This paper explores some of the ways in which a cognitive appreciation of landscape may be matched to the GIS context of DEM analysis. Physiographical characteristics of landscape cognition can be modelled using the technology associated with viewshed analysis. Relief, depth of view, horizon characteristics and shape could all be measured using GIS functionality. Uncertainty in surface feature classification may be identified by examining scale dependencies in the landsape and storing the results as feature membership functions. It is suggested that cognitive criteria such as drama, mystery and coherence may have measurable surrogates by using the modelled view as a basis for their definition.

Introduction

There are many aspects of non-deterministic decision making within the sphere of spatial information. This paper will take the example of landscape within GIS, and explore some ways to convey a subjective analysis of the human perception to it. Current GIS functions used to address landscape tend to have their foundation based upon Digital Elevation Models (DEMs) or surrogates thereof (Bishop and Hulse, 1994). They appear to have been dominated by the hydrological modelling community relying on deterministic functions to isolate Boolean concepts such as the drainage network and the catchment area. Generation of the visible area has developed as part of this modelling functionality (Lange, 1994), and in the usual implementation of the operation shows this deterministic heritage (De Floriani and Puppo, 1994).

Visible area analysis is thought to be among the most widely used function in landscape planning with GIS (Davidson et al, 1993). Given the possible discrepancy between the objective result of visible area analysis in GIS and our subjective landscape perception, it is perhaps surprising that such an approach has been so broadly adopted. It is the contention of this paper that future guidance given to landscape planners and architects using GIS in relation to visual experiences needs to be enhanced by the introduction of cognitive elements within the digital environment. Further discussion will suggest reasons for this and how problems associated with the combination of such different data types may be addressed.

Significant research has been undertaken in the area of landscape cognition as it relates to personal experience of landscape (Daniel and Boster, 1976). The literature relating to landscape architecture and environmental psychology abounds with qualitative phrases such as interest, drama, mystery and quality (Kaplan and Kaplan, 1982; Preece, 1991) but it appears that the identification of the spatial extent of many classes of landscape feature (such as valley and hill) remains inconsistent between individual approaches (Shafer et al, 1969, 1973, 1974). No attempt known to the authors has been made to address this problem or model such cognitively defined elements within GIS.

What we can do already

The viewshed operation has received considerable attention with respect to its operational reliability (Fisher, 1991, 1992, 1993) and the optimisation of the line-of-sight operation (De Floriani et al, 1994; Lee, 1991, 1994). Significantly, however, this research has remained in the realm of deterministic modelling. Most assume that a location is either visible or it is not, although Fisher (1994, 1995) has explored a probabilistic approximation and its application and the use of fuzzy set membership functions has been proposed to determine a subjective version of the viewshed.

Researchers in environmental perception have concluded that personal experience of landscape can be classed into four general categories: physiographical characteristics, the presence of specific physical features, cognitive variables and viewer interest (Kliskey and Kearsley, 1993). However, the comparatively basic physical elements of relief, depth of view and the identification of specific features within a given visible area remain difficult to accurately describe in both subjective and digital environments. Cognitive variables such as drama, mystery and coherence (Kaplan and Kaplan, 1982) and their association to elevation, landcover and digital terrain information within the reporting of human perception poses an important challenge to the established manner in which the interrogation of landscape information is implemented.

Most studies attempting to assess landscape perception have employed questionnaire surveys which incorporate the comparison or compilation of significant landscape components (Potter and Wagar, 1971; Leopold, 1974; Countryside Commission, 1986). However, the manner in which the results are collated often make it difficult to relate the data to digitally defined information such as DEM values.

What we need to achieve

This paper seeks to investigate the cognitive and digital interface of landscape value assessment by examining several elements of ‘landscape experience’ so that their inclusion in GIS may be facilitated. Before examining any associations between the physiographical and cognitive elements, it is suggested that a revised approach to the interpretation of the variables themselves needs to be adopted. This discussion is structured around physiographical, specific feature, cognitive, and viewer interest variables (Kliskey and Kearsley, 1993) and will generate suggestions that are specifically designed with GIS application in mind. Current GIS functionality could support the generation of some of these variables (Orland B,1992), for example, components that could be defined as ‘physiographical’, but it is not possible to derive a measure of landscape value from them. It is accepted that the choice of elements discussed is based upon the assumption that their inclusion will contribute to landscape cognition and provide the foundation for a series of indexed GIS variables. As part of their inclusion within GIS many of the implementations discussed assume some form of Boolean modelling which would dictate the point at which individual measures become significant or insignificant in landscape assessment. It is suggested that rather than seeking to ascertain specific points, an exploration of fuzzy membership functions needs to be undertaken which would be based upon context dependent values (Robinson,1988).

