Letters to the Editor

April 2006 » Letters
Report missing content

Rainfall probabilities

I read with interest ["Researchers predict dramatic climate changes and infrastructure failures (December 2005, page 11)] concerning the October 2005 rain storm in Keene, N.H. which was described as being a 1-in-100,000 year event with 10 inches of rainfall. Perhaps it was a typo and should have read 1-in-100 year or 1-in-1,000 year event? By example, the Oct. 14-15, 2005, rain storm dumped almost 11 inches of rainfall over a 48-hour period in areas of Burlington and Canton, Conn., as recorded by a U.S. Geological Survey (USGS) rain gage. The runoff for Burlington Brook in Burlington, Conn., exceeded the prior gaging station record from the August 1955 flood caused by two hurricanes within a month of one another. Tentatively, the USGS is saying that the flood was 1-in-500+/- event. Kenneth Wassall,P.E., L.S.

via e-mail [The CE News] article does not contain any errors, we stand by our conclusion that the October storm is the equivalent of a 1- in-100,000 year event. Numbers this large, of course, do not reflect actual return periods (we’ve had two ice ages during the last 100,000 years!), but rather probabilities of occurrence based on the historical record.

And assigning probabilities to events as large as the October storm is somewhat academic due to the large amount of leverage involved in projecting to the extreme regions of the probability curve. However, I think Mr. Wassall can agree with us that these storms are extremely rare events. Please note that for Keene, N.H., the October 11.4 inches of rain was 2.6 times the next largest 24-hour rainfall event on record, 4.37 inches, which fell on Sept. 17, 1999.

We believe there is a simple explanation for the apparent disparity between the tentative USGS conclusion of a 1-in-500 year return period for the Connecticut storm and our result for Keene. Mr. Wassall reports that the Connecticut storm dropped it’s rainfall over a 48-hour period, whereas the Keene rainfall occurred during a 24-hour period.

Please note that others in the Keene area recorded rainfall amounts similar to ours (we checked before going public), confirming the accuracy of our instrumentation. In addition, there is about 100 miles between Canton, Conn., and Keene, N.H., so there is no reason to expect that Keene’s rainfall should have equaled Canton’s.

For the curious, here’s a condensed version of our methodology. Our source for historical rainfall data was the NCDC Summary of the Day dataset from NOAA. For Keene, this represents 76 years of records from 1926 through 2004 (most data for 2002-2003 is missing). We used the Peaks-over-Threshold method to establish the intensity-frequency relationship; 1.5 inches was selected for the threshold because at this value confidence intervals were reasonably tight and scale and shape parameters were stable. The set of rainfall events greater than the threshold was fit to the Generalized Pareto Distribution (GPD) using the Maximum Likelihood method to estimate scale and shape parameters.

The GPD provided a good fit to Keene’s data. With a good fit having been obtained, it was a simple matter to plug the resulting scale and shape parameters, along with the 11.4-inch rainfall value, into the GPD equation to derive the probability of occurrence.

This probability is actually 8.780184e-6, or an 8.8 chance in a million years. We rounded this up to 1-in-100,000 years. Please note that for the culvert capacity study (as opposed to the analysis of the October storm), we regionalized the analysis, incorporating data from NCDC weather stations proximate to Keene to average out any anomalies in Keene’s historical record.

Taken in isolation, the October storm cannot be attributed to climate change.

However, it is a pointer to what the climate models predict. Our study echoes, in a tangible way, the findings of previous research: There will likely be a fundamental change in the coefficients of the [risk = exposure x (probability of occurrence)] equation. The design-storm for common culverts in New Hampshire has historically been the 1-in-25 year event. Our work shows that for Keene, what had been a 1-in-25-year event will become a 1-in-5-year event, and what had been between a 1-in- 500 year and 1-in-1,000 year event will become the new 1-in-25 year event. Either communities will be assuming the higher degree of risk associated with the compression of the risk curve, or culverts must be upgraded to maintain the hist Infrastructure elements that will still be in use as mid-century approaches should be spec’d out according to carefully and conservatively developed studies of future design storms. Although there remains uncertainty in predictions of climate-change impacts, this should not deter preparation. A core facet of leadership is the necessity of decisionmaking under conditions of uncertainty.

The "unknowns" inherent in projections of the long-term cost of capital, population growth, and the vicissitudes of both public opinion and the regulatory environment are routinely assimilated in the design of municipal infrastructures. Without a tangible quantification of vulnerabilities and remedial costs, decision-makers have no means by which to incorporate climate change preparation into plans and budgets.

— Latham Stack and Michael Simpson Antioch
New England Graduate School, Keene, N.H.


A noble profession

I really appreciated Geomatics World View [in February] ("Statesman-surveyors," page 16). I had never thought about Mount Rushmore including "three surveyors and the other guy." My son is taking a job with a firm as a "rod man." Although I consider it a less than desirable job for him (he has a college finance degree), this will give me a more positive perspective to encourage him about the noble history of land surveying. Thanks again.

Dennis W.Lang, P.E., P.L.S.

Fort Worth, Texas


When I taught surveying, even though most of my students were in civil engineering, agricultural engineering, and forestry, I was at first surprised and later quite proud of the fact that students from all over the campus came to take my courses. A lot of the time, the students had a motive such as, "My father owns a farm and he said it would be helpful if I knew how to survey so that we could grade the fields properly." But the one you and your son may appreciate is, "One of the highest-paid unskilled job areas for a college student to take in the summer is construction, and if you have a skill like surveying, then the pay is even higher. So, I’m expecting this course to help keep my college loans down." Perhaps as a student of finance, your son’s already figured out the latter.

— Joseph V.R.Paiva, Ph.D., P.S., P.E

Upcoming Events

See All Upcoming Events