Abstract
Originally developed in California, earthwork terracing has proliferated worldwide as an expedient method for grading terrain for subdivision development. In Western Australia, terracing is now entrenched as conventional practice for facilitating suburban development on highly malleable coastal dunes. Although the adoption of international urban design models was partially intended to reduce topographic modification, retaining walls continue to increase in height and extent. To address this disjunction between theory and practice in current suburban development, this article tests landscape-based urban design mechanisms for improving the conservation and expression of natural topography on a coastal site in Perth, Western Australia. With the international growth of coastal suburbanization, the article contributes to the establishment of more topographically sensitive suburban planning practices on steep coastal settings.
INTRODUCTION: SUBURBAN BENCHING
Terraced site works first occurred in California in the 1950s in response to suburban expansion into mountainous areas. Referred to as “mountain cropping” (Bronson 1968, 35), the practice permitted standardized flatland suburban development models to be established irrespective of site conditions (Banham 1971). Following the influence of U.S. suburban ideals and practices, large-scale earthworks began to emerge in Australian suburban development in the 1970s.
In Western Australia in particular, mining-scaled earthmoving processes enabled steep coastal sand dune terrain to be readily remodeled to facilitate rapid suburban sprawl. By the mid-1980s, site preparation had evolved to the total re-engineering of natural topography into suburb-scaled systems of level lots retained with limestone block walls. Termed benching in the local industry, the practice offers numerous performance advantages over older suburbs subdivided with limited reshaping of the natural terrain. These include the ability to develop steep sites with increased lot densities, larger house footprints, standardized construction techniques, and improved flow between indoor and outdoor entertainment areas. Additional performance benefits associated with earthwork include more efficient and cost-effective sewerage and stormwater systems, standardized road grades, simplified street layout design, and the capacity to maximize valuable views from each lot (Kullmann 2014a).
While leveling land is integral to the history of human inhabitation (Rykwert 1976; Leatherbarrow 2004) and is convenient for the development industry, the expansive scale of benching in the suburban context produces numerous negative biophysical and experiential consequences. The process of bulk earthwork necessitates removing all native vegetation, which eliminates significant sources of habitat, biodiversity, and human amenity in the suburban environment (Cary and Williams 2000). Moreover, established soil-strata are disturbed or eliminated and micro-ecologies cannot be readily reconstructed, which has ramifications for soil stability, fertility, and ground water quality (Grose 2010).
From an experiential perspective, landform is a significant feature of landscape character, which in turn is fundamental to spatial navigation, perception, and placemaking (Tuan 1974). In particular, terrain exerts substantial agency over the legibility of urban environments, whereby “topographic gradients” (slope) are fundamental to the formation of directional differentiation (Lynch 1960, 96). Where large-scale suburban benching substitutes natural terrain with an artificially re-engineered topographic system, topographic gradients are likely to be obscured. Moreover, suburban benching potentially undermines how residents physically, psychologically, and creatively interface with their environment. Corporeal interaction with complex and variable landscapes acts as a vital catalyst of creative expression, which enhances the establishment of placemaking (Sennett 1998).
In response to the negative impacts of large scale earthwork, Western Australian planning policies adopted in the twenty-first century aim to minimize the impact of benching while simultaneously mandating improvements to a range of standard urban design performance criteria1 (WAPC 2002; WAPC 2004a; WAPC 2004b). In practice, the two objectives appear incongruent, whereby the application of urban design principles developed in Europe and the US coincides with higher retaining walls in the most recent coastal suburban developments (Kullmann 2014b) (Figure 1).
Northern Perth suburb of Jindalee illustrating retaining walls up to 20 feet high. (Image courtesy of Google © 2012.)
Research Scope: Enhanced Topographic Expression in Suburban Development
Given the apparent incongruity between local terrain and unmodified urban design models imported from abroad, this article aims to test mechanisms for improving the conservation and expression of natural topography in current coastal suburban development. The objective is explored by examining comparative design scenarios for a greenfield subdivision development site on the northern periphery of Western Australia’s capital city of Perth.
A substantial body of international literature identifies both the vulnerability of coastal landscapes and the environmentally, socially, and economically unsustainable nature of unconstrained suburban expansion (Bruegmann 2006). For this reason, Perth’s coastal sprawl—which has been officially sanctioned for nearly half a century (MRPA 1970)—ideally will be slowed or stopped. Towards this goal, government policy over the past decade has sought to limit sprawl through higher densities in new development and transit-oriented densification in older inner-city suburbs (WAPC 2004a; WAPC 2004b). Nevertheless, strong community opposition to coastal high-rise development continues to limit the efficacy of densification policy, with coastal sprawl continuing to expand at a rate of approximately half a mile per year (WAPC 2013). Therefore, the article takes the position that concurrent to improving the effectiveness of urban densification strategies, the ongoing improvement of current design and planning practices in greenfield subdivisions remains valuable for the foreseeable future (Gleeson 2006).
