Wednesday, July 23, 2008

Wetlands for hurricane protection might make cents

A study published in the June 2008 journal of the Royal Swedish Academy of Sciences attempts to calculate the economic value of wetlands as hurricane and storm surge defense features. I first heard of the study online and was fortunate to find the full text of the journal article here.

It is first important to note what this paper is NOT. It is not an engineering analysis; it is an exercise in econometrics. It does not propose any new design methods, nor does it test any of the natural mechanisms by which wetlands can provide storm protection for human populated areas. Early on the authors acknowledge this fact and note that the best empirical studies they find place the storm surge reduction benefit of wetlands at about 3 inches per mile. Regular readers of this blog will likely recognize that factoid.

What the researchers have done is gathered a ream of existing hurricane data, applied some simplifying assumptions to fill data gaps, and crunched the numbers through a series of statistical tests. The result, the authors can claim with some credibility, is that the value of wetlands as hurricane storm barriers can be expressed in dollars.

This is not a new problem. Economists have struggled to quantify the monetary value of natural features for decades. The vexing part of the problem is how to properly and fully equate wild lands and habitats in economic units of dollars used to store and trade wealth in human commerce.

Take any forest, for example. What is it worth? Is it simply the market value of its trees cut down and shipped to the mill plus the going rate for its developed acreage? We know this is not the right answer because forests have ecological value as carbon storage centers and aesthetic value as places humans enjoy seeing in person and in pictures.

Government agencies have for this reason studiously avoided placing economic value on wild lands. If forced to complete in the market of capitalist values, naturally occurring features will almost always lose out to commercial development. Environmental scientists instead have been using other more nature-based measures such as the Wetland Value Assessment and assigning "Habitat Units" in lieu of pure dollars to compare alternative land uses.

The authors of the Royal Swedish study set out to define the relationship between raw acreage of natural features as storm defense systems and the avoided damage from those same storms. First, they obtained a data set of hurricanes that struck the US Atlantic and Gulf coasts since 1980. This data included storm path, intensity and estimates of damage. They found a wealth of scientific and economic data in the Emergency Events Database ( created and maintained by the World Health Organization, the Centre for Research on the Epidemiology of Disasters, and the Belgian Government.

Next, they needed to calculate the total possible damage of strike locations. By comparing the experienced damage with the theoretical maximum possible damage, the researchers aimed to quantify the value of avoided damage. And this is where the econometric gymnastics kicks in. The authors took statistical data on US and individual state gross domestic product (GDP) and assigned a GDP value to a map of coastal America on a grid resolution of 1 kilometer square. Because GDP data is not readily available in this resolution, they interpolated and assigned GDP value by looking at satellite photographs taken at night. Areas brightly lit were given higher GDP values than dim or dark areas of the country.

Armed with this data, the authors ran the numbers and found some statistically significant relationships. Their primary discovery was that damage in areas near wetlands was reduced proportionally with the quantity of wetlands. Everyone has long acknowledged this fact, but here was proof--expressed in dollars.

The researches went further and attempted to calculate the average annual benefit of wetlands as storm protection. Their conclusion: wetlands save America from billions in storm damage year in and year out.

But let's not get too far ahead here. I have some serious misgivings about the methods employed and with the way the results were tabulated and presented. I will elaborate on some of them here.

1. It is interesting to note that the methods employed here failed to provide satisfactory results in several key instances. For instance, researchers had to limit their analysis to only the US Atlantic and Gulf coasts because of the scarcity of data. They admit to wishing they could apply their skills to the thousands of known disastrous tsunamis, monsoons and other coastal storm events, but found insufficient data to do so. The resulting data set includes just 34 hurricanes out of the nearly 270 major storms to strike the US during the selected period.

2. The researchers initially set out quantify the storm barrier benefits of forested wetlands, too, but found the statistical correlation not significant enough to support their hypothesis. Several widely published papers have touted the benefits of mangrove and cypress swamps in knocking down storm effects. The authors of this study could not validate that claim, and it makes me wonder why they proceeded with their work in spite of that glaring problem.

3. The use of GDP as a measure of potential storm damage leaves much to be desired. GDP is an expression of economic productivity; it says nothing about the present value of homes, businesses and other infrastructure that fall victim to nature's fury. Likewise, the Emergency Events Database records the total estimated damage wrought by storms. This would include wind and rain damages, but it is primarily the storm surge damage that is thought to be muted by wetlands.

4. To convert their calculation of avoided damage into an annually recurring value, the authors developed hurricane return frequencies using a data set with a period of just 25 years. This is hardly a large enough sample to extrapolate reliable return frequencies.

5. The authors conclude their paper with an editorial on the virtues of wetlands far beyond the apparent correlation to protection of human economic activity. They opine how coastal wetlands are "maintained by nature" and that such features are "far more cost-effective than constructed levees." But there is absolutely nothing in this article to support these conclusions. Indeed, the authors themselves note that Hurricane Katrina alone destroyed some 50,000 acres of wetlands.

These are just a few of the shortcomings of the study.

It is also important to note that the study tells us nothing about how and where to build wetlands as part of a storm defense system. The study considers gross acreage in the path of hurricanes assumed to be 100-km wide; nowhere does the study enlighten us as to how we would go about designing a wetlands plan.

I was disappointed that the researchers included the rather colloquial analogy that wetlands function as "horizontal levees." In fact, all physical features exist in three dimensions: earthen levees and wetlands both have length, width and height. The difference is the primary storm defense benefit of levees lies in their height, while the frictional resistance of wetlands is manifest in their area--length and width. But to equate the two, as if to suggest one could substitute some acreage of wetlands for some or all of a levee is ludicrous--if not dangerous.

It is noted that as I was preparing this blog entry, The Times-Picayune ran a story on the study that seized upon the expression "horizontal levees" as if it were a valid scientific result of this study. It is not.

I think the study represents a remarkable effort to quantify the hurricane defense benefits of wetlands in dollars--not an easy task--even if the results seem exaggerated to my eye. As an engineer, I find little in the study to help me understand and design these vital systems. But perhaps the study gives our society the justification and incentive to take wetland preservation and construction seriously as a bonafide contributor in our overall strategy to protect coastal developments. In the end, much research remains to be done before this general appreciation of wetlands can become a plan of action with predictable results.


Schroeder said...

Ya bastard. I'm gonna have to print this to read and understand it. Nice work.

Mark Folse said...

It would be intersting to see the same techniques applied to the loss of inland wetlands and their relationship as water retention features for inland flooding. It's not just a coastal issue, as I learned when I lived in pancake-flat Fargo behind 40 foot dikes and learned about the wonder of overland flooding (when rivers get so backed up sheets of water start looking for another path.)

Mark Folse said...

After a more careful reading of just the post (not the study) whether their questionable approach to GDP is really that important. Wouldn't a simple comparison of damage estimates between storms in coastal zones with and without buffer zones been as valid based on known Census data (as simple as population, but it is probably possible to make some reasonable assignments by zip code, for example, from Census data. Some correlation of the census tracts or zip codes in the storm path, comparing the "wealth" (for lack of a better word) or density to reported damage estimates would seem more valid that looking at Google Maps.

Tim said...

The reason they used GDP was to estimate the value of what was NOT damaged. There are lots of estimates of what was damaged; they started with GDP and subtracted the estimated damage and that is what they used as avoided damages. I don't know why they chose illuminated satellite photos to distribute the GDP--seems to me you'd be better off using census data as you suggest. But then again, I'm not an econometric researcher.