A comparison of past small dam removals in highly sediment-impacted systems in the U.S.

Abstract

The ability to predict the effects of dam removal in highly sediment-filled systems is increasingly important as the number of such dam removal cases continues to grow. The cost and potential impacts of dam removal are site-specific and can vary substantially depending on local conditions. Of specific concern in sediment-impacted removals is the volume and rate of reservoir deposit erosion. The complexity and potential accuracy of modeling methods used to forecast the effects of such dam removals vary substantially. Current methods range from predictions based on simple analysis of pre-dam channel geometry to sophisticated data-intensive, three-dimensional numerical models. In the work presented here, we utilize data collected from past dam removals to develop an additional tool for predicting the rate and volume of sediment deposit erosion. Through the analysis of sediment, discharge, deposit, removal timeline, channel, and watershed data, in conjunction with post-removal monitoring data from a wide range of dam removal projects, some significant trends in the evolution of reservoir deposits following dam removal can be seen. Results indicate that parameters such as median grain size, level of cohesion, spatial variability of the deposit, and removal timeline are among the most influential factors in determining the rate and volume of sediment erosion. By comparing local conditions of dams and reservoirs slated for removal with those of past removals, we hope that predictions of the rate and volume of sediment deposit erosion can be usefully constrained.

Introduction

There have been over 700 documented dam removals or decommissions in the past century (Gleick et al., 2009), with upwards of 350 such removals having occurred in the last decade (American Rivers, 2009). With dam removal continuing to increase in popularity and an inventory of over 80,000 registered dams in the United States (FEMA, 2009), there is a definite need for a set of reliable tools capable of predicting the geomorphic effects of the many dam removals to come. The potential geomorphic effects of dam removal are most extreme in situations where reservoirs have been substantially filled with deposited sediments that are then left to erode naturally via streamflow after removal. The fate of these deposited sediments following dam demolition is acknowledged as being the most poorly understood aspect of dam removal projects (Heinz Center, 2002, Stewart and Grant, 2005). There are numerous process-based models, conceptual (Doyle et al., 2002, Pizzuto, 2002) and numerical (Cui et al., 2006a, Cui et al., 2006b, Greimann and Huang, 2006, Langendoen, 2010), that have been developed over the past decade to predict the erosion and evolution of stored reservoir sediments following dam removal. However, comparison of model-generated predictions of deposit evolution following dam removal to actual post-removal monitoring data is quite rare. In studies where such testing of forecasts has been done, comparisons are made with only one or two removal cases. Most published dam removal studies deal specifically with the analysis of data gathered from individual removal projects. In a few instances, multiple removals have been monitored and compared in a single study (Doyle et al., 2003a, Wildman and MacBroom, 2005), and occasionally there has been some comparison of results from a new removal project with those of past removals (Riggsbee et al., 2007). However, in general there appears to be a significant lack of comparative analysis of post-removal monitoring data, particularly of the evolution of highly sediment-filled reservoirs following dam removal. Fortunately, as the number of independent case studies has grown over the past decade, so too has our ability to learn from, and make use of, the numerous data sets that are now available (Stewart and Grant, 2005, Kibler et al., 2010).

Included in the need for a broad comparison of currently available dam removal data sets is the need to test and validate some of the most basic concepts regarding which parameters are most influential in determining rates and volumes of streamflow-driven deposit erosion. Because of concerns over downstream channel and ecological impacts, stabilization of reservoir deposits, new channel development, and other issues (Grant, 2001, Heinz Center, 2002), a major question surrounding natural erosion of stored reservoir sediments is the volume, rate, and timing of their release following dam removal. Relatively accessible variables such as deposit geometry, sediment grain size and level of cohesion, and annual watershed sediment yield have long been suggested to have specific impacts on the erosional processes that determine the rate, volume, and form of deposit evolution (Harbor, 1993, Egan, 2001, Doyle et al., 2002, Pizzuto, 2002, Stewart and Grant, 2005). With the empirical support of the compiled data sets presented in this paper, we are able to identify the influence of specific parameters on the evolution of stored sediment deposits following dam removal.