Flash flooding is one of the costliest weather hazards, both in economic losses and in human casualties. Research showing that high intensity rainfall will be more frequent in our current trend of climate change seems to imply that flash flooding will also become more frequent. However, because flooding results from complex interactions of meteorological and hydrological processes, it is not obvious what will be the true trends in flash flooding. It is well established that flash flooding results from the combination of the spatio-temporal distribution of rainfall with antecedent soil moisture and basin-scale hydrological variables. Our recent work suggests that storm runoff in the northeastern US is more often a product of total rainfall accumulation (saturation excess) rather than a product of high instantaneous rainfall rates (Hortonian runoff). In this project we will evaluate whether this is true for runoff that contributes specifically to flash floods by exploring the fundamental processes that link the atmospheric processes with surface-hydrological processes for the northeast U.S. The primary research questions are: (Q1) Are northeastern flash-flood-producing storm events substantially different from those that have been characterized for the midwest and southeast? (Q2) Are northeastern flash floods more often the result of saturation excess or Hortonian storm runoff processes? We have identified 186 flash flood events from 2003-2007 throughout the northeastern U.S that we will use as our sample set. There are fewer gaps in the radar data during this period than in previous years and the relatively high frequency of flooding events during this period may be typical under climatic conditions expected during this century. Q1 will be addressed by characterizing observations of both synoptic-scale (1000s km) and mesoscale (10-100s km) structures of storm events associated with these flash flood events, as well as the resultant watershed-scale rainfall amounts and intensities based on radar (using standard or tropical reflectivity-rainfall relationships as conditions warrant) and ground station (point-scale) data. These data will be compared with published information about flash flood storms in the Midwest and Southeast, which is more plentiful than that for the Northeast. Because distributed hydrologic information like soil moisture is not widely measured, to address Q2, we will estimate this information with an ensemble of four watershed models ranging from fully distributed/fully mechanistic to semi-distributed/empirical that have been well vetted with regional hydrologic observations of both discharge and soil moisture. We will use bias-corrected NEXRAD rainfall data to run our watershed models. Once we ?know? our antecedent conditions (from our watershed model results) and the flash-flood-causing rainfall (from Q1), we can use bivariate probability analysis to determine the probability of the coincidence of watershed wetness and rainfall needed to produce a flash flood. By analyzing how often flash-flood-generating rain intensities produce Hortonian runoff, i.e., exceed soil infiltration capacity, which is a function of soil moisture, we can determine how frequently flash floods are ?Hortonian.? We will use an ensemble of four watershed models with different mechanistic and empirical underpinnings in order to bracket our results; note: this approach is taken because there is little consensus about which approach, mechanistic vs empirical, provides the ?best? result. This work will foster meaningful collaboration between atmospheric and hydrologic scientists to help put climate change predictions in context as to how they are likely to influence flood frequency. Results will be shared with National Weather Service Forecast Offices ideally resulting in more timely and spatially precise flash flood warnings. Graduate students will conduct much of the research, contributing to their training as scientists. We are designing a Freshman Writing course around society-science-policy interactions in the context of global climate change that uses the hydrometeorology of flash flooding as a consistent theme or case study, and will develop course modules to incorporate the findings and tools from this study into other science and engineering courses.
This project investigated atmospheric and surface landscape conditions associated with recent flash flooding events over the northeastern United States. Flash floods are a natural hazard that can impact lives and property, but they have not been studied as thoroughly over the northeastern United States as elsewhere. The project identified patterns in radar imagery that can alert weather forecasters to the likelihood of flash flooding. Flooding weather systems were found to be often larger in size than those observed in other parts of the country. The study also provided scientific explanations for the existence of these patterns. Notably, the flooding weather system was typically located at the end of a plume of warm, moist air. Landscapes that were relatively impervious and capable of rapid runoff were found to be more likely to flood under the different storm types than other landscapes. This finding can help identify areas that are more susceptible to flooding than others. A surpising result was that some floods were associated with unremarkable rains and runoffs, possibly due to undersized culverts built to divert the flooding rains. Two students experienced scientific training while working on the project. They conducted research, wrote scientific reports, and presented findings in formal and informal settings. One student is now employed as a college teacher while the other is completing her Ph.D., anticipating future employment.