There are three principle aims:

a) to extract certain physiographical characteristics from the available digital data that are expected to contribute to landscape evaluation

b) to identify suitable cognitive criteria used in the subjective appreciation of landscape quality

c) to combine the extracted physiographical and cognitive elements in a GIS environment and subsequently, through functionality development, investigate their related roles in landscape value assessment.

The structure of discussion in this paper will be in a question and answer format where challenges will be described with implementable and researchable answers suggested. It is accepted that in some cases, the indices created may prove less suitable in a GIS environment than others. The incorporation of cognitive elements and their interpretation still requires more research.

Physiographical characteristics

The landscape is the product of a multitude of related components which interact at a range of differing scales. For the purpose of this paper, relief, depth-of-view, horizon characteristics and shape are examined and GIS operations are suggested that could identify and analyse them as specific landform morphology components (Gobster and Chenoweth, 1989).

Q. How can we accurately assess the significance of relief in the viewshed?

A. ’Relief’ is an ambiguous concept that is generally considered to be a function of elevation (Mark, 1975; Evans, 1980). Using distance and the viewing elevation data in conjunction with relief angles, a measure of relief may be derived which is sensitive to perspective (Fig.1). An answer may also be derived from the extraction of maximum and minimum viewing angles from a given viewpoint [eg. the line of sight operation r.los in the GRASS GIS]. The viewing angle may in itself be considered a simple measure of relief when examining a series of independent points, but may not be a sufficient indicator as individual elevation values do not allow for concepts such as the relative importance of landscape features. It is therefore suggested that a better indicator of relief is volume. It is considered that viewers of a landscape perceive quality of the view to be related to the amount of land or sky within the given viewshed. The volume of ‘air-space’ within a particular view and the ratio of land to sky can be obtained from digital data and it may be possible to incorporate a weighted function to describe their importance within a given view. For instance, the inclusion of both two and three dimensional geometric characteristics of the view area could provide insight into the relief component by combining volume and perspective viewing within a digital analysis toolpack.

Q. How can we interpret the depth of view or the effects of perspective in a given viewshed?

A. It is simple to extract a summary ‘depth of view’ value from the viewing angle function described above (Fig.2). However, the appropriate inclusion and significance of the incorporation of such a measure within landscape value assessment remains unclear as such analysis could produce several different components. For example, the distance of the furthest point from the viewer may be a high or low value depending on the shape of the viewshed and as a result viewer position has to be considered carefully in the interpretation of the landscape (Unwin,1975). An alternative approach may be to generate an area weighted mean value (from viewer to all points within the viewshed) or a standard deviation component for all such points. The application of a differential weighting of distances could prove effective in resolving problems surrounding the perspective component in such a model.

Q. In what way can the characteristics of the skyline and other intermediate horizons be incorporated within operational GIS functionality? Is it possible to quantify the way in which different types of horizons contribute to the view quality?

A. Although the skyline is obviously an important factor in the appreciation of the landscape, its extraction is not usually available within GIS functionality. If this was obtainable, its length and that of intermediate horizons could be associated in the analysis of the two-dimensional planimetric measures for a given viewshed area. An assessment of the contribution of linear horizon features could then be undertaken including, for instance, their density within the view and the area they screen compared to the viewshed area. Characteristics of each horizon such as their smoothness and the number of times the horizon is broken could also be incorporated which would provide the first steps to producing a measure of horizon dominance and the subsequent description of individual horizon qualities which may affect view quality.

Q. Can the ways in which different view-shapes affect cognitive assessment be parameterised? For example, are different views more pleasing than others as a result of their viewshape characteristics, and if so, can this association be measured when interrogating digital data?

A. Following on from the generation of a skyline and horizon function, a complementary component would enhance the characteristics of the local landscape as viewed from a point whilst reducing the importance of distant landscape components. This function would incorporate consideration of the linear boundary features and horizons of a viewshed and would be designed to emphasise the dominance of a given landscape element in the foreground whilst reducing the impact of an identical feature in the background (Craik,1972). A proportional scale could be applied to morphological features in the context of assessing their contribution to landscape value with local outliers and extreme features being an important consideration in this weighting function.

