Accepted Preprint
Original Article
Integrated Environmental Assessment and Management DOI 10.1002/ieam.1944
Modeling aesthetics to support an ecosystem services approach for natural resource management decision making†
Running title: Modeling aesthetic ecosystem services Pieter N. Booth*1
, Sheryl A. Law1¥, Jane Ma2, John Buonogurio3, James Boyd4 and Jessica Turnley5
1
Ramboll Environ, 901 Fifth Avenue, Suite 2820, Seattle, WA 98164
2
Exponent, 15375 SE 30th Place, Suite 250, Bellevue, WA 98007, Email: [email protected]
3
Exponent, 1800 Diagonal Road, Suite 500, Alexandria, VA 22314, Email: [email protected]
4
Resources For the Future, 1616 P Street NW, Washington, DC 20036, Email: [email protected]
5
Galisteo Consulting, 4004 Carlisle Blvd NE, Suite H, Albuquerque, NM 87101, Email: [email protected]
†
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1002/ieam.1944] All Supplemental Data may be found in the online version of this article at the publisher’s website.
This article is protected by copyright. All rights reserved Submitted 16 Auguest 2016; Returned for Revision 12 April 2017; Accepted 17 April 2017
This article is protected by copyright. All rights reserved
Accepted Preprint
*Address correspondence to: Ramboll Environ, 901 Fifth Avenue, Suite 2820, Seattle, WA 98164 Email: [email protected] Tel: 206 336 1656 Fax: 206 336 1651 ¥
Email: [email protected]
This article is protected by copyright. All rights reserved
Accepted Preprint
ABSTRACT This paper reviews literature on aesthetics and describes the development of vista and landscape aesthetics models. Spatially explicit variables were chosen to represent physical characteristics of natural landscapes that are important to aesthetic preferences. A vista aesthetics model evaluates the aesthetics of natural landscapes viewed from distances over 1,000 m. and a landscape aesthetics model evaluates the aesthetic value of wetlands and forests within 1,000 m from the
viewer. Each of the model variables is quantified using spatially-explicit metrics on a pixelspecific basis within EcoAIM™, a GIS-based ecosystem services decision analysis support tool.
Pixel values are “binned” into ranked categories and weights are assigned to select variables to represent stakeholder preferences. The final aesthetic score is the weighted sum of all variables and is assigned ranked values from 1 to 10. Ranked aesthetic values are displayed on maps by
patch type and integrated within EcoAIM™. The response of the aesthetic scoring in the models was tested by comparing current conditions in a discrete area of the facility with a Development scenario in the same area. The Development scenario consisted of two 6-story buildings and a trail replacing natural areas. The results of the vista aesthetic model indicate that the viewshed
area variable had the greatest effect on the aesthetics overall score. Results from the landscape aesthetics model indicate a 10% increase in overall aesthetics value, attributed to the increase in landscape diversity. The models are sensitive to the weights assigned to certain variables by the user and these weights should be set to reflect regional landscape characteristics as well as
stakeholder preferences. This demonstration project shows that natural landscape aesthetics can
be evaluated as part of a nonmonetary assessment of ecosystem services, and a scenario-building exercise provides end users with a tradeoff analysis in support of natural resource management decisions.
This article is protected by copyright. All rights reserved
Accepted Preprint
Key Words: aesthetics, ecosystem services, decision analysis, geospatial modelling
This article is protected by copyright. All rights reserved
Accepted Preprint
INTRODUCTION
The U.S. Department of Defense (DOD) has stewardship responsibilities for more than
30 million acres of land (Hewett 2008). Natural areas comprise a large proportion of these lands, and are critically important to the success and sustainability of military missions. Military activities and land use affect natural resources and their ecosystem services (ES). DoD, via their
Environmental Technology Demonstration and Validation Program (ESTCP) sponsored demonstration projects to determine how analysis of ES might better inform land management decisions at military installations to ensure a sustainable balance between ES preservation and mission prerogatives. The work presented in this paper was part of a larger demonstration project
of ES tradeoff modelling using the modelling platform EcoAIMTM at U.S. Army Aberdeen Proving Ground (APG) in Maryland (see Booth et al. 2014) (Figure 1 provides a map showing the location of APG).
APG has been an active military facility since 1918, and the size of the installation allows
for extensive research, development and testing of materials, vehicles, ordnance, and weaponry (APG 2009). The installation is home to more than 70 different Garrison-Support Organizations (GSOs) involved with various military and scientific research, testing, and development, such as Army Public Health Command, Army Intelligence, and Army Research and Development. It also hosts 11 major commands. Because APG is an active test and evaluation installation, there are several geographical areas of the installation for which geospatial information was not provided for security reasons, and consequently, for which the demonstration was not conducted. This article is protected by copyright. All rights reserved
Accepted Preprint
Undeveloped areas of the installation consist of 50% hardwood forest, 34%
mowed/grassy areas, 13% marsh or marsh shrub, 2% bare earth, and 1% shrub habitat. Forest areas are important not just to APG, but also regionally because of the loss of forest in the Chesapeake Bay watershed due to development. Most of the forested area is located within APG’s secured area and forest patches range in size from less than 1 hectare to more than 100
ha. and consist of early successional to mature forests. The most ecologically important forested areas are un-fragmented and riparian forests, which support a great variety of wildlife. The most common forest type includes sweet gum/water oak, mixed oak, and yellow poplar/transition hardwoods. Wetlands constitute about 2,226 ha at APG and include emergent, forested, and scrub/shrub wetlands (APG 2011). During structured stakeholder engagement at APG, it was determined that the aesthetics
of natural landscapes is an important ES to the sustainability of APG’s stated notional mission to attract and retain top talent. This notional mission was articulated as part of an overall mission at APG which is to be a work center for world-class research and development. For this demonstration project, aesthetics was incorporated as an ES into an existing decision analysis support tool (EcoAIM™) that uses geospatial analysis of land use/land cover (LU/LC) changes to assess tradeoffs in other ES such a nutrient sequestration in wetlands, recreation potential, and biodiversity support (see Booth et al. 2014). This paper details the model background and
development and discusses how aesthetics analysis can inform natural resources management decision-making. Aesthetics is classified as a cultural ecosystem service and valued because it fosters
inspiration, social relations, a sense of place, recreational opportunities, and cultural heritage (MEA 2005). In this paper, we refer to “aesthetics” as the beauty and appealing qualities
This article is protected by copyright. All rights reserved
Accepted Preprint
provided by natural landscapes. “Aesthetics” is synonymous with terms such as “scenic beauty,” “visual landscapes,” “landscape qualities,” “aesthetic benefits” and other terms used in diverse
publications. Favorable aesthetics in the workplace, such as natural views from office spaces is an
important element in both the physical and psychological well-being of employees, and among other things has been shown to result in decreased stress levels (e.g., Bolund and Hunhammar
1999), and decreases in sick leave taken (e.g., Ulrich 1979; Lottrup et al. 2012). Natural areas can play a large role in efforts to attract employees (Bolund and Hunhammer 1999) and they can
provide informal and neutral gathering areas for the exchange of ideas among employees. Beneficiaries at APG (natural resource personnel and higher ranking officials) stated that
the aesthetics of the natural areas at the installation play a key role in the notional mission to attract and retain talent. The aesthetics component of natural resources management has been discussed as an important aspect of landscaped areas that are frequented by visitors and personnel at APG, especially areas near the gate entrances and cantonment areas (where most personnel and offices are located). The objective of this study is to support APG’s use of natural landscape aesthetics as one
of the key ways to attract and retain talent at the installation. The vision of APG leadership is to provide an aesthetically pleasing cantonment area, which consists of patches of forest, wetlands, maintained lawns, office buildings, roads, and parking lots. Although there is a significant body of research on natural landscape aesthetics, including development of models relating
biophysical attributes to aesthetic preferences, there has been very little work done to use geospatial analysis of typically mapped land use land cover features to assess relative changes in aesthetics resulting from land use changes (see Background and Model Development below). This article is protected by copyright. All rights reserved
Accepted Preprint
To achieve the objective of this exercise, two aesthetics models were created to combine
the measurable biophysical features of natural areas with the preferences of ecosystem service beneficiaries for certain landscape aesthetics. T This paper discusses: (1) the development of two
aesthetics models, a landscape aesthetics model to capture near field aesthetic value and a vista aesthetics model to capture far field aesthetics value; (2) the results of model application to a baseline case and a hypothetical Development scenario; and (3) the role of the models in stakeholder communications and natural resource management decisions.