Although the practice and morphology of suburban benching—along with the majority of research on the topic of suburbanization—originates in the U.S., the Australian context differs in several key aspects. First, suburban growth in Australia is more centrally controlled at the state level than in the U.S., where planning decisions with regional impact are typically generated at the county or municipal level (Troy 1996; Gleeson 2006). State government oversight results in a planning process that benefits from consistency across jurisdictions, but conversely is potentially less reactive to local conditions and issues than in the U.S. Second, changes in topographic slope in the undulating coastal landform of Australia tend to be more gradual than the geomorphology of the Western U.S. (where benching originated), which is typified by distinct demarcations between flatlands and rugged mountain chains. Where competent regional planning potentially restricts development in steep zones, this well-defined topographic/residential demarcation is not as evident in continuously undulating terrain. And third, whereas topographic elevation strongly influences property values and socio-economic patterns in the U.S., proximity to the ocean is typically more influential over property values in Australia (Fraser and Spencer 1998).
Design Study Site: Burns Beach Northern Residential Precinct
Originally platted in the 1920s, Burns Beach is an isolated coastal settlement located 19 miles north of Perth. In the 1990s, metropolitan suburban sprawl leapfrogged the community (Figure 2). Since 2005, coastal heathland immediately to the north of the original Burns Beach settlement has undergone phased development for suburban housing. As is typical of current practices along the Perth coastal plain, development necessitates the complete clearing of virgin heathland and installation of large-scale earthwork benching prior to the sale of individual lots.
Location of the Burns Beach study site in the context of Australia and Perth. Coastal settlement patterns are evident in the map of Australia. Each dot represents 500 people.
Designated as the Northern Residential Precinct (NRP), the final phase of the Burns Beach development is planned for a particularly steep area of stable secondary dunes (Figure 3). The topography of the 62-acre site is characterized by a diagonal ridgeline that separates lower areas in the southwest corner from the highest points in the northeast. Aeolian depressions and knolls are dispersed complexly throughout the site, with slopes on the steepest sand formations reaching the natural angle of repose of 73%. Although the impact of uncontrolled recreational off-road vehicles results in some dune degradation, the indigenous coastal heath and grasses remain mostly intact prior to forthcoming development (author’s field survey, July 2014).
Burns Beach with the Northern Residential Precinct (NRP) development site demarcated with a dashed line. Areas of exposed sand are undergoing development. (Image courtesy of Daryl Jones © 2012.)
Development Framework: Aspirational Guidelines versus Entrenched Practices
The Adopted Structure Plan for the NRP acknowledges the extremely steep topography of the location. As directed by state government planning policy, the key development objective is to provide “quality residential outcomes” while “retaining the general landform of the site” (City of Joondalup[CoJ] 2007, 27). To retain the general landform while providing residential lots “developed upon a standard manner,” the Structure Plan states the need for retaining walls that “stabilize” and “retain the general slope and height of the ‘original’ landform” (CoJ 2007, 27). To achieve this compromise, development must be benched up to the natural high point in the center of the site. In order to minimize visual impact from the street, retaining walls are limited to a height of 13 feet, and must be located near the rear boundaries of lots (CoJ 2007).
The Adopted Design layout for the NRP claims to spatialize these objectives. The layout is structured around a narrow linear park that forms an axis between the highest point in the center of the site and a low point in the northwest corner of the site that is used for storm water detention (Figure 4). Streets running perpendicular to this spine establish a stepped sequence of benches with equal distribution of ocean views. In the southern portion of the site, the residential streets are parallel to the coast to match the street grid established in earlier phases of the Burns Beach development.
3D model of the NRP Adopted Design, placed above a 3D mesh of the site’s natural topography.
In practice, this structure accommodates a subdivision layout typical of recent development in Perth’s northern suburbs. The 233 conventional street access suburban lots fall within a narrow size range of between 5,920 ft2 and 8,070 ft2, for an average of 6,940 ft2. When combined with the 27 smaller rear access lots of approximately 3,770 ft2, the 260 lots vary 210% in size. All lots average 6,670 ft2 to achieve an overall net density of 6.5 dwelling units per acre (du/ac) and gross density of 4 du/ac including road reserves and parks. Based on dwelling footprints in the initial phases of the Burns Beach development, houses in the NRP are assumed to average 3,880 ft2 on street access lots, and 2,690 ft2 on rear access lots. Dwelling footprints cover an average of 55% for the conventional lots and 70% for the compact rear-access lots.
Three-dimensional modeling demonstrates that the 48,980,000 ft3 of re-grading required to implement the Adopted Design heavily modifies the original terrain (Figure 4). The plan generally retains the higher and lower areas of the site, an advancement over the leveling of the entire fore-dune systems and elimination of the complex sand dune terrain that characterizes nearby conventional developments (Figure 5).
Northern Perth suburb of Alkimos, illustrating the elimination of retaining walls through the total leveling of sand dune topography. The dunes visible in the background originally covered the entire site. (Image courtesy of author 2014.)
Nevertheless, the restrictions of benching eliminate all indication of the topographic nuances between the highest and lowest points. This result suggests that coastal sand dune terrain is too complex and intricate to be retained under the conventional development model.
In attempting to conserve more of the natural shape of the terrain, developers claim to have considered alternatives to benching with retaining walls on lot boundaries. They conclude, however, that these unspecified alternative design and construction practices require lot sizes to be significantly larger than the compact lots that government policies presently mandate in an attempt to limit suburban sprawl (WAPC 2004a; WAPC 2004b). The developers argue that the provision of very large lots “to accommodate level rises without … retaining walls” would lower residential yields and as such “directly conflicts with government sustainability objectives” (CoJ 2007, 31).