Presence of specific features

The established conventions applied to the analysis of specific landscape features combine generalised characterisations of a topographical surface with individual landscape elements (Hadrian et al, 1988). Whilst the ‘shapes and forms of the world’s surface’ can be modelled within the GIS environment it is not so simple to define the specific boundaries of ‘mountains and valleys, plains and plateaus’ for digital analysis. In addition, questions remain as to the manner in which scale dependent classes of such relief components may be extrapolated. Specific features may be seen as landsurface elements such as mountains, valleys, plateaus, natural landcover for example rivers, woods, moors and cultural features including settlements, archaeological features and other human impacts.

Q. How do we identify surface shape?

A. The process of systematically identifying surface form from mapped data has a long history dating from at least the mid-nineteenth century (Cayley, 1859). More recently, automated procedures for characterising surface form have concentrated on extraction from raster DEMs (eg, Evans, 1980; Pike, 1988; Meisels et al, 1995). One of the consistent problems observed with such extraction is that results are, in part, dependent on the scale and resolution implied by the raster data model (Evans, 1979; Hodgson, 1995; Polidori, 1995; Wood, 1996a). Any characterisation of surface form should be as independent of the data model as possible.

The approach adopted here to model surface form is to fit a bivariate quadratic surface through local 'kernels' passed over a regular gridded DEM. The derivatives of the modelled surface allow six morphometric feature types to be identified (Evans, 1979). These are, pits (local concavities in all directions); channels (local concavity in one direction, planar in an orthogonal direction); passes (local convexity and concavity in orthogonal directions); ridges (local convexity in one direction, planar in an orthogonal direction); and peaks (local convexities in all directions). Any part of the surfaces not identified in one of the categories above is regarded as planar. An algorithm for deriving such measures is detailed in Wood (1996b).

Q. How do we incorporate the uncertainty of feature definition?

A. Uncertainty in feature definition arises in part because the same location can be considered part of a number of different features simultaneously. While this may be for a variety of contextual reasons, it is asserted here that scale is a primary cause of this uncertainty. Landscape in the foreground of a view will inevitably be viewed at a contrasting scale to that which makes up a distant horizon. Equally an observer may consider a landscape as a whole, or be concerned with only small parts in greater detail. It is additionally possible that the very change of surface from with scale is itself an important part of a landscape view.

A procedural solution to this question is offered by deriving surface features over a variety of scales and recording not a single value, but a feature membership function for any one location. This is illustrated in Fig. 3 which shows the graphical output from an interactive interrogation of a DEM. Here, size of kernel increases along the X-axis by 100m increments. In this example, Mickledore which lies between Scafell and Scafell Pike in the English Lake District), is seen to be part of a channel at the finest scale (< 200m), a pass at intermediate scales (200m - 2km) and a peak feature at the coarsest scales (> 2km).

Q. How can the contribution of individual artificial features present in the view in question be identified and parameterised?

A. Although it is possible to link tabular information to features within a GIS, this functionality is not, in itself, able to assess specific feature relationships. Given that viewer focusing occurs, does the presence of such a feature preclude the importance of other landscape elements to the extent of reducing their contribution to the overall landscape? For example, given the visual dominance of a named Mountain Peak, does the viewer become sufficiently focused on it to the point of ignoring a pipeline running down its side, that in any other view would be considered an eyesore? By identifying specific features and naming them according to their physiographical characteristics, it is suggested that it should be possible to relate cognitive information to such features in an effort to assess the differences in perceived contributions of both micro and macro landscape components within the viewshed. Subsequent identification of the contribution made by the different components of specific features can then be analysed by simple database abstraction in addition to the perspective and scale criteria already defined by the viewshape characteristics.

It may also be possible to class the ‘feature’ as a polygon with an assigned dominance value where the difference in the feature value to that of the surrounding landscape would provide a means of categorising it as integrated, intrusive, dominant etc.