BACKGROUND AND MODEL DEVELOPMENT
Unlike some ecosystem services that can be quantified on the basis of objective units of
measure (e.g., mass of pollutants sequestered per acre per year as a measure of the regulating service provide by a wetland), the concept of aesthetics is more difficult to quantify because it is subjective and driven by personal preferences. Researchers suggest that understanding the visual qualities of preferred landscapes and their emotional attachments to viewers would help achieve sustainability goals (e.g., Schadler et al. 2013). The concept that aesthetics can be quantified has been approached in several ways by
researchers. Daniel (2001) reviewed the history of quantifying landscape quality by comparing “expert” and “perception-based” approaches. The former translates the biophysical features of the landscape into design parameters such as “form, line, variety, and unity” as indicators of aesthetic quality. Studies based on the “expert” approach are often used in forestry management (Daniel and Meitner 2001). The perception-based approach is based on the premise that natural landscape aesthetics will evoke psychological responses such as “mystery, and prospect-refuge” This article is protected by copyright. All rights reserved
Accepted Preprint
(feelings of discovery or finding safety) (Daniel 2001; Rudell et al. 1989). In most case studies reviewed, survey participants were shown pictures of landscapes and asked to rate or rank their preferences of the scenery (Clay and Daniel 2000; Daniel and Boster 1976). Studies in recent years have sought to combine the two approaches, such that people’s
preferences are translated into the biophysical attributes used for experts’ judgments. Various indicators or metrics measured using GIS tools are related (either statistically or qualitatively) to the overall aesthetic preferences by stakeholders. For example, Ritters et al. (1995) evaluated 55 metrics that describe landscape patterns and structure. A principal component analysis determined that metrics that describe patch size, shape, density, and heterogeneity can be used for long-term monitoring of landscape conditions and as an indicator for aesthetic preferences. Ode et al. (2008) assessed a different set of indicators (e.g., complexity, coherence, and
disturbance) in a proposed framework meant to capture landscape visual character. Plexida et al. (2014) used the spatial pattern analysis program FRAGSTATs (McGarigal, et al. 2012) and found that 10 metrics were significant in measuring spatial heterogeneity. Schirpke et al. (2013) evaluated several landscape metrics with a stepwise linear regression model and found that shape
complexity and landscape diversity were positively related to visual quality. Sakieh et al. (2016) develop an informed modelling approach to assess the aesthetic value of urban landscapes that includes several spatially explicit variables including topographic variability, vegetation density and ecotone, and vegetation type and diversity. These authors used Analytical Heirarchy Process to derive preference weights and determined that topographic variability and vegetation density account for over 60% of the preference values among 8 attributes modelled. Given that the existing research indicates that aesthetics are founded on the physical
landscape, we have concluded that by measuring spatially explicit attributes that define the
This article is protected by copyright. All rights reserved
Accepted Preprint
landscape, we can describe aesthetic preferences on a spatial scale. Many important attributes that define landscape aesthetic quality can be directly derived from widely available geospatial data such as land cover classifications. In cases where such data do not exist or are outdated, land cover classifications can be derived and quantified from remote sensing data such as satellite images or georeferenced aerial photography. In the present case, a variety of geospatial data were readily available from public sources
as well as LU/LC (thematic) maps developed by natural resource managers at APG. For modelling approaches of this nature it is important to determine whether the accuracy of the underlying data used to generate the thematic maps is adequate to support decision making. For this demonstration project, detailed GIS layers were available that were generated from intensive surveys by APG staff, therefore it was determined that ground-truthing of the thematic maps was not needed. However we did perform a site-specific accuracy assessment by determining the concurrence between thematic maps and reference data consisting of APG’s field data and high resolution satellite images. Concurrence was determined by randomly generating a set of 60 pixel locations in ArcGIS for land use/land cover features in the thematic map (e.g., wetland, forest, agriculture). A pixel in a LU/LC thematic map is assessed as “accurate” if the dominant feature for the pixel area is correctly identified compared to reference data. The threshold for meeting the quantitative criteria is >85% concurrence between thematic maps and reference data (e.g., McCormick 1999; Scepan 1999; Wulder et al. 2006) and is seen by many as a universal standard for thematic mapping in remote sensing (e.g., Fisher and Langford 1996; Weng 2002; Rogan et al. 2003; Bektas and Goksel 2004). An aerial photo for 2013 was compared to the LU/LC GIS layer (2007), because it is the underlying base layer for the aesthetic models. Of the
This article is protected by copyright. All rights reserved
Accepted Preprint
60 points selected from the LU/LC map, 52 correctly identified the LU/LC from the aerial photo for a concurrence of 88.3%
METHODOLOGY A review of human aesthetic preference studies indicates that landscape features can be
separated into two categories based on proximity to the viewer: (1) “vista” aesthetics and (2)
“landscape aesthetics. Vista aesthetics are determined by features perceived at large distances out to the horizon. Such views can be from the ground level, or from a certain height, such as looking out a window in a building or standing on a hill. Landscape aesthetics are determined by features perceived by a viewer within a much closer range, typically by viewers standing within a natural area and looking at a landscape patch around them (Rolston 1995; Bishop and Hulse 1994). The landscape aesthetic category reflects the experiential aspect of a viewer interacting with and immersed in the natural environment. The vista and landscape models are linked to the EcoAIM™ modelling platform but can also be run independently. For the purposes of this study,
the landscape aesthetics model is constrained to evaluating features within 1,000-m of the observer and the vista aesthetics model is constrained to evaluating features greater than 1,000-m
from the observer. These constraints were determined based on professional judgment and knowledge about the study area and can be modified by the model user to accommodate sitespecific characteristics (e.g., large vertical features such as mountain ranges) and user preferences.
This article is protected by copyright. All rights reserved
Accepted Preprint
Vista Aesthetics Model
Studies of vista aesthetics or “viewsheds” have typically focused on specific geographic
locations and habitat types (e.g., Appalachian Mountains [Hammitt et al. 1994], Dutch countrysides [de Vries et al. 2007], the Alps in central Europe [Grêt-Regamey et al. 2007], and spatial patterns in urbanized lands [Sakieh et al. 2016]). Despite the differences in locations and types, three main variables emerge as important aspects of vista aesthetics: heterogeneity, size,
and diversity of features. Studies of what makes natural areas aesthetically pleasing often cite variation as a key part of the experience. Landscape variations can include different colors, textures, forms, and densities of objects. For example, Tyrvainen et al. (2005) found that people
valued large spaces that were broken up and defined and framed by discrete foregrounds and backgrounds more than homogeneous landscapes. The theory that three main characteristics can
describe vista aesthetics was translated into three measurable attributes in EcoAIM™ as follows. Patch Richness: Patch richness is the number of LU/LC patch types in the field of view
where patch types correspond to types of LU/LC in the GIS database--this approach is consistent with that applied by Schirpke et al. (2013). The model returns a count of the number of patch types based on the data set. The model allows weights to be assigned by the user to correspond to the relative ranking of viewers’ preferences for patch types . This enables the tool be customized for application in a wide range of environmental and social settings. For example, a user in a savannah environment may have a higher preference for views of patches of trees or forests than a user in a forest environment. If the user applied preference weights to patch types, the patch richness metric is calculated as the weighted sum of the different patches in the field of view.
