Research Idea Details

Improving Transportation Infrastructure Design to Accommodate Climate Change

Research Idea Scope

In recent years, northern New England has experienced several unusually large storm events, such as the Mother’s Day storm of 2006 and the Patriot’s Day storm of 2007. These events resulted in extreme flooding across the region and caused great material losses, collapse of lifeline infrastructure, and the breakdown of public health services among other things.  Many culverts and small bridges succumbed to the intense floods, indicating that past design guidelines may now be obsolete, especially considering the likely increased flooding due to future climate change.  Transportation structures are typically designed to pass a flood event associated with a specified probability of exceedance, such as the so-called 25-year or 100-year flood, which has a theoretical 4% and 1% chance, respectively, of being equaled or exceeded in any given year. Traditionally, classical frequency analysis, which assumes that climatic processes remain constant over time, has been used to estimate the magnitude and frequency of these design events.  The standard of practice for design storm selection, Technical Paper 40 (TP-40) was published by the National Weather Service in 1961, and used precipitation time series through 1958 for design storm estimation.  An update to TP-40 for the northeastern US was published by the Northeast Regional Climate Center (NRCC) in 1993.  Both reports use a frequency analysis approach, but design storm estimates from these two reports are quite different.  For example, the 100-yr, 24-hr storm for Boston is approximately 6.5 inches according to TP-40 and approximately 8 inches according to the NRCC report.  Two important questions that can be raised from the differences in these estimates are: 1)  Is the difference in design storm estimates due to differences in the statistical models used or is it due to changes in the frequency of extreme events over the different periods of record?; and 2)  How can our method for estimating design quantiles be improved to accommodate non stationarity and projected climate change? To address the first question, we analyzed annual maximum precipitation depths extracted from 48 stations in Maine, New Hampshire and Massachusetts that had relatively continuous daily precipitation records from 1954 to 2005.  We evaluated the presence of trends over two time frames (1954-2005 and 1970-2005) using two statistical methods (linear regression and a modification of the Mann-Kendall trend test) and at two scales (at-site and regional).  Five factor groups (regions) were delineated using a factor analysis of the extreme precipitation time series, and four of them were spatially coherent, although the physical reasons for these factors is not yet fully understood.  The results of trend tests with p-values less than 0.05 were defined as highly significant, while trend test results with p-values less 0.10 were defined as significant.  At-site analyses showed highly significant trends at about 25% of the sites, but that the magnitude of these trends was small.  We did find evidence of increasing extreme precipitation depths after 1970.  The regional analysis showed a highly significant trend in extreme precipitation depths in the lower Connecticut River Valley, but the magnitude was also small.  Estimation of the 100-year storm using the Generalized Extreme Value function (in the same family as the Gumbel function used in TP-40) yielded values in the same range as the TP-40 estimates for northern New England.  Hence, this analysis suggests that observed trends are not large enough to have affected the design quantiles based on the TP-40 methodology. We are currently in the process of investigating different statistical distributions for northern New England annual maximum time series and also methods that can accommodate nonstationarity in the time series properties.  There have been numerous methods reported in the literature over the last decade that can account for climate variability and change.  We are also extending our analysis to include southern New England as well. We propose to 1) extend our trend analysis to precipitation time series in southern New England and also investigate the presence of trends flood flows across New England, 2) evaluate the best methods for estimating design quantiles for precipitation and floods that can accommodate climate variability and change, 3) assess the potential change in design storm and design flood estimates under three climate change scenarios (IPCC low emissions, high emissions and moderate emissions scenarios) and 4) hold workshops at selected locations across New England to disseminate our results and to educate engineers, town planners and public works departments on how to address the potential impacts of future climate change.

Urgency and Payoff

Most engineers would agree that TP-40 is out-of-date and needs to be updated, simply in light of the fact that we now have 50 more years of data with which to make our design storm estimates.  In addition, over the past several decades, New England has observed many of the changes in our climate and hydrology that were predicted by climate change models; these changes include increased summer and winter temperatures, decreases in overall snow cover and changes in the timing and magnitude of snow melt and flooding.  All of these changes are predicted to continue by even the most conservative of climate changes scenarios.  The damage costs that have been incurred across New England by recent extreme events attest to the need for improved guidelines for infrastructure design and construction.  In light of the inherent uncertainty in future climate scenarios, there is also a great deal of confusion amongst practitioners and planners alike as to just how to interpret and prepare for the potential impacts of climate change. Our proposed research will not only improve engineering design under climate change but will also educate the people that need to implement these improvements.

Suggested By:
Ellen M Douglas, University of Massachusetts Boston
Submitted:
03/26/2009