Topography and Urbanism: International and Australian Perspectives
Reconciling the apparently divergent objectives of economically viable urban densities and topographic conservation is complicated by the political, impassive, or nostalgic nature of relevant critiques, models, and precedents. Although the earliest examples of suburban benching in the U.S. were criticized for perpetuating the environmental destruction typical of industrialized society (Blake 1964; Bronson 1968; Banham 1971), the aerial perspective of the practice has more recently been characterized as a type of infrastructural land art (Corner and MacLean 1996; Light 2015).
Limited explicit integration of topography into key contemporary urban models exacerbates the problematic relationship between urban/suburban development and landform conservation. Twentieth-century models of modern urbanism consolidate the city into towers floating above a terrain that flows unhindered beneath and between buildings. Although sound in theory, when realized on a large scale in postwar reconstruction in Europe and urban renewal in the U.S., decoupling urbanism from the ground plane unintentionally diminished stewardship of the landscape (Ingersoll 2006). Consequently, the pastoral ideal that characterises Modernism deteriorated into automobile dominated wastelands (Jacobs 1961; Sennett 1990). Ecological planning addresses this deficiency by integrating topography into the planning of residential density and distribution at the regional scale (McHarg 1969). Although the suitability analysis techniques of ecological planning successfully repelled developments and lowered residential densities on steeper terrain, the commercial returns from developing desirable locations often outweigh this reasoning. Moreover, suitability analysis is criticised for actually accelerating sprawl by dispersing suburban development (Hill 1992). Taking a more compact approach to cities, traditional urbanism systematically positions topography as a scenic rural/natural backdrop or as a parkland landmark within the urban zone. As the widely adopted Rural-Urban Transect (Center for Applied Transect Studies n.d.) illustrates, framing landform in this manner is most relevant to cities sited on flat river floodplains and deltas (Duany 2002). In contrast, expanding coastal cities often exhibit a far more complex interaction of topography and suburbanization (Bosselmann 2011).
In the Australian context, numerous architects continue to pursue the Modernist ideal of touching the earth lightly at the lot scale through topographically sensitive design (Paolella and Quattrone 2008). Although providing compelling architectural solutions, these prototypes are often isolated projects for privileged clients on spacious peri-urban lots with limited application to the mass-suburban context. Current literature directly addressing suburban benching in Australia is limited to incidental negative discussion in botanical and ecological studies (Rokich et al. 2001; Williams et al. 2005). As such, the most sustained critique of topographic modification in suburban development remains the work of George Seddon (1979; 1990), who criticized the earliest examples of suburban benching for extinguishing the character associated with the nuances of natural topography. As a public figure and internationally significant multidisciplinary scholar of the humanities, sciences, planning, and landscape architecture, Seddon synthesized typically disconnected biophysical and cultural perspectives on the environment. From this position, Seddon advised planners and designers to build sympathetically with the natural landform, and used suburbs platted prior to the 1960s as examples to be emulated in good practice.
The Evolution of Suburban Development Standards
Although the older costal suburbs that Seddon references are indeed characterized by limited modification of the natural topography, their development standards differ markedly from present day standards. Comparison of the NRP with the Perth coastal garden suburb of City Beach, constructed in the 1950s, illuminates these distinctions. First, the landforms underpinning the two subdivisions are different. Whereas City Beach is sited on undulating terrain with the steepest landforms set-aside as bushland reserves (Figure 6), the NRP is sited on steep fore-dunes that will be fully developed (Figure 4). Second, planning metrics vary between the two suburbs. Lots at City Beach are on average 60% larger and 20% more diverse in size than those at the NRP. Furthermore, site coverage at the NRP is up to ten times higher than City Beach, which equates to three times more unbuilt space in the older suburb with which to accommodate changes in topography. And third, infrastructural performance differs markedly; while the NRP is optimized for efficient gravity-fed storm water and sewerage systems, City Beach remains dependent on outdated soakwells (that is dry wells) and septic tanks. The natural depressions in the terrain complicate the retrofitting of modern infrastructure and necessitate expensive vacuum systems.
3D model of the 1950s Perth coastal suburb of City Beach, illustrating low degree of modification between original and current topography.
Changing construction techniques evolving after the 1950s also interacted with higher site coverage ratios to transform suburban development. In the initial coastal settlements, shacks were constructed of wood frames set above the original landform on stilts. In affluent suburbs such as City Beach, double-brick construction with suspended wood floors predominated. Typically, this technique included a limestone foundational plinth that absorbed variations in topography with minimal disturbance of the original landform (Figure 7).
Circa 1950s houses set on limestone plinths that accommodate variations in topography without major earthwork. (Image courtesy of author 2015.)
The widespread adoption of concrete slabs supplanted both of these building techniques in the 1960s. Cost efficiencies dictated that each slab be constructed on a level sand pad significantly larger than the footprint of the dwelling. By the 1980s, space limitations required that sand pads be retained with walls at the lot boundaries. Since the 1990s, site coverage ratios above 40% resulted in level sand pads filling out the entire lot (Figure 8).
Circa 2010 house set on sand pad that fills out the entire lot and is retained at the property boundaries. (Image courtesy of author 2014.)
DESIGN SCENARIOS: ALTERNATIVE SUBURBAN LAYOUTS
Real world factors that drive current high lot densities and large dwelling footprints limit the applicability of alternative and historical suburban design practices. Although government policies, developer yield expectations, and owner lifestyle expectations are unlikely to shift significantly in the near term, other design factors remain more flexible. In this article, different scenarios explore the potential for factors as diverse as suburban layouts, construction techniques, distribution of lot sizes and public open space to conserve natural landform at typical suburban densities. The scenario methodology permits equitable detailed comparison of the impacts of diverse planning approaches (Chakraborty et al. 2011).