This article is protected by copyright. All rights reserved
Accepted Preprint
Diversity: Landscape heterogeneity and patch variation is an important variable in
aesthetics (Ritters et al. 1995; Plexida et al. 2014). In a study by Schirpke et al. (2013), participants preferred to see a variety of landforms, such as buildings, forests, and water, and in different proportions. This variable is measured using the Shannon Diversity Index calculation, which reflects different patch type amounts and accounts for the proportion that each patch type constitutes in the field of view. The more patch types there are with varying proportions, the
greater the Shannon Diversity Index value. The calculation is as follows: Shannon Diversity Index
where S is the total number of patch types in the data set and pi is the proportion of S made up of the ith patch. Viewshed Area: The viewshed area is the area of land (e.g., in acres or square meters to
be selected by the model user) that is visible by an unimpeded line of sight from a single point (Grêt-Regamey et al. 2007). In general, studies found that participants preferred to see more, rather than less in their field of view or have unimpeded views, as opposed to blocked views (Ritters et al. 1995). This variable is measured in GIS with a digital elevation model (DEM) by
accounting for the location and elevation of the observer, the angle of view, and obstructions to the view. The point of view is an important component of aesthetics (Grêt-Regamey et al. 2007).
Accordingly, the EcoAIM™ tool allows for choosing locations on a map and defining the
resulting field of view. The viewshed is delineated by drawing a line on the DEM with the observation point as the starting point of the line and the line direction as the desired direction of This article is protected by copyright. All rights reserved
Accepted Preprint
view by the observer. For the model run presented here, the default elevation is 4.6 m (15 ft) above ground level based on the ground elevation (from the DEM) at the point of observation. However, the user can designate other elevations (e.g., 3 or 6 m above ground elevation) to simulate a viewshed from an office window on a second or third floor. The view angle is set to a
default value of 90° to simulate a general view that does not include peripheral vision; however the default view can be set from 0-360 degrees by the model user. The tool extracts visible pixels of land use in the viewshed area and calculates the patch richness, Shannon Diversity Index, and viewshed area, all of which are necessary for characterizing the vista aesthetics quality. All else being equal, the larger the viewshed area, the better the vista aesthetics quality. Viewshed area is largely a function of elevation above ground, ground topography, and other obstructions to views (forest or buildings). In its default mode, the modelled attributes contribute equally to the overall aesthetics score; however, the user can assign preference weights in the model interface (e.g., 05) to selected attributes such as LU/LC or patch type to reflect stakeholder preference or sitespecific landscape features.
Landscape Aesthetics Model
Studies of the aesthetics of natural landscapes in close proximity to the viewer have
identified certain characteristics as being preferred by viewers, such as diversity, size, contrast, heterogeneity, and “naturalness”. In addition, landscapes with more preferable aesthetics are
those with a combination of less development, greater number of distant views, a more varied terrain, presence of water, and a sense of engagement with the environment (Steinitz 2001;
This article is protected by copyright. All rights reserved
Accepted Preprint
Shang and Bishop 2000). For this application of EcoAIMTM, the landscape aesthetics model was developed for two LU/LC patches that predominate at APG: forests and wetlands. In addition to their considerable ecological value, wetlands provide culturally important
services. Wetlands with public access are also valued for their educational and recreational potential. Smardon (1983) found that study participants prefer wetlands with more open waters
compared to highly vegetated sites. Forests, in general, are considered to be good representations of natural areas. Nassauer
(1995) found that survey respondents indicated that areas that are too formal, open, bare, flat, or monotonous and lack trees are unattractive. The forest condition is also important, with people preferring a “natural” stand compared to areas that have been recently harvested, tree plantations where there is a poorly developed under- and mid-story, or trees planted in rows. The “mystery”
associated with forest openings is considered a positive attribute of forest landscapes (Herzog and Miller 1998; Herzog and Bryce 2007). The variables measured for each forest and wetland patch include the following. Landform contrast: Differences in vertical stratification of landforms are cited by several
investigators as important contributors to the aesthetic value of landscapes (e.g., Sakieh et al 2016; Junge et al. 2014). In particular, Smardon (1983) found that steep slopes on the sides of wetlands provide landform contrast which makes a wetland more aesthetically pleasing than
wetlands with gradual slopes and minimal contrast with surrounding areas. Similarly, people prefer perceiving a distinct difference between adjacent patches, such as forested and open areas (Smardon and Fabos 1983). The landform contrast variable is the absolute value of the difference between the elevations (in meters) of the wetland/forest patch and the adjacent landform. The landform contrast variable is calculated in two steps. First, relative relief was calculated as the
This article is protected by copyright. All rights reserved
Accepted Preprint
absolute difference between the average elevation of a wetland and its adjacent landform (100-m buffer ring). Second, the land contrast variable is calculated by dividing the relative relief (meter) by the average width (meter) of the wetland.
Edge complexity: Forests with maximized edge-to-area ratios are perceived as more
aesthetically pleasing than those with lower edge-to-area ratios. The shape of a forest patch is assumed to indicate the intensity of human impact. For example, people prefer forest patch edges to be undulating, or sinuous, making them less noticeable and more “natural” compared to straight edges (that suggest maintenance by humans) (Gobster 1999; Frank et al. 2013; Smardon
1983). The calculation to measure the sinuosity of the wetland/forest edge is:
where P = perimeter and A = area (Smardon and Fabos 1983).
Surrounding land-cover contrast: People prefer variation in the scenery as they travel
along a corridor (Ribe 1989). Wetlands with surrounding contrasting upland landforms are preferred because the uplands define a sharp visual image of the wetland and provide a feeling of enclosed space. Through preference testing, wetlands that are adjacent to open water, forest, and agricultural lands were deemed most visually appealing (Smardon 1983). The surrounding landcover contrast variable is measured as the number of different land cover types found within a 100-m buffer delineated around each forest or wetland. The 100-m buffer corresponds to an assumed “experiential” aesthetics envelope of 100-m from the point of observation. Although no studies cite the 100-m point, it was assumed this is an appropriate distance at which an observer could see in the distance and is consistent with the selection of constraining distances for the two
This article is protected by copyright. All rights reserved
Accepted Preprint
models The distance constraint can be modified by the use based on site-specific attributes such as significant elevation changes or viewpoint of the observer from a promontory, where a larger buffer are may be more appropriate. Surrounding land-cover diversity: The preference for seeing different types of land cover
varies by individuals and is typically assessed during the stakeholder implementation phase of a study of this type. Researchers have studied the importance of psychological factors and stakeholder attitudes that influence preferences for landscape types (Ribe 1989; Ribe 2009; Gobster and Chenoweth 1989; Kearney and Bradley 2011). Thus, the model allows weights to be assigned to land use or land cover types by the user to represent preferences. Incorporation of
preferences is an important functionality to account for the widely varied landscapes occupied by DOD installations nationwide (e.g., desert, high mountain, shoreline/coastal, lowland mixed
forest, western fir forests, etc.). For example, whereas open water may rank highest in a coast/shoreline site like APG, it would be irrelevant for desert sites like White Sands, New Mexico, or the Yakima Training Center in Washington State. Surrounding land cover diversity is measured as the weighted sum of the total different land cover types within a 100-m buffer delineated around each forest or wetland. Patch Size: In contrast to vista aesthetics, where diversity of patch types is important,
for landscape aesthetics, people generally prefer experiencing large patches of land rather than
small patches. Studies have determined that the largest wetlands are the most aesthetically pleasing, possibly because wetland size is related to recreational value and increased biodiversity (Smardon 1983). Similarly, large forest tracts are also more desirable than small patches. Size is a measure of the total area of the wetland or forest patch.
