the fact that events such as so called "100 year floods" appear to be happening a lot more often than that time line?

I can see why models would need to include anthropogenic factors in them, but could you expand a bit on why "welfare economics" would be one of the factors that should be included in the models?

E. R. Dunhill said...

Say, you look a lot like a guy who used to write for this blog... Welcome back.
This reminds me a little of a discussion I was party to a while back about extinctions. The parallel is that some extinctions are perfectly natural, which raises the question as to whether or not conservation policy should be mindful of this fact. I think we do need to recognize that natural systems are moving targets. All of the bits that make them up are constantly moving and changing, influencing each other and the whole, and humans are powerless to impact certain parts of these changes. Perhaps this understanding is an even greater charge to be good stewards of those parts we do indeed control.
Chris M also raises a good point about the unholy master of all things governmental: Money. The federal fiscal machine has a number of fundamental flaws. The one that irks me the most (and I dare say should irk every tax-payer), is that efficiency is punished by the budgeting process. An agency's worth and its officals' esteem are measured by the size of its budget. And, if an agency finds ways to reduce costs as compared to budget, the agency finds its budget in the next FY reduced to reflect the new, lower fiscal need. A common result is that when offices find that they have saved costs, they find something else to spend the money on to avoid coming up short. Woe to those agencies that save one year, but actually need the money again the next year.
The truth is that we have some money to invest in infrastructure, but the political nature of funding means that it's difficult to move that money to where it's needed.