Set on the NRP site, the five scenarios aim to maintain the total yield of the Adopted Design layout (260 dwellings), and are limited to locally accepted suburban densities of between 3 and 12 du/ac (for reasoning refer to Research Scope). These densities are manifested predominantly as fully detached, front access, single and double story houses with twenty-foot (car length) street setbacks. This standard template is augmented with medium-density fully detached, rear access, double story houses (and in one scenario with low-density front access bush-lots). Additionally, contiguity between suburban houses and their private outdoor spaces are maintained in this study. This contrasts with residential patterns common in Northern European cities where dwellings and private gardens are often located in different parts of the city.
As does the Adopted Design, each scenario interfaces with the existing street grid on the southern side and maintains the perimeter road with a mandatory 600-foot coastal setback. As government policy dictates, approximately 10% of the site is designated as public open space in each scenario, with open space retained in the northwestern corner of all scenarios for storm water detention. Where included in the scenarios, retaining walls are capped at 13 feet for consistency with the Adopted Design. Fifty-foot road easements and 20-foot laneway standards are also maintained. By contrast, whereas road grades in the Adopted Design and other contemporaneous suburban subdivisions rarely exceed 10%, maximum road grades for the scenarios are increased to 15%. This adjustment acknowledges that far steeper grades were incorporated into older coastal suburbs without significant reductions in usability.
The topographic expression of each scenario is evaluated using earthwork volumes and criteria drawn from the subjectivist framework for visual landform character analysis developed by Tveit, Ode, and Fry (2006) and Ode, Tveit, and Fry (2008). This framework offers a range of criteria adaptable to the particular characteristics of the Perth northern coastal heathland environment, which include visual fragmentation, vulnerability to anthropomorphic disturbance and variation in topographic complexity. To apprehend these key characteristics, the following criteria were selected for their applicability to the morphology of the study site: a) imageability (the contribution of landmarks and other topographic elements to a strong visual image); b) disturbance (the deviation of topographic features from the original context); c) complexity (the diversity and richness of topographic features). In addition to this visual framework, landform is also perceived and evaluated kinesthetically (through the body while in motion) (Lynch, Appleyard, and Meyer 1964).
Scenario One: Adopted Layout with Earthworks Limited to Roads
Prior to 1970, minimal modification of the original landform prior to the subdivision and sale of individual lots characterized the development of Perth suburbs. Site earthwork was generally limited to minor re-grading to accommodate street profiles and storm water detention sumps. Suburban lots were sold as mostly vegetated, with site clearing the responsibility of individual landowners. This practice—combined with low site coverage ratios and low-impact dwelling construction techniques—resulted in significant retention of original topography and native vegetation. The first scenario tests the historical practice of limited earthwork at current densities and site coverage ratios. For comparison, the Adopted Design layout is retained in plan form and superimposed over the original topography (Figure 9). Earthwork is limited to localized cut and fill to accommodate street grades of up to 15%. With the removal of flat lots, each dwelling and associated private outdoor space negotiates topographic variations on site.
3D model of scenario one, illustrating the topographic impact of the Adopted Design with only minimal topographic modification to accommodate road profiles.
Urban performance
With 226 lots, the residential yield of scenario one is 15% lower than the identical Adopted Design layout with benching. This yield loss results from several factors. First, storm water detention basins displace lots situated in natural low points. This arrangement is common in suburbs constructed prior to the 1960s, and while improving local infiltration, creates less effective overall drainage infrastructure. Retaining the natural depressions also complicates modern sewerage infrastructure, with complex vacuum-sewerage systems required to counteract gravity. Second, additional yield loss results from lots rendered inaccessible and undevelopable due to a high proportion of slopes set near the steepest natural angle of repose of 70%. And third, road alignments originally designed for benching are not topographically optimized, resulting in cuts up to 40 feet deep and encroaching up to 80 feet into adjacent lots.
While numerous retained natural high points enable panoramic views from the highest lots, views are distributed less evenly than in the Adopted Design. Vehicular permeability for this scenario is also slightly lower than in the Adopted Design as vehicular roadways are discontinued in several instances where roadway easements exceed the 15% grade limit. While pocket parks with staircases could potentially occupy the road reserve in these locations, vehicular access to adjacent lots remains impacted.
Topographic expression
The 8,300,000 ft3 of earthwork required to accommodate the layout equates to 17% of the earth movement volume of the Adopted Design. In comparison with the original landform, the overall topographic imageability of the scenario is moderate. The road cuts dissect significant portions of the main ridgeline, which diminishes the legibility of the remaining landform from a distance. The road cuts also create significant landform disturbance, although conserved fragments of topographic complexity elsewhere in the development counterbalance this impact. Even though dwellings cover more than half of each lot’s total area, enough of the natural landform is retained in front, behind, and along the street to convey a coherent image of the original topography. Houses elevated on plinths or framed structures act as datums within the landscape that enhance that image. Furthermore, the physical barriers that discontinued streets create on the steepest slopes ensure that topography actively influences legibility and wayfinding in the suburb.