This article is protected by copyright. All rights reserved
Accepted Preprint
Vegetative Interspersion: The surface pattern or “texture” of a wetland is an important
aspect of aesthetics. Surveys show that people prefer completely interspersed, grouped vegetation patterns on the water surface more than completely vegetated waters. Vegetative interspersion is a count of the total number of pixels along the perimeters of vegetation groups within each wetland. Similarly, people prefer forested areas with “natural-looking” groupings of trees over evenly spaced plantation-like forests (Ribe 1989). If detailed forest survey information is available, vegetative interspersion is measured as the number of different tree species within each forest patch; otherwise, this variable is measured the same way as wetlands. Forest density: Stamps (2010) suggested that the concept of ‘mystery’ or the distance
which people can see through the vegetation is extremely important to aesthetic preferences. Study participants who were shown pictures had greater appreciation of forests with greater canopy cover and areas with smaller openings (Gobster 1999). Forest density is a measure of the percentage of canopy cover in each forest patch. Water-body size: Views than contain water bodies provide a sense of being most natural,
attract and hold a viewer’s attention, and create a calming effect (Nasar and Li 2004). For wetlands, as well as with other water bodies, the larger the water area, the more aesthetically pleasing the water body is assumed to be, with all other attributes being equal. Water-body size is a measure of the area of water inside a wetland patch boundary. Associated water-body diversity: Wetlands that are adjacent to other water bodies, (i.e.,
rivers, small lakes, ponds, and saltwater bays and inlets) are optimum environments from a visual perspective (Smardon 1983). Associated water body diversity is a count of the total number of open water bodies (excluding other wetlands) within a 100-m buffer delineated around the wetland. This article is protected by copyright. All rights reserved
Accepted Preprint
Data Processing Each of the landscape aesthetics variables (described above) is measured for each LU/LC
patch, quantified, and plotted on a distribution. The Jenks natural breaks classification method within the ArcGIS® software is used to categorize (data bin) the values within a distribution. This is a data clustering, optimization method commonly used in ArcGIS which runs iterations of the data (patch variables in this study) such that the average variance is minimized for patches within a group and maximized between groups. By grouping patches with this method, ten
distinct categories of ES values can be created and displayed on maps. The landscape variable values for each patch are averaged and processed with a second
run with Jenks classification. Each habitat patch is then classified in one of ten categories and
color-coded for mapping purposes (red = 1 [worst] to green = 10 [best]). As with the vista aesthetics model, in the model’s default mode, the attributes contribute equally to the overall landscape aesthetics score; however, the user can assign preference weights (e.g., 0-5) to selected attributes to reflect stakeholder preference or site-specific landscape features.
Statistical Analysis
To assess whether changes in either vista or landscape aesthetics values were significant
between a baseline condition and a hypothetical development condition (the Development scenario), statistical methods were applied to measure the differences in aesthetics scores. To
avoid potentially violating assumptions of normality of the data distributions, a two-tailed nonparametric Wilcoxon signed rank test on paired observations was used to determine whether the difference between the Baseline case and the Development scenario is significantly different This article is protected by copyright. All rights reserved
Accepted Preprint
from zero at p Timmins, K.B. Jones and B.L. Jackson. 1995. A factor analysis of landscape pattern and structure metrics. Landsc. Ecol., 10(1): 23-39. Rogan, J., J. Miller, D. Stow, J. Franklin, L. Levien and C. Fischer. 2003. Land-cover change monitoring with classification trees using Landsat TM and ancillary data. Photogram. Eng. Remote Sensing, 69:793–804. Rolston, H. III. 1995. Does aesthetic appreciation of landscapes need to be science-based? British Journal of Aesthetics, 35(4): 374-386. Rudell, E.J., J.H. Gramann, V. A. Rudis, and J.M. Westphal. 1989. The psychological utility of visual penetration in near-view forest scenic-beauty models. Environment and Behavior, 21(4): 393-412. Sakieh, Y., A. Salmanmahiny, S. Hamed Mirkarimi, and S. Saeidi. 2016. Measuring the relationships between landscape aesthetics suitability and spatial patterns of urbanized lands: an informed modelling framework for developing urban growth scenarios. Geocarto Intl. Published online 2 May 2016. http://www.tandfonline.com/doi/abs/10.1080/10106049.2016.1178817 Schädler, S., M.Finkel, A. Bleicher, M. Morio and M. Gross. 2013. Spatially explicit computation of sustainability indicator values for the automated assessment of land-use options. Landsc. Urban Plann., 111: 34-45. Schirpke, U., E. Tasser, and U. Tappeiner. 2013. Predicting scenic beauty of mountain regions. Landsc. Urban Plann., 111: 1-12. Scepan, J. 1999. Thematic validation of high-resolution global land-cover data sets. Photogram. Eng. Remote Sensing, 65:1051–1060.
This article is protected by copyright. All rights reserved
Accepted Preprint
Shang, H. and I.D. Bishop. 2000. Visual thresholds for detection, recognition and visual impact in landscape settings. Journal of Environmental Psychology, 20: 125-140. Smardon, R. C. 1983. State of the art in assessing visual-cultural values.
Smardon, R.C. and J.G. Fabos. 1983. Chapter 9. A model for assessing visual-cultural values of wetlands: A Massachusetts case study. Stamps, A.E. 2010. Effects of permeability on perceived enclosure and spaciousness. Environment and Behavior, 42(6): 864-886. Steinitz, C. 2001. Visual evaluation models: some complicating questions regarding memorable scenes. Landsc. Urban Plann., 54: 283-287. Tyrvainen, L., S. Pauleit, K. Seeland and S. de Vries. 2005. Chapter 4. Benefits and uses of urban forests and trees. Urban Forests and Trees, 81-114. Ulrich, R.S. 1979. Visual landscapes and psychological well-being. Landscape Research, 4(1): 17-23. Weng, Q. 2002. Land use change analysis in the Zhujiang delta of China using satellite remote sensing, GIS and stochastic modelling. J. Environ. Manage., 64:273–284. Wulder, M.A., S.E. Franklin, J.C. White, J. Linke, and S. Magnussen. 2006. An accuracy assessment framework for large-area land cover classification products derived from mediumresolution satellite data. Intl. J. Remote Sensing, 27:663–683.
This article is protected by copyright. All rights reserved
Accepted Preprint
LIST OF FIGURES
Figure 1. Location of U.S. Army’s Aberdeen Proving Ground
Figure 2. a) Site-wide wetlands landscape aesthetics under baseline conditions; and b) Site-wide forest landscape aesthetics under baseline conditions.
Figure 3. Vista aesthetics model screen shots under Baseline and the Development scenario.
Figure 4. Landscape aesthetics model output for wetlands under Baseline and the Development scenario.
Figure 5. Landscape aesthetics model output for forests under Baseline and the Development scenario.
This article is protected by copyright. All rights reserved
Accepted Preprint
LIST OF TABLES Table 1. Statistical analysis of model results: Landscape Aesthetics--Forest Table 2. Statistical analysis of model results: Landscape aesthetics—Wetland
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 1
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 2(a)
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 2(b)
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 3
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 4
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 5
This article is protected by copyright. All rights reserved
Accepted Preprint
Table 1. Statistical analysis of model results: Landscape Aesthetics--Forest Surrounding Surrounding Landform Edge Land Cover Land Cover Contrast Complexity Contrast Diversity
Patch Size
Vegetation Interspersion
Forest Density
V.stat 56 5 7.5 26 10 16 p.value 0.484642 0.422678 0.589774 0.287636 0.100348 0.821098 sig.diff no no no no no no no
This article is protected by copyright. All rights reserved
5 1
overall 183 0.018729 yes
Accepted Preprint
Table 2. Statistical analysis of model results: Landscape Aesthetics--Wetland
Landform Contrast
Edge Complexity
Waterbody Size
Water Body Diversity
Surrounding Land Cover Contrast
Surrounding Land Cover Diversity
Wetland Size
VegetationWater Interspersion
overall
V.stat 106 5 9 3.5 7 7 6 17 193 p.value 0.985101 0.422678 0.78353 0.343028 0.577469 0.577469 0.181449 0.000649 0.000502 sig.diff no no no no no no no yes yes
This article is protected by copyright. All rights reserved
Original Article
Integrated Environmental Assessment and Management DOI 10.1002/ieam.1944
Modeling aesthetics to support an ecosystem services approach for natural resource management decision making†
Running title: Modeling aesthetic ecosystem services Pieter N. Booth*1
, Sheryl A. Law1¥, Jane Ma2, John Buonogurio3, James Boyd4 and Jessica Turnley5
1
Ramboll Environ, 901 Fifth Avenue, Suite 2820, Seattle, WA 98164
2
Exponent, 15375 SE 30th Place, Suite 250, Bellevue, WA 98007, Email: [email protected]
3
Exponent, 1800 Diagonal Road, Suite 500, Alexandria, VA 22314, Email: [email protected]
4
Resources For the Future, 1616 P Street NW, Washington, DC 20036, Email: [email protected]
5
Galisteo Consulting, 4004 Carlisle Blvd NE, Suite H, Albuquerque, NM 87101, Email: [email protected]
†
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1002/ieam.1944] All Supplemental Data may be found in the online version of this article at the publisher’s website.
