Deconstructing Hydrologies: Reviving the Memory of Water in Dumbarton Oaks Park



Washington, DC | Elizabeth Anderson, Associate ASLA | Graduate | Faculty Advisor: Thaïsa Way, ASLA; Ken Yocom, John Findlay | University of Washington


Cultural landscapes are too often stewarded in ways that privilege their visual appearance as a static character rather than a dynamic response to shifting systems. Dumbarton Oaks Park in Washington, DC is a powerful illustration of the conflict between a site’s evolving ecology and federal landscape preservation guidelines that favor a lack of change over time. The c. 1921 Beatrix Farrand-designed valley, now a national park unit, has been endangered by stormwater runoff for decades, but managers have struggled to implement effective mitigation measures that engage the landscape’s performance in diverse domains. This stormwater retention proposal demonstrates the crucial role of design to the fields of landscape preservation and history. Likewise it promotes history as a tool of inquiry in design, vital to creating landscapes of ecological resilience, cultural relevance, and beauty.


The stormwater crisis in Dumbarton Oaks Park is the result of 1) the valley’s shifting ecological context, 2) a simplified reading of Beatrix Farrand’s work that evaluates her praxis on predominantly visual terms, and 3) management guidelines for the cultural landscape that similarly focus on its appearance and strive to effectively freeze the site in time. An iterative, performance-based exploration of the park reveals new possibilities for ecological resilience and improved cultural landscape stewardship through a respectful and elegant design for stormwater mitigation. Originally a naturalistic garden created by Farrand in the 1920s, the park is a 27-acre valley that encompasses a small tributary of Washington’s Rock Creek, including a series of eighteen constructed dams and associated structures. Once a part of the famous Dumbarton Oaks gardens, located just uphill, the valley was an integral informal component of a composition that balanced architectonic and organic elements. Winding footpaths led through a carefully curated sequence of dynamic spaces shaped by plants and water. Farrand exploited the dramatic slope between the stream and the upper gardens for aesthetic and functional purposes, inscribing the site’s hydrological patterns with buried irrigation pipes and decorative above-ground channels for spring water and stormwater runoff alike.

In 1940 the valley was separated from the rest of the garden and given to the National Park Service (NPS) by its owners, Mildred and Robert Woods Bliss. Since then it has suffered increasing degradation due to acute levels of stormwater runoff, encroachment by invasive species, and heavy public use. Although the NPS has viewed the abundant presence of water and people in the park as threats to its resources, a deeper look at the valley’s history indicates that water and people have always played an active role in shaping the landscape. The park’s location in a topographical bowl means that it has always received water from its surrounding watershed, and later, sewershed. Ancient hydrological patterns formed the valley’s slopes and the course of its stream, called “the Branch,” which Farrand incorporated into her design. Moreover, the Branch—like Rock Creek and its other tributaries—was viewed by early city planners as a practical conveyance for wastewater, as evidenced by the combined sewer line that traces its course. People have long manipulated the site, whether for forage, hydropower, livestock grazing, waste removal, or recreation. Today the park relies on a volunteer staff for its maintenance. Instead of threats, water and people should be seen as contributing resources to be encouraged in the design.

Park management is currently regulated by the Secretary of the Interior’s Guidelines for the Treatment of Cultural Landscapes, codified in 1996 and heir to an historic preservation philosophy—first developed for architecture—that is object-oriented and emphasizes “integrity” and a lack of change over time. This notion of stopping time and change is clearly problematic for landscapes, composed of constantly shifting systems. Furthermore, the cultural landscape is evaluated according to visual metrics, for example “character-defining features,” that address only how the landscape looks (and looked) rather than how it works. Dumbarton Oaks Park expresses the shortcomings of this approach in an especially salient manner. Increased runoff due to an increasingly developed urban context is sending stormwater into the park in greater quantity and at greater speed. Farrand’s small dams, intended to enhance the experience of the stream by creating pools and waterfalls, were not sized to accommodate the amount of water the stream receives in today’s storm events. The dams limit the stream’s capacity, which is further reduced by sedimentation deposited on the valley floor from highly erosive slopes. As stream capacity is reduced, the force of the runoff has begun to break the dams apart and take over Farrand-era pathways, even uprooting trees. Conditioned by the cultural landscape guidelines described above, however, the significance of Farrand’s design in the valley has been reduced to the dam structures themselves. This creates a Catch-22 for park managers who are charged to preserve the very elements that are contributing to the valley’s unstable ecology.

Approaching the site as a designer, however, allows a performance-based analysis and proposal that restores ecological function while protecting cultural landscape features. Instead of appraising Farrand’s work in purely visual terms and attempting to recreate the visual experience of the valley in the 1920s, analyzing the landscape’s historical performance both before and during Farrand’s tenure reveals a site that was highly adept at conveying and absorbing large quantities of stormwater.

Historical maps indicate that several sub-tributaries originally fed the Branch, before they were piped and buried in the early 20th century. Not surprisingly, the paths of these former sub-tributaries correspond to locations of concentrated runoff and erosion today. Similarly, their historical confluence points receive the greatest amount of sediment load and flooding during storm events. A careful study of Farrand’s design reveals that she sited pools along the stream course to conform to these original floodplains, working with the valley’s existing hydrological and topographical patterns and using them as aesthetic space-shaping devices.

Ancient fluvial terraces form a swath of meadows in the northwest section of the park: these more gentle landforms historically dispersed and absorbed runoff. Farrand preserved the meadows while planting the steep banks of the original sub-tributaries with native trees. These plantations effectively slowed the surface water in the channels while re-establishing a native hardwood forest on the slopes and lending rhythm and depth to the experience of the valley.

Farrand clearly engaged ecological systems to support her aesthetic objectives for the valley, enlisting the landscape’s performance to shape space and dynamic, multi-sensory experiences. Like many women practitioners in the early 20th century, however, Farrand’s achievement has been largely reduced to a collection of Gilded Age gardens designed for well-heeled relatives and acquaintances. Almost without exception, her work has been evaluated according to visual measures only—her command of architectural detailing, for example, or her expert use of plant material, or her skill at balancing American and European stylistic traditions. This simplistic reading of Farrand’s praxis ignores her lifelong interest in landscape ecology and a career-long experimentation with ecological principles, which began in 1901 with her design of a stormwater conveyance system and restored native woodland for her aunt, Edith Wharton, at The Mount in Massachusetts. Farrand would deploy the same principles two decades later in her plan for the Branch valley, and yet this systems-based understanding of the park’s design has been overlooked in favor of its appearance.

A proposal for stormwater mitigation, based on an analysis of the landscape’s historical performance, suggests a new paradigm for cultural landscape preservation. Stormwater resumes its role as a space-making device. The fluvial terraces and Farrand’s meadows are scooped out to more actively slow runoff and trap sediment. Farrand-era pools at confluence points are allowed to flood beyond their current boundaries during storm events, relieving pressure on the dams while encouraging the return of riparian habitat and many of the hydrophilic species from the 1920s naturalistic garden. Toe-of-slope wetlands fed by parking lot runoff encourage the return of native species and discourage invasive plants. Park visitors experience the constant interplay of water, people, and plants in the valley, understanding the cultural landscape as a palimpsest of ongoing relationships in which they play an active role.

"How do we do a contemporary intervention in an historic space? Wonderfully filled with all the things we ought to think about."

- 2015 Awards Jury