Design challenges
In contrast with floating wood floors typical in pre-1970s suburban housing, dwellings constructed on concrete slabs require substantial sand importation to backfill each plinth. Given that contemporary suburban houses are approximately 80 feet long, the finished floor level of dwellings set on the steepest slopes of 70% are elevated up to 66 feet above ground level. Typically, 3-foot setbacks from the side property boundaries result in 7-feet wide chasms between adjacent houses with little natural light or amenity. Moreover, private open space is likely to be set significantly higher than the finished floor level of the house, which complicates the indoor/outdoor flow that is typical of contemporary suburban homes.
To reduce excessively high plinths, houses may be partially excavated into sloping sites and/or set over two or more split-levels. Challenges associated with these methods include increased complexity of the concrete slab and diminished internal accessibility due to level changes within the house. While substitution of brick on slab construction with lightweight framing methods eliminates bulky plinths, this construction technique continues to impact the ground plane, with the space underneath the house typically devoid of sunlight and requiring artificial stabilization.
Scenario Two: Grid Layout with Partially Leveled and Split-Level Lots
Prior to the mid-century introduction of curvilinear layouts, Perth coastal suburbs were generally platted into east-west grids without regard to the local landform. Despite this apparent disjunction, many older gridded suburbs involved minimal topographic modification. Although low site coverage ratios were clearly a factor, the structure of the grid is also significant, whereby the repetitive geometry creates a set of parallel section lines that trace variations in the ground (Figure 10). Following this historical precedent, the second scenario extends the existing parallel streets at the southern end of the NRP into a grid layout across the site. Street grid intersections are set close to the natural contour and long streets are aligned with the natural contour at three equidistant locations and smoothed to the maximum gradient of 15%. Each lot is benched using conventional retaining walls at the side and rear property boundaries but not at the front. In contrast to the Adopted Design, where lots are set to match the road level on both sides of the street, each lot level is set to the natural contour at its center. On steep slopes that would necessitate retaining walls higher than 13 feet, lots are split into two benches set 9 feet apart.
1940s Perth coastal suburb of Watermans Bay, illustrating a warped street grid expressing topographic undulations. (Image courtesy of Google © 2013.)
Urban performance
The grid for Scenario two (Figure 11) delivers a residential yield equal to the Adopted Design. However, with 50 compact rear access lots supplementing 210 front access lots, this yield is contingent on the inclusion of nearly double the number of compact rear access lots included in the Adopted Design (n=23). The loss of front access lots is counterbalanced by improved urban permeability. The permeability of pedestrian circulation is higher than that for vehicles due to the conversion of side streets to public open space. While the yield target dictates this configuration, converting side streets to parks is common in pre–WWII grid suburbs and has the advantage of distributing public open space evenly throughout the neighborhood.
3D model of Scenario two, illustrating the topographic impact of the grid layout with partially leveled lots.
Topographic expression
The 16,740,000 ft3 of earthwork required to accommodate the layout equates to 34% of the earth movement volume of the Adopted Design. Although the grid significantly modifies the original terrain, the general features of the landform are maintained, resulting in a moderate visual expression of the main topographic features from a distance. The grid and wall system creates a pixelated version of the original topography that lacks complexity. However, while the steepest slopes and more intricate sand dune features are eliminated, the overall topographic impression retains key traces of the landform. The imageability of these traces is amplified along parallel streets, which lead directly up and over topographic features, and offer revelatory views at hillcrests. Views out of the suburb are widespread and are incorporated into private lots, public space and streets. The setback of retaining walls 20 feet into the lots also reduces their visibility from the street, although vertical drops of up to 13 feet continue to dominate the interface between adjacent lots.
Design challenges
Numerous disjunctions between the arbitrary grid and the natural topography occur along the southwest facing slopes of the main ridgeline through the site. In several locations the street grid encounters natural slopes of up to 60%, which is almost four times the maximum permissible road grade. In numerous instances, the natural levels between parallel streets significantly exceeds the maximum retaining wall height of 13 feet. Lots situated in areas that remain excessively steep after slope moderation are bisected with 9-feet high retaining walls. This split-level configuration enables back-to-back lots to transcend almost 40 feet in elevation difference between parallel streets without breaching the retaining wall maximum height of 13 feet. The 9-foot step-down is calibrated to accommodate a partial ground floor, either at the front or rear of the house depending on the direction of the slope. Advantages of this configuration include relocating level changes from the edges to the center of a lot, and the potential for dwelling floor plans that make innovative use of the dual levels. Disadvantages include reduced access between the front and rear levels of a split-lot, and increased design and construction costs in comparison with conventional level lots.
Scenario Three: Roads Along Swales with Earth Smoothing
In sandy terrain, swales normally provide the most level routes for movement. As a consequence, circulation patterns in unplanned coastal shack settlements traditionally followed swale lines. Similarly, in the 1950s, the first coastal suburban layouts to depart from the time-honored grid used curvilinear streets aligned to swales. Placing roads in swales creates a funneling effect along the streetscape, with houses generally sited above the road on both sides (Figure 12). Swale lines also generally simplify storm water engineering and reduce the amount of cut and fill required to accommodate road profiles when compared with roads that contour along hillsides. In the third Scenario (Figure 13), local streets set at maximum 15% grades are designed to follow the natural swale lines at the NRP site. Bulk earthwork is used to smooth back the natural terrain to establish a balance between retaining the topographic form and ensuring that lots are sufficiently moderated to be buildable. During the bulk grading process, protruding knolls are rounded off, the steepest slopes are smoothed back to 50%, and depressions are raised to match the street level. As was true in Scenario one, houses are set on solid plinths or lightweight sub-structures.
