.|  Baltimore Ecosystem Study
Reconceptualizing Urban Land Cover: The HERCULES Model
Prepared by M.L. Cadenasso
Urban landscapes are characteristically heterogeneous and this heterogeneity is a result of fine scale variation in both natural and built features. Existing classifications of urban landscapes focus on the way the land is being used, such as for residential, commercial, or industrial purposes. But this view of the system neglects important features of the spatial heterogeneity in cities and suburbs. We developed a new classification for urban landscapes called HERUCLES

Figure 1. HERCULES model template
(High Ecological Resolution Classification for Urban Landscapes and Environmental Systems). This classification complements existing classifications, but focuses on the biophysical elements on the landscape rather than how the land is being used. HERCULES recognizes elements of urban land heterogeneity - buildings, surface materials, and vegetation (Ridd 1995), and divides these three elements into six features: (1) coarse-textured vegetation (trees and shrubs), (2) fine-textured vegetation (herbs and grasses), (3) bare soil, (4) pavement, (5) buildings, and (6) building typology (Cadenasso et al. 2007; Figure 1).
Urban Ecosystems and Spatial Heterogeneity

Figure 2. False color infra-red photo - vegetation is red and pavement and built structures are grey.
The classification is also intended to be applied at fine spatial scales which is appropriate for urban landscapes (Figure 2). Because HERCULES keeps the physical structure of the system separate from the function of that system, it can be used to test the link between structure and function. To learn more about the conceptual underpinnings of HERCULES, please see Cadenasso, Pickett and Schwarz (2007) found under the publications tab.
How are the patches created?

Figure 3. HERCULES boundaries in bright green on a false color infra-red image
Patch mapping and classification commonly consists of two steps: 1) delineating patch boundaries, and 2) deriving attribute values of with-patch features such as percent cover of trees and impervious surfaces. Humans are exceptionally adept at visually recognizing and interpreting complex spatial patterns through comprehensively using shape, size, color, orientation, pattern, texture and context in interpretations. Patches were digitized on-screen using high-spatial resolution imagery with the HERCULES classification scheme (Figure 3).
How are patches attributed?

Figure 3. HERCULES boundaries in bright green on a false color infra-red image

While visual interpretation is a good approach for patch delineation, it is not ideally suited for quantifying area estimates of within-patch land cover features. We initially estimated the proportional cover of each of the first five features listed in the model template (Figure 1). The proportional cover is divided into four ranges: (0) absent, (1) present to 10% cover, (2) 11-35% cover, (3) 36-75% cover, and (4) > 75% cover. We compared the estimates derived from visual interpretation to those derived from an object-based approach. Under the object-based approach, HERCULES patches served as pre-defined boundaries for finer-scale segmentation and classification of within-patch land cover features. Addition datasets including LIDAR data and building footprints were used to both facilitate the finer scale object segmentation and obtain greater classification accuracy (Figure 4). We compared the agreement of two approaches - visual interpretation and object-based - to estimate the proportion cover of landscape features within delineated patches, and investigated the spatial patterns of patches with large disagreement between the two approaches. To learn more about this comparison please see Zhou, Schwarz and Cadenasso (2010) under the publications tab. We are currently working to publish a more detailed description of the methods used to delineate and classify HERCULES patches. For additional information please see the urban design working group page and the featured blog post by Victoria Marshall entitled "Patch Dynamics: Urban Design and Ecology as Mosaic".
The Urban Design Working Group (UDWG) members Brian McGrath, Victoria Marshall and Phanat Xanamane first developed a matrix showing the array of possible HERCULES patches (Cadenasso, M.L., S.T.A. Pickett, and K. Schwarz. 2007. Spatial heterogeneity in urban ecosystems: Reconceptualizing land cover and a framework for classification. Frontiers in Ecology and Evolution 5: 80-88.). The matrix consists of patch "family" clusters which are arranged from left to right from built to vegetated dominated and from top to bottom from homogenous to heterogeneous in terms of landcover mix. (Figure 1) Figure 2 illustrates a forest (coarse vegetation) dominated patch, Figure 3 illustrates a grass (fine vegetation) dominated patch, Figure 4 illustrates a bare soil dominated patch, Figure 5 illustrates a pavement dominated patch, while Figure 6 illustrates a building dominated patch. The graph in Figure 7 extrudes the number of patches in the Gwynns Falls Watershed within each family. In modeling the prevalence of certain landcover patch types in the Gwynns Falls Watershed, the UDWG has created a landcover "signature" for Gwynns Falls. Figure 8 illustrates the presence of coarse vegetation patches greater than 75%, Figure 9 illustrates the presence of fine vegetation patches greater than 75%, Figure 10 illustrates the presence of building dominated patches between 50 and 75%, while Figure 11 illustrates the co-presence of building and pavement dominated patches between 25 and 50%. Figure 12 illustrates the how various landcover patches interact with the topography of Gwynns Falls in vertical sections cut transversally through the watershed at four BES water gauging locations.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12