This article is protected by copyright. All rights reserved Submitted 16 Auguest 2016; Returned for Revision 12 April 2017; Accepted 17 April 2017
This article is protected by copyright. All rights reserved
Accepted Preprint
*Address correspondence to: Ramboll Environ, 901 Fifth Avenue, Suite 2820, Seattle, WA 98164 Email: [email protected] Tel: 206 336 1656 Fax: 206 336 1651 ¥
Email: [email protected]
This article is protected by copyright. All rights reserved
Accepted Preprint
ABSTRACT This paper reviews literature on aesthetics and describes the development of vista and landscape aesthetics models. Spatially explicit variables were chosen to represent physical characteristics of natural landscapes that are important to aesthetic preferences. A vista aesthetics model evaluates the aesthetics of natural landscapes viewed from distances over 1,000 m. and a landscape aesthetics model evaluates the aesthetic value of wetlands and forests within 1,000 m from the
viewer. Each of the model variables is quantified using spatially-explicit metrics on a pixelspecific basis within EcoAIM™, a GIS-based ecosystem services decision analysis support tool.
Pixel values are “binned” into ranked categories and weights are assigned to select variables to represent stakeholder preferences. The final aesthetic score is the weighted sum of all variables and is assigned ranked values from 1 to 10. Ranked aesthetic values are displayed on maps by
patch type and integrated within EcoAIM™. The response of the aesthetic scoring in the models was tested by comparing current conditions in a discrete area of the facility with a Development scenario in the same area. The Development scenario consisted of two 6-story buildings and a trail replacing natural areas. The results of the vista aesthetic model indicate that the viewshed
area variable had the greatest effect on the aesthetics overall score. Results from the landscape aesthetics model indicate a 10% increase in overall aesthetics value, attributed to the increase in landscape diversity. The models are sensitive to the weights assigned to certain variables by the user and these weights should be set to reflect regional landscape characteristics as well as
stakeholder preferences. This demonstration project shows that natural landscape aesthetics can
be evaluated as part of a nonmonetary assessment of ecosystem services, and a scenario-building exercise provides end users with a tradeoff analysis in support of natural resource management decisions.
This article is protected by copyright. All rights reserved
Accepted Preprint
Key Words: aesthetics, ecosystem services, decision analysis, geospatial modelling
This article is protected by copyright. All rights reserved
Accepted Preprint
INTRODUCTION
The U.S. Department of Defense (DOD) has stewardship responsibilities for more than
30 million acres of land (Hewett 2008). Natural areas comprise a large proportion of these lands, and are critically important to the success and sustainability of military missions. Military activities and land use affect natural resources and their ecosystem services (ES). DoD, via their
Environmental Technology Demonstration and Validation Program (ESTCP) sponsored demonstration projects to determine how analysis of ES might better inform land management decisions at military installations to ensure a sustainable balance between ES preservation and mission prerogatives. The work presented in this paper was part of a larger demonstration project
of ES tradeoff modelling using the modelling platform EcoAIMTM at U.S. Army Aberdeen Proving Ground (APG) in Maryland (see Booth et al. 2014) (Figure 1 provides a map showing the location of APG).
APG has been an active military facility since 1918, and the size of the installation allows
for extensive research, development and testing of materials, vehicles, ordnance, and weaponry (APG 2009). The installation is home to more than 70 different Garrison-Support Organizations (GSOs) involved with various military and scientific research, testing, and development, such as Army Public Health Command, Army Intelligence, and Army Research and Development. It also hosts 11 major commands. Because APG is an active test and evaluation installation, there are several geographical areas of the installation for which geospatial information was not provided for security reasons, and consequently, for which the demonstration was not conducted. This article is protected by copyright. All rights reserved
Accepted Preprint
Undeveloped areas of the installation consist of 50% hardwood forest, 34%
mowed/grassy areas, 13% marsh or marsh shrub, 2% bare earth, and 1% shrub habitat. Forest areas are important not just to APG, but also regionally because of the loss of forest in the Chesapeake Bay watershed due to development. Most of the forested area is located within APG’s secured area and forest patches range in size from less than 1 hectare to more than 100
ha. and consist of early successional to mature forests. The most ecologically important forested areas are un-fragmented and riparian forests, which support a great variety of wildlife. The most common forest type includes sweet gum/water oak, mixed oak, and yellow poplar/transition hardwoods. Wetlands constitute about 2,226 ha at APG and include emergent, forested, and scrub/shrub wetlands (APG 2011). During structured stakeholder engagement at APG, it was determined that the aesthetics
of natural landscapes is an important ES to the sustainability of APG’s stated notional mission to attract and retain top talent. This notional mission was articulated as part of an overall mission at APG which is to be a work center for world-class research and development. For this demonstration project, aesthetics was incorporated as an ES into an existing decision analysis support tool (EcoAIM™) that uses geospatial analysis of land use/land cover (LU/LC) changes to assess tradeoffs in other ES such a nutrient sequestration in wetlands, recreation potential, and biodiversity support (see Booth et al. 2014). This paper details the model background and
development and discusses how aesthetics analysis can inform natural resources management decision-making. Aesthetics is classified as a cultural ecosystem service and valued because it fosters
inspiration, social relations, a sense of place, recreational opportunities, and cultural heritage (MEA 2005). In this paper, we refer to “aesthetics” as the beauty and appealing qualities
This article is protected by copyright. All rights reserved
Accepted Preprint
provided by natural landscapes. “Aesthetics” is synonymous with terms such as “scenic beauty,” “visual landscapes,” “landscape qualities,” “aesthetic benefits” and other terms used in diverse
publications. Favorable aesthetics in the workplace, such as natural views from office spaces is an
important element in both the physical and psychological well-being of employees, and among other things has been shown to result in decreased stress levels (e.g., Bolund and Hunhammar
1999), and decreases in sick leave taken (e.g., Ulrich 1979; Lottrup et al. 2012). Natural areas can play a large role in efforts to attract employees (Bolund and Hunhammer 1999) and they can
provide informal and neutral gathering areas for the exchange of ideas among employees. Beneficiaries at APG (natural resource personnel and higher ranking officials) stated that
the aesthetics of the natural areas at the installation play a key role in the notional mission to attract and retain talent. The aesthetics component of natural resources management has been discussed as an important aspect of landscaped areas that are frequented by visitors and personnel at APG, especially areas near the gate entrances and cantonment areas (where most personnel and offices are located). The objective of this study is to support APG’s use of natural landscape aesthetics as one
of the key ways to attract and retain talent at the installation. The vision of APG leadership is to provide an aesthetically pleasing cantonment area, which consists of patches of forest, wetlands, maintained lawns, office buildings, roads, and parking lots. Although there is a significant body of research on natural landscape aesthetics, including development of models relating
biophysical attributes to aesthetic preferences, there has been very little work done to use geospatial analysis of typically mapped land use land cover features to assess relative changes in aesthetics resulting from land use changes (see Background and Model Development below). This article is protected by copyright. All rights reserved
Accepted Preprint
To achieve the objective of this exercise, two aesthetics models were created to combine
the measurable biophysical features of natural areas with the preferences of ecosystem service beneficiaries for certain landscape aesthetics. T This paper discusses: (1) the development of two
aesthetics models, a landscape aesthetics model to capture near field aesthetic value and a vista aesthetics model to capture far field aesthetics value; (2) the results of model application to a baseline case and a hypothetical Development scenario; and (3) the role of the models in stakeholder communications and natural resource management decisions.