1960s Perth coastal suburb of Quinns Rocks, illustrating streets set along swale lines with houses set on ridge lines on both sides of the street. (Image courtesy of Google © 2013.)
3D model of Scenario three, illustrating the topographic impact of aligning roads along swales and light smoothing of terrain on residential lots.
Urban performance
The swale-based lot layout of this scenario comprises 227 front access lots and 27 compact rear access lots for a total yield of 256 lots. While the overall yield and allocation of lot sizes is comparable to the Adopted Design, greater variability of lot sizes and geometries occurs within this range. As a consequence of following alignments dictated by the landform, the layout of streets and lots necessitates several cul-de-sacs and numerous looping roads that change direction significantly along their length. In two-dimensional plan view, this configuration appears to provide lower urban legibility and permeability than the grid based systems of previous scenarios. However, the layout is more legible when perceived three-dimensionally within the context of the dune structure. By framing the streetscape on both sides and setting houses above the street, lots sloping up from the street level also enhance urban legibility. In addition, the swales create a funnel effect that supports standard gravity fed storm water and sewerage infrastructure.
Topographic expression
The 7,130,000 ft3 of earthwork required to accommodate the layout equates to 15% of the earth movement volume associated with the Adopted Design. Despite rounding of some knolls, the smoothed landforms remain highly imageable. With road alignments closely calibrated to the original swales, site circulation approximates natural desire lines that exist in the unmodified landscape. Following swale lines minimizes changes in elevation, particularly at street termini where natural saddles represent the most efficient location for cresting ridgelines. Situating roads along swales and over saddles also amplifies the sense of topographic immersion, which contrasts with the panoramic external views of Scenario two. Although expression of the most pronounced dune formations is diminished, a consistently moderate level of topographic disturbance across the site mitigates this impact. As a consequence, a general sense of the complexity of the original topography is present throughout the modified landform.
Design challenges
As distinct from the long, contiguous and parallel swales that characteristically separate primary and secondary dune systems, the NRP site comprises many short disconnected swales and gullies. Fitting standardized lot sizes and roads into this highly variable and intricate dune structure results in complex lot geometries. Given site coverage ratios above 50%, these irregular lot shapes directly impact dwelling layouts and reduce the capacity for standardized housing design. In addition, although houses elevated slightly above the street are ideal from the perspective of real-estate “street presence”, many dwellings are set more than a full floor above street level. Reintroducing the undercroft garages that were common in vernacular architecture up until the 1960s is one method for utilizing this street-front level differential between house and lot.
Scenario Four: Conserved Ridgeline with Unmodified and Conventional Benched Lots
Even when occupying a relatively small portion of a site, the high visibility of ridgelines amplifies their influence over the topographic character of a neighborhood (O’Shea and Ross 2007). In Perth coastal suburbs established prior to the 1980s, the steepest dune ridgelines typically remained as undeveloped bushland reserves (Figure 14). Since the 1980s, development yield pressures, escalating land values, and earthmoving technology have led to increased residential development of ridgelines. In addition, the shift to combining parks with storm-water detention basins displaced public open space from ridge tops to the lowest elevations. The fourth scenario aims to reverse this trend by conserving the most elevated areas of the site (Figure 15). To achieve this goal while meeting residential yield targets, all public open space not essential for storm water detention is situated on the ridge. In addition, the remainder of the ridge is platted as large “hill” lots. These lots are developed under the caveat that minimal earthwork is undertaken for dwelling construction and the remainder of each lot remains vegetated with indigenous coastal heath. The remaining lowlands are benched as per current conventional practice.
1970s Perth coastal suburb of Kallaroo, illustrating dune ridgeline conserved as bushland. (Image courtesy of author 2014.)
3D model of Scenario four, illustrating the topographic impact of siting large lots and public open space on the highest areas of the site.
Urban performance
With 254 lots, the total yield for Scenario four is close to the Adopted Design yield. In addition to 128 regular front access lots, the scenario comprises 54 hill lots, which average 12,920 ft2 or 3 du/ac. Offsetting this area of low-density development requires 72 compact rear access lots, which equates to 2.5 times more compact rear access lots than in the Adopted Design. Moreover, while lot sizes vary 430% between the smallest and largest lots, actual size variation within each of the three lot types is minimal. As there are few roads in the topographically unmodified hill lot area on the ridgeline, urban permeability is low and connectivity is restricted across the site. Conversely, the elevation of the ridge and the elimination of natural depressions amongst the leveled lots facilitates efficient storm water and sewerage infrastructure.
Topographic expression
The 15,400,000 ft3 of earthwork required to accommodate the layout equates to 31% of the earth movement volume of the Adopted Design. As is consistent with the twofold development strategy, the topographic expression of the scenario is bimodal. The ridgeline is visible from many parts of the surrounding development and conveys the topographic silhouette of the site. With numerous streets terminating at the base of the ridge, this central elevated position amplifies the topographic imageability of the ridge within the suburb. Along the ridge, dwellings scattered at low densities with retained natural vegetation maintain a high degree of complexity. In addition, the hilltop park allows the ridge to be physically experienced, rather than only visually registered from a distance. In contrast to the high imageability of the ridge, natural topographic character is entirely eliminated in the conventionally benched lowlands. These areas, which comprise the majority of the development area and residential yield, exhibit very high degrees of topographic disturbance and very low degrees of imageability and complexity.
Design challenges
In order to maintain overall residential yield, more compact rear access lots are required to offset the significant number of large hill lots. Although additional compact lots align with demographic shifts towards smaller households (ABS 2010), the conventional benching required to accommodate both compact and conventional lots eliminates direct interaction with landform for 80% of the residences in this scenario. Privatizing the desirable ridge outlook to 20% of households creates clear segregation from the higher density leveled lots in the lower areas of the site. The incorporation of public open space on the ridgeline is essential to offset the impression of the ridge as a privatized and segregated barrier that bisects the suburb. Maintaining topographic conservation requires clear urban guidelines for hill lots that define building envelopes and construction methods, and limit indigenous vegetation clearing and earthwork in the remainder of each lot. While these controls are straightforward to define and enforce during the initial permitting and construction phase, the piecemeal site modifications that individual homeowners make over time are far more difficult to police. The incremental clearing of vegetation and leveling of slopes to accommodate ancillary structures and entertainment spaces is therefore likely to erode the topographic advantages of hill lots over time.
Scenario Five: Conserved Depressions with Conventional and Split-Level Benched Lots
Due to the absorbent substratum, the action of wind—rather than water—predominantly shapes sandy terrain. The numerous depressions that result from this process connect the ground surface with the ground water table. Despite this important hydrological role, low-lying areas have been filled in and built over throughout the history of Perth’s settlement. Although environmental policy now mandates that larger wetlands be conserved during suburban development, they continue to be degraded through poor design integration and parkland maintenance regimes (Davis and Friend 1999) (Figure 16). Moreover, smaller depressions are routinely eliminated during the site engineering process to simplify construction of gravity-fed storm water and sewerage infrastructure. The fifth scenario retains all but the smallest depressions and their natural vegetation on the NRP site (Figure 17). To minimize degradation of natural systems through direct adjacencies with residential lots, a street or laneway buffers each depression. A meandering grid of streets connects the depressions among suburban lots developed using conventional benching practices.
Local park in the Perth coastal suburb of Craigie, illustrating a depression retained as a poorly integrated storm water detention basin. (Image courtesy of author 2014.)
3D model of Scenario five, illustrating the topographic impact of retaining depressions within conventional benching practices.
Urban performance
With 247 lots, the total yield for Scenario five falls 5% short of the Adopted Design yield. This shortfall results from the increased area allocated to retaining the depressions, which produces 40% more open space than the Adopted Design. To compensate, 70 compact rear access lots are incorporated into the yield, which is more than double the compact lot allocation in the Adopted Design. The meandering street layout that links the depressions creates moderate lot size diversity. Irregular lot geometries adjacent to depressions do, however, present challenging proportions for accommodating dwelling footprints. The absence of cul-de-sacs in the street network facilitates high urban permeability. As a result of conventional benching, sewerage and storm water infrastructure efficiency is good, albeit with the important environmental caveat that retained depressions are not converted to storm water detention in the future.
Topographic expression
The 23,450,000 ft3 of earthwork required to accommodate the layout equates to 48% of the earth movement volume of the Adopted Design. Since depressions are naturally distributed across a wide range of elevations and locations throughout the site, their retention inherently restricts the extent to which the site’s natural landform may be leveled. Public spaces are created in the depressions making them more evident and contributing to the topographic character of the neighborhood. Having center elevations that are below ground level, the depressions cannot be seen from a distance, and they are therefore unable to function as imageable landmarks in the same capacity as ridges and knolls. Moreover, the elimination of highpoints and other complex topographic features removes the natural topographic context of the depressions. When experienced in isolation, the conserved depressions contribute minimal information to the imageability of the original landform, which is highly disturbed throughout the development.
Design challenges
Expressing the site’s natural depressions within the suburban development presents several design complications. Unlike development in dendritic terrain—in which creek lines typically form the basis of public open space—the fragmented non-linear distribution of depressions does not support an integrated open space system. Moreover, small fragmented remnant depressions are highly vulnerable to ecological degradation as a result of high perimeter-to-area ratios (Stenhouse 2004). In addition to improving general imageability, encircling depressions with roads minimizes weed ingress and prevents incremental privatization from adjacent private property. Nevertheless, road buffers may position depressions as an inconvenience to motorists and unintentionally perpetuate their historical devaluation. For this reason, the implementation of effective long-term land management policies is essential to prevent the conserved depressions from becoming depositories for waste or being incrementally leveled and grassed to accommodate active recreation.
EVALUATION: RECONCILING LANDFORM AND URBAN PERFORMANCE
The relative performance of the five scenarios and the Adopted Design across key quantitative and qualitative criteria is summarized in Table 1. To reflect the broadly opposing values of conservation and development, performance metrics are grouped into landform expression and urban design/development criteria.
Scenarios ranked against selected landform expression and urban design/development criteria. 1 = best performance ↔ 6 = poorest performance.
All of the scenarios express the site’s pre-development landform more comprehensively than the Adopted Design, which ranks last across all landform criteria. Ranked second to last, Scenario five is the least effective of the scenarios, with the conserved depressions and conventional benching generally exhibiting poor landform expression across most metrics. At the other end of the spectrum, Scenario three is most effective, with the system of topographic smoothing and swale-aligned layout ranking first in the majority of landform expression criteria. Scenario four exhibits the second most effective expression, with the bimodal strategy of ridgeline conservation and conventional benching ranking highly in most criteria.