BACKGROUND AND MODEL DEVELOPMENT
Unlike some ecosystem services that can be quantified on the basis of objective units of
measure (e.g., mass of pollutants sequestered per acre per year as a measure of the regulating service provide by a wetland), the concept of aesthetics is more difficult to quantify because it is subjective and driven by personal preferences. Researchers suggest that understanding the visual qualities of preferred landscapes and their emotional attachments to viewers would help achieve sustainability goals (e.g., Schadler et al. 2013). The concept that aesthetics can be quantified has been approached in several ways by
researchers. Daniel (2001) reviewed the history of quantifying landscape quality by comparing “expert” and “perception-based” approaches. The former translates the biophysical features of the landscape into design parameters such as “form, line, variety, and unity” as indicators of aesthetic quality. Studies based on the “expert” approach are often used in forestry management (Daniel and Meitner 2001). The perception-based approach is based on the premise that natural landscape aesthetics will evoke psychological responses such as “mystery, and prospect-refuge” This article is protected by copyright. All rights reserved
Accepted Preprint
(feelings of discovery or finding safety) (Daniel 2001; Rudell et al. 1989). In most case studies reviewed, survey participants were shown pictures of landscapes and asked to rate or rank their preferences of the scenery (Clay and Daniel 2000; Daniel and Boster 1976). Studies in recent years have sought to combine the two approaches, such that people’s
preferences are translated into the biophysical attributes used for experts’ judgments. Various indicators or metrics measured using GIS tools are related (either statistically or qualitatively) to the overall aesthetic preferences by stakeholders. For example, Ritters et al. (1995) evaluated 55 metrics that describe landscape patterns and structure. A principal component analysis determined that metrics that describe patch size, shape, density, and heterogeneity can be used for long-term monitoring of landscape conditions and as an indicator for aesthetic preferences. Ode et al. (2008) assessed a different set of indicators (e.g., complexity, coherence, and
disturbance) in a proposed framework meant to capture landscape visual character. Plexida et al. (2014) used the spatial pattern analysis program FRAGSTATs (McGarigal, et al. 2012) and found that 10 metrics were significant in measuring spatial heterogeneity. Schirpke et al. (2013) evaluated several landscape metrics with a stepwise linear regression model and found that shape
complexity and landscape diversity were positively related to visual quality. Sakieh et al. (2016) develop an informed modelling approach to assess the aesthetic value of urban landscapes that includes several spatially explicit variables including topographic variability, vegetation density and ecotone, and vegetation type and diversity. These authors used Analytical Heirarchy Process to derive preference weights and determined that topographic variability and vegetation density account for over 60% of the preference values among 8 attributes modelled. Given that the existing research indicates that aesthetics are founded on the physical
landscape, we have concluded that by measuring spatially explicit attributes that define the
This article is protected by copyright. All rights reserved
Accepted Preprint
landscape, we can describe aesthetic preferences on a spatial scale. Many important attributes that define landscape aesthetic quality can be directly derived from widely available geospatial data such as land cover classifications. In cases where such data do not exist or are outdated, land cover classifications can be derived and quantified from remote sensing data such as satellite images or georeferenced aerial photography. In the present case, a variety of geospatial data were readily available from public sources
as well as LU/LC (thematic) maps developed by natural resource managers at APG. For modelling approaches of this nature it is important to determine whether the accuracy of the underlying data used to generate the thematic maps is adequate to support decision making. For this demonstration project, detailed GIS layers were available that were generated from intensive surveys by APG staff, therefore it was determined that ground-truthing of the thematic maps was not needed. However we did perform a site-specific accuracy assessment by determining the concurrence between thematic maps and reference data consisting of APG’s field data and high resolution satellite images. Concurrence was determined by randomly generating a set of 60 pixel locations in ArcGIS for land use/land cover features in the thematic map (e.g., wetland, forest, agriculture). A pixel in a LU/LC thematic map is assessed as “accurate” if the dominant feature for the pixel area is correctly identified compared to reference data. The threshold for meeting the quantitative criteria is >85% concurrence between thematic maps and reference data (e.g., McCormick 1999; Scepan 1999; Wulder et al. 2006) and is seen by many as a universal standard for thematic mapping in remote sensing (e.g., Fisher and Langford 1996; Weng 2002; Rogan et al. 2003; Bektas and Goksel 2004). An aerial photo for 2013 was compared to the LU/LC GIS layer (2007), because it is the underlying base layer for the aesthetic models. Of the
This article is protected by copyright. All rights reserved
Accepted Preprint
60 points selected from the LU/LC map, 52 correctly identified the LU/LC from the aerial photo for a concurrence of 88.3%
METHODOLOGY A review of human aesthetic preference studies indicates that landscape features can be
separated into two categories based on proximity to the viewer: (1) “vista” aesthetics and (2)
“landscape aesthetics. Vista aesthetics are determined by features perceived at large distances out to the horizon. Such views can be from the ground level, or from a certain height, such as looking out a window in a building or standing on a hill. Landscape aesthetics are determined by features perceived by a viewer within a much closer range, typically by viewers standing within a natural area and looking at a landscape patch around them (Rolston 1995; Bishop and Hulse 1994). The landscape aesthetic category reflects the experiential aspect of a viewer interacting with and immersed in the natural environment. The vista and landscape models are linked to the EcoAIM™ modelling platform but can also be run independently. For the purposes of this study,
the landscape aesthetics model is constrained to evaluating features within 1,000-m of the observer and the vista aesthetics model is constrained to evaluating features greater than 1,000-m
from the observer. These constraints were determined based on professional judgment and knowledge about the study area and can be modified by the model user to accommodate sitespecific characteristics (e.g., large vertical features such as mountain ranges) and user preferences.
This article is protected by copyright. All rights reserved
Accepted Preprint
Vista Aesthetics Model
Studies of vista aesthetics or “viewsheds” have typically focused on specific geographic
locations and habitat types (e.g., Appalachian Mountains [Hammitt et al. 1994], Dutch countrysides [de Vries et al. 2007], the Alps in central Europe [Grêt-Regamey et al. 2007], and spatial patterns in urbanized lands [Sakieh et al. 2016]). Despite the differences in locations and types, three main variables emerge as important aspects of vista aesthetics: heterogeneity, size,
and diversity of features. Studies of what makes natural areas aesthetically pleasing often cite variation as a key part of the experience. Landscape variations can include different colors, textures, forms, and densities of objects. For example, Tyrvainen et al. (2005) found that people
valued large spaces that were broken up and defined and framed by discrete foregrounds and backgrounds more than homogeneous landscapes. The theory that three main characteristics can
describe vista aesthetics was translated into three measurable attributes in EcoAIM™ as follows. Patch Richness: Patch richness is the number of LU/LC patch types in the field of view
where patch types correspond to types of LU/LC in the GIS database--this approach is consistent with that applied by Schirpke et al. (2013). The model returns a count of the number of patch types based on the data set. The model allows weights to be assigned by the user to correspond to the relative ranking of viewers’ preferences for patch types . This enables the tool be customized for application in a wide range of environmental and social settings. For example, a user in a savannah environment may have a higher preference for views of patches of trees or forests than a user in a forest environment. If the user applied preference weights to patch types, the patch richness metric is calculated as the weighted sum of the different patches in the field of view.
This article is protected by copyright. All rights reserved
Accepted Preprint
Diversity: Landscape heterogeneity and patch variation is an important variable in
aesthetics (Ritters et al. 1995; Plexida et al. 2014). In a study by Schirpke et al. (2013), participants preferred to see a variety of landforms, such as buildings, forests, and water, and in different proportions. This variable is measured using the Shannon Diversity Index calculation, which reflects different patch type amounts and accounts for the proportion that each patch type constitutes in the field of view. The more patch types there are with varying proportions, the
greater the Shannon Diversity Index value. The calculation is as follows: Shannon Diversity Index
where S is the total number of patch types in the data set and pi is the proportion of S made up of the ith patch. Viewshed Area: The viewshed area is the area of land (e.g., in acres or square meters to
be selected by the model user) that is visible by an unimpeded line of sight from a single point (Grêt-Regamey et al. 2007). In general, studies found that participants preferred to see more, rather than less in their field of view or have unimpeded views, as opposed to blocked views (Ritters et al. 1995). This variable is measured in GIS with a digital elevation model (DEM) by
accounting for the location and elevation of the observer, the angle of view, and obstructions to the view. The point of view is an important component of aesthetics (Grêt-Regamey et al. 2007).
Accordingly, the EcoAIM™ tool allows for choosing locations on a map and defining the
resulting field of view. The viewshed is delineated by drawing a line on the DEM with the observation point as the starting point of the line and the line direction as the desired direction of This article is protected by copyright. All rights reserved
Accepted Preprint
view by the observer. For the model run presented here, the default elevation is 4.6 m (15 ft) above ground level based on the ground elevation (from the DEM) at the point of observation. However, the user can designate other elevations (e.g., 3 or 6 m above ground elevation) to simulate a viewshed from an office window on a second or third floor. The view angle is set to a
default value of 90° to simulate a general view that does not include peripheral vision; however the default view can be set from 0-360 degrees by the model user. The tool extracts visible pixels of land use in the viewshed area and calculates the patch richness, Shannon Diversity Index, and viewshed area, all of which are necessary for characterizing the vista aesthetics quality. All else being equal, the larger the viewshed area, the better the vista aesthetics quality. Viewshed area is largely a function of elevation above ground, ground topography, and other obstructions to views (forest or buildings). In its default mode, the modelled attributes contribute equally to the overall aesthetics score; however, the user can assign preference weights in the model interface (e.g., 05) to selected attributes such as LU/LC or patch type to reflect stakeholder preference or sitespecific landscape features.
Landscape Aesthetics Model
Studies of the aesthetics of natural landscapes in close proximity to the viewer have
identified certain characteristics as being preferred by viewers, such as diversity, size, contrast, heterogeneity, and “naturalness”. In addition, landscapes with more preferable aesthetics are
those with a combination of less development, greater number of distant views, a more varied terrain, presence of water, and a sense of engagement with the environment (Steinitz 2001;
This article is protected by copyright. All rights reserved
Accepted Preprint
Shang and Bishop 2000). For this application of EcoAIMTM, the landscape aesthetics model was developed for two LU/LC patches that predominate at APG: forests and wetlands. In addition to their considerable ecological value, wetlands provide culturally important
services. Wetlands with public access are also valued for their educational and recreational potential. Smardon (1983) found that study participants prefer wetlands with more open waters
compared to highly vegetated sites. Forests, in general, are considered to be good representations of natural areas. Nassauer
(1995) found that survey respondents indicated that areas that are too formal, open, bare, flat, or monotonous and lack trees are unattractive. The forest condition is also important, with people preferring a “natural” stand compared to areas that have been recently harvested, tree plantations where there is a poorly developed under- and mid-story, or trees planted in rows. The “mystery”
associated with forest openings is considered a positive attribute of forest landscapes (Herzog and Miller 1998; Herzog and Bryce 2007). The variables measured for each forest and wetland patch include the following. Landform contrast: Differences in vertical stratification of landforms are cited by several
investigators as important contributors to the aesthetic value of landscapes (e.g., Sakieh et al 2016; Junge et al. 2014). In particular, Smardon (1983) found that steep slopes on the sides of wetlands provide landform contrast which makes a wetland more aesthetically pleasing than
wetlands with gradual slopes and minimal contrast with surrounding areas. Similarly, people prefer perceiving a distinct difference between adjacent patches, such as forested and open areas (Smardon and Fabos 1983). The landform contrast variable is the absolute value of the difference between the elevations (in meters) of the wetland/forest patch and the adjacent landform. The landform contrast variable is calculated in two steps. First, relative relief was calculated as the
This article is protected by copyright. All rights reserved
Accepted Preprint
absolute difference between the average elevation of a wetland and its adjacent landform (100-m buffer ring). Second, the land contrast variable is calculated by dividing the relative relief (meter) by the average width (meter) of the wetland.
Edge complexity: Forests with maximized edge-to-area ratios are perceived as more
aesthetically pleasing than those with lower edge-to-area ratios. The shape of a forest patch is assumed to indicate the intensity of human impact. For example, people prefer forest patch edges to be undulating, or sinuous, making them less noticeable and more “natural” compared to straight edges (that suggest maintenance by humans) (Gobster 1999; Frank et al. 2013; Smardon
1983). The calculation to measure the sinuosity of the wetland/forest edge is:
where P = perimeter and A = area (Smardon and Fabos 1983).
Surrounding land-cover contrast: People prefer variation in the scenery as they travel
along a corridor (Ribe 1989). Wetlands with surrounding contrasting upland landforms are preferred because the uplands define a sharp visual image of the wetland and provide a feeling of enclosed space. Through preference testing, wetlands that are adjacent to open water, forest, and agricultural lands were deemed most visually appealing (Smardon 1983). The surrounding landcover contrast variable is measured as the number of different land cover types found within a 100-m buffer delineated around each forest or wetland. The 100-m buffer corresponds to an assumed “experiential” aesthetics envelope of 100-m from the point of observation. Although no studies cite the 100-m point, it was assumed this is an appropriate distance at which an observer could see in the distance and is consistent with the selection of constraining distances for the two
This article is protected by copyright. All rights reserved
Accepted Preprint
models The distance constraint can be modified by the use based on site-specific attributes such as significant elevation changes or viewpoint of the observer from a promontory, where a larger buffer are may be more appropriate. Surrounding land-cover diversity: The preference for seeing different types of land cover
varies by individuals and is typically assessed during the stakeholder implementation phase of a study of this type. Researchers have studied the importance of psychological factors and stakeholder attitudes that influence preferences for landscape types (Ribe 1989; Ribe 2009; Gobster and Chenoweth 1989; Kearney and Bradley 2011). Thus, the model allows weights to be assigned to land use or land cover types by the user to represent preferences. Incorporation of
preferences is an important functionality to account for the widely varied landscapes occupied by DOD installations nationwide (e.g., desert, high mountain, shoreline/coastal, lowland mixed
forest, western fir forests, etc.). For example, whereas open water may rank highest in a coast/shoreline site like APG, it would be irrelevant for desert sites like White Sands, New Mexico, or the Yakima Training Center in Washington State. Surrounding land cover diversity is measured as the weighted sum of the total different land cover types within a 100-m buffer delineated around each forest or wetland. Patch Size: In contrast to vista aesthetics, where diversity of patch types is important,
for landscape aesthetics, people generally prefer experiencing large patches of land rather than
small patches. Studies have determined that the largest wetlands are the most aesthetically pleasing, possibly because wetland size is related to recreational value and increased biodiversity (Smardon 1983). Similarly, large forest tracts are also more desirable than small patches. Size is a measure of the total area of the wetland or forest patch.