Overall, the scenarios exhibit decreased urban design and urban development performance in comparison with the Adopted Design. Scenario two ranks second highest overall, with only low lot diversity and lower storm water and sewerage infrastructure performance undermining the generally high urban performance of the traditional grid layout. Scenario four ranks a close third, with the combination of ridgeline conservation and conventional benching rating well across most urban metrics. Significantly, Scenario four substantially outperforms the Adopted Design in the criterion of lot size diversity through the inclusion of large hill lots and increase in compact rear access lots. Scenario two, on the other hand, exhibits the lowest overall ranking. In this scenario, the translation of the Adopted Design layout onto a non-benched landform strategy compromises all urban metrics.
When both landform expression and urban design and development are evaluated together, Scenario four exhibits the highest and most balanced overall performance across both groups of metrics. Somewhat counter-intuitively, this balance is achieved by segregating urban development areas from landscape conservation areas within the development site. Two other scenarios also rate highly in the combined rankings, albeit with significant compromises from divergent priorities. The grid strategy of Scenario two scores highly in urban criteria and moderately in landform expression criteria for a high combined performance. The central function of retaining walls in structuring this scenario suggests the potential for benching to be reconfigured for improved topographic responsiveness. From an alternate perspective, Scenario three exhibits a similarly high combined performance. With a very strong expression of original landform offsetting comparatively urban poor performance, this scenario suggests the potential for foregoing performance in certain urban criteria to achieve high overall landform expression.
CONCLUSION: SITE SPECIFICITY
The development scenarios for the NRP site demonstrate that government-mandated suburban densities can be reconciled with improved design responsiveness to natural sand-dune topography. Conserving major ridgelines with open space and low-density housing offers a satisfactory counter-balance to conventional practices elsewhere in the development. Retaining walls can express terrain if laid out on a grid in small enough units, including within individual lots. This assessment demonstrates that improved topographic expression is not necessarily contingent upon eliminating or reducing the height and extent of retaining walls. Aligning roads to swales and moderately smoothing the landform requires the least earthwork and conveys a consistently strong image of original landscape character across a wide range of scales.
Integrating elements of the scenarios into current suburban development requires significant re-evaluation of industry conventions and urban design policies. Adopting increased road grades and slower, tighter road curves enables suburban street networks to respond more closely to natural variations in topography. To achieve this, engineering standards that are closely linked to infrastructure criteria require recalibration. Increased diversity of lot and dwelling sizes also improves opportunities for suburban form to respond to topographic variation. To achieve this, the suburban monoculture of similarly sized houses requires augmentation with smaller and larger dwellings, which have the added benefit of accommodating increasing demographic diversity (ABS 2010). Alternative housing construction techniques enable lots to be developed without first being leveled. To achieve adoption, the industry-perpetuated cultural-bias against lightweight construction must be overcome. This represents an opportunity for residential architects, who currently design an insignificant percentage of suburban housing stock (Dunham-Jones 2000). Finally, looser urban structures enable street networks to more effectively incorporate natural landform. To achieve this, rigid urban performance criteria may be relaxed to allow variations in topography to distort neat, ordered urban structures.
These innovations require policy innovation to refocus an industry that is entrenched around bulk earthwork and benching as the most economic and expedient model for suburban development. Given the strong link between ocean views and property value in Perth (Fraser and Spencer 1998), poorer view distribution represents a major obstruction to widespread acceptance of lower topographic modification in coastal suburban development. With ocean views now more valuable than the cost of moving sand, extensive topographic modification has a strong economic rationale. This imperative will need to be systematically addressed by reorienting perceptions towards the value of inland views and lots that are sheltered from the elements. The suburban layouts most capable of expressing the topographic character of coastal dunes are likely to be more variable than the formulaic plan-oriented templates that currently drive development patterns. Most critically, this research indicates that expressing the topographic character of sand dunes necessitates a profoundly site-specific approach to suburban development.
With Perth projected to add 2 million people by 2056—and current government urban densification objectives proving difficult to realize in practice (WAPC 2010)—the sheer extent of ongoing suburban sprawl is likely to overwhelm any topographic improvements at the subdivision scale. That this sprawl is planned to extend into dune systems that are steeper than the more moderate topographies of older coastal districts to the south further exaggerates this disjunction. In this topographic and demographic context, extending the scope of future research beyond established low/medium suburban densities is required. Calibrating higher urban densities and novel dwelling arrangements to landform potentially offers further improvements of topographic expression while concurrently demonstrating the positive qualities of urban density.
Footnotes
Karl Kullmann is a landscape architect, urban designer and Associate Professor at the University of California, Berkeley, where he teaches design studios and courses in landscape theory, digital modeling and publications. Kullmann’s research and creative work applies landscape frameworks to urban processes across diverse topics, including urban topography, green infrastructure, rural and urban decline, urban wastelands, public gardens, landscape perception, mapping and representation. These topics are actively tested through design practice that includes constructed urban landscape projects in China, Australia and Germany, and numerous design competition prizes and exhibitions.
↵1. Urban performance criteria typically include: density (number of dwellings per unit of area); diversity (range of dwelling sizes and types); legibility (ease with which an urban landscape can be comprehended and navigated at ground level); permeability (degree of public accessibility and connectivity); and walkability (comfort and proximity of the pedestrian experience) (Carmona 2010).