This article is protected by copyright. All rights reserved
Accepted Preprint
Vegetative Interspersion: The surface pattern or “texture” of a wetland is an important
aspect of aesthetics. Surveys show that people prefer completely interspersed, grouped vegetation patterns on the water surface more than completely vegetated waters. Vegetative interspersion is a count of the total number of pixels along the perimeters of vegetation groups within each wetland. Similarly, people prefer forested areas with “natural-looking” groupings of trees over evenly spaced plantation-like forests (Ribe 1989). If detailed forest survey information is available, vegetative interspersion is measured as the number of different tree species within each forest patch; otherwise, this variable is measured the same way as wetlands. Forest density: Stamps (2010) suggested that the concept of ‘mystery’ or the distance
which people can see through the vegetation is extremely important to aesthetic preferences. Study participants who were shown pictures had greater appreciation of forests with greater canopy cover and areas with smaller openings (Gobster 1999). Forest density is a measure of the percentage of canopy cover in each forest patch. Water-body size: Views than contain water bodies provide a sense of being most natural,
attract and hold a viewer’s attention, and create a calming effect (Nasar and Li 2004). For wetlands, as well as with other water bodies, the larger the water area, the more aesthetically pleasing the water body is assumed to be, with all other attributes being equal. Water-body size is a measure of the area of water inside a wetland patch boundary. Associated water-body diversity: Wetlands that are adjacent to other water bodies, (i.e.,
rivers, small lakes, ponds, and saltwater bays and inlets) are optimum environments from a visual perspective (Smardon 1983). Associated water body diversity is a count of the total number of open water bodies (excluding other wetlands) within a 100-m buffer delineated around the wetland. This article is protected by copyright. All rights reserved
Accepted Preprint
Data Processing Each of the landscape aesthetics variables (described above) is measured for each LU/LC
patch, quantified, and plotted on a distribution. The Jenks natural breaks classification method within the ArcGIS® software is used to categorize (data bin) the values within a distribution. This is a data clustering, optimization method commonly used in ArcGIS which runs iterations of the data (patch variables in this study) such that the average variance is minimized for patches within a group and maximized between groups. By grouping patches with this method, ten
distinct categories of ES values can be created and displayed on maps. The landscape variable values for each patch are averaged and processed with a second
run with Jenks classification. Each habitat patch is then classified in one of ten categories and
color-coded for mapping purposes (red = 1 [worst] to green = 10 [best]). As with the vista aesthetics model, in the model’s default mode, the attributes contribute equally to the overall landscape aesthetics score; however, the user can assign preference weights (e.g., 0-5) to selected attributes to reflect stakeholder preference or site-specific landscape features.
Statistical Analysis
To assess whether changes in either vista or landscape aesthetics values were significant
between a baseline condition and a hypothetical development condition (the Development scenario), statistical methods were applied to measure the differences in aesthetics scores. To
avoid potentially violating assumptions of normality of the data distributions, a two-tailed nonparametric Wilcoxon signed rank test on paired observations was used to determine whether the difference between the Baseline case and the Development scenario is significantly different This article is protected by copyright. All rights reserved
Accepted Preprint
from zero at p Timmins, K.B. Jones and B.L. Jackson. 1995. A factor analysis of landscape pattern and structure metrics. Landsc. Ecol., 10(1): 23-39. Rogan, J., J. Miller, D. Stow, J. Franklin, L. Levien and C. Fischer. 2003. Land-cover change monitoring with classification trees using Landsat TM and ancillary data. Photogram. Eng. Remote Sensing, 69:793–804. Rolston, H. III. 1995. Does aesthetic appreciation of landscapes need to be science-based? British Journal of Aesthetics, 35(4): 374-386. Rudell, E.J., J.H. Gramann, V. A. Rudis, and J.M. Westphal. 1989. The psychological utility of visual penetration in near-view forest scenic-beauty models. Environment and Behavior, 21(4): 393-412. Sakieh, Y., A. Salmanmahiny, S. Hamed Mirkarimi, and S. Saeidi. 2016. Measuring the relationships between landscape aesthetics suitability and spatial patterns of urbanized lands: an informed modelling framework for developing urban growth scenarios. Geocarto Intl. Published online 2 May 2016. http://www.tandfonline.com/doi/abs/10.1080/10106049.2016.1178817 Schädler, S., M.Finkel, A. Bleicher, M. Morio and M. Gross. 2013. Spatially explicit computation of sustainability indicator values for the automated assessment of land-use options. Landsc. Urban Plann., 111: 34-45. Schirpke, U., E. Tasser, and U. Tappeiner. 2013. Predicting scenic beauty of mountain regions. Landsc. Urban Plann., 111: 1-12. Scepan, J. 1999. Thematic validation of high-resolution global land-cover data sets. Photogram. Eng. Remote Sensing, 65:1051–1060.
This article is protected by copyright. All rights reserved
Accepted Preprint
Shang, H. and I.D. Bishop. 2000. Visual thresholds for detection, recognition and visual impact in landscape settings. Journal of Environmental Psychology, 20: 125-140. Smardon, R. C. 1983. State of the art in assessing visual-cultural values.
Smardon, R.C. and J.G. Fabos. 1983. Chapter 9. A model for assessing visual-cultural values of wetlands: A Massachusetts case study. Stamps, A.E. 2010. Effects of permeability on perceived enclosure and spaciousness. Environment and Behavior, 42(6): 864-886. Steinitz, C. 2001. Visual evaluation models: some complicating questions regarding memorable scenes. Landsc. Urban Plann., 54: 283-287. Tyrvainen, L., S. Pauleit, K. Seeland and S. de Vries. 2005. Chapter 4. Benefits and uses of urban forests and trees. Urban Forests and Trees, 81-114. Ulrich, R.S. 1979. Visual landscapes and psychological well-being. Landscape Research, 4(1): 17-23. Weng, Q. 2002. Land use change analysis in the Zhujiang delta of China using satellite remote sensing, GIS and stochastic modelling. J. Environ. Manage., 64:273–284. Wulder, M.A., S.E. Franklin, J.C. White, J. Linke, and S. Magnussen. 2006. An accuracy assessment framework for large-area land cover classification products derived from mediumresolution satellite data. Intl. J. Remote Sensing, 27:663–683.
This article is protected by copyright. All rights reserved
Accepted Preprint
LIST OF FIGURES
Figure 1. Location of U.S. Army’s Aberdeen Proving Ground
Figure 2. a) Site-wide wetlands landscape aesthetics under baseline conditions; and b) Site-wide forest landscape aesthetics under baseline conditions.
Figure 3. Vista aesthetics model screen shots under Baseline and the Development scenario.
Figure 4. Landscape aesthetics model output for wetlands under Baseline and the Development scenario.
Figure 5. Landscape aesthetics model output for forests under Baseline and the Development scenario.
This article is protected by copyright. All rights reserved
Accepted Preprint
LIST OF TABLES Table 1. Statistical analysis of model results: Landscape Aesthetics--Forest Table 2. Statistical analysis of model results: Landscape aesthetics—Wetland
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 1
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 2(a)
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 2(b)
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 3
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 4
This article is protected by copyright. All rights reserved
Accepted Preprint
Figure 5
This article is protected by copyright. All rights reserved
Accepted Preprint
Table 1. Statistical analysis of model results: Landscape Aesthetics--Forest Surrounding Surrounding Landform Edge Land Cover Land Cover Contrast Complexity Contrast Diversity
Patch Size
Vegetation Interspersion
Forest Density
V.stat 56 5 7.5 26 10 16 p.value 0.484642 0.422678 0.589774 0.287636 0.100348 0.821098 sig.diff no no no no no no no
This article is protected by copyright. All rights reserved
5 1
overall 183 0.018729 yes
Accepted Preprint
Table 2. Statistical analysis of model results: Landscape Aesthetics--Wetland
Landform Contrast
Edge Complexity
Waterbody Size
Water Body Diversity
Surrounding Land Cover Contrast
Surrounding Land Cover Diversity
Wetland Size
VegetationWater Interspersion
overall
V.stat 106 5 9 3.5 7 7 6 17 193 p.value 0.985101 0.422678 0.78353 0.343028 0.577469 0.577469 0.181449 0.000649 0.000502 sig.diff no no no no no no no yes yes
This article is protected by copyright. All rights reserved