Cognitive Criteria

The landscape architecture literature suggests that particular landscape components can be considered to have an effect on the quality of any view, e.g. water is generally considered a positive attractor (Leopold, 1969), whereas a road is considered to be negative (Potter and Wagar, 1971). In seeking to build upon such conclusions, a questionnaire has been devised whereby the responses obtained will be used to provide a means of assessing how beneficial or negative these affects may be. The structure of the questionnaire is designed to generate responses that will aid in the identification of certain cognitive elements that are considered particularly suitable to association with the digital data. These include drama, mystery and coherence. (Kaplan and Kaplan, 1982). In addition, trade off assessments may be evaluated through the utilisation of multi-criteria-decision-making models (Jankowski,1995) in an attempt to gauge relative impacts of specific features within the viewshed of interest (Higuchi, 1983). This type of feature description can then be related to the textural analysis within the digital environment and could provide a cognitively defined application to classify the landscape using ‘harmony’ or ‘chaos’ values. For example, is the accepted beauty of a scene compromised by the presence of a certain detractor to the extent that the view looses its beauty?

At present, none of these components can be addressed with GIS functionality, but it is suggested that the following questions may be proposed and subsequent answers explored through the combination of cognitive and digital data.

Q. It is anticipated that the presence of certain features make certain landscapes more dramatic to view than others. Is it possible to identify from the cognitive response data combinations of physiographical and geometrical components that may contribute to differing landscape quality and facilitate value analysis within a GIS?

A. It is expected that the viewers perception of the landscape can be related to the plan geometry characteristics derived from digital data. It is proposed that a viewing position from the top of a mountain is dramatic, but a view from the base of the same mountain peak could be similarly described. It may be possible to assess drama within a GIS by categorising the viewshed into proximal, intermediate and distant viewing areas and combining this element with the maximum and minimum viewing angle. For example:

i) In the proximal viewing region (0m - 1km from viewer) drama may be created by the presence of a cliff or precipice where the angle of relief is significantly greater than the viewing angle. This could be seen as particularly dramatic.

ii) In the middle region, (1km - 5km) drama tends to be created by the presence of a peak or significant visible topographic variation to the surrounding area. The viewing angle would be closer to the relief angle and the drama would then be derived from a combination of angle, feature and scale information.

iii) In the distant viewing area, (5km - skyline horizon) drama is created by a large-scale landscape feature such as a volcano or mountain range, and as a result, the impact of the viewing angle may be a lesser consideration. In this case, the skyline shape would be combined with viewangle and relief components.

Further investigation may identify geometrical elements that are of particular importance when relating the concept of drama to those of scale, perspective and more importantly the relief component(Litton, 1974). It is proposed that drama is a function of the corporate effects of physiographical, planimetric and cognitive criteria. The result of such investigation should provide the functionality to combine data from such categories to provide a cumulative descriptive index of landscape drama.

Q. It is suggested that the concept of mystery contributes greatly to landscape assessment but how can this contribution be assessed and then substantiated?

A. Mystery is closely linked to the distance that a viewer can see and the viewshape of the visible area (Gimblett et al, 1985). Typically, it is suggested that a view that has a mysterious component contains areas that the viewer knows must be present but are out of view and yet within relatively easy access. The concept of discovery is linked to that of mystery, and in the context of landscape assessment is usually made more acute by the presence of some form of access to the areas out of sight. For instance, in a valley scene, intrigue may heighten the concept of discovery by a pathway leading along its floor or, in a ‘corridor view’ by the visibility of features beyond a local or intermediate horizon. In both examples, the viewer is drawn into the landscape by the intrigue of what lies ahead or within. By analysis of the horizon characteristics and masking of visible areas, it should be possible to generate a mystery component when combined with landsurface and landcover information. The generation of such an indicator will depend greatly on the ratio of land in and out of view within the viewshed and the areas lying beyond the horizons in question.

Q. Coherence is used to describe the fractal nature of the landcover within the visible area. How is coherence measured, and what effect does it have on the way in which people assess the quality of the landscape?

A. It is the contention of the authors that coherence affects how people feel and thus the manner in which cognitive responses are generated. For example, if a given area has a large number of landscape components that are highly spatially auto-correlated (negatively or positively) the viewer may be less likely to consider it attractive. This is because the landscape will appear visually uniform and possibly ‘boring’, whereas if the land surface elements are more randomly arranged, the viewer is more likely to identify areas of interest and variation (Crofts, 1975). Examples of the former can be found in regions dominated by conifer afforestation practices and large-scale grain crop planting. By composing a digital representation of the landsurface and landcover information, this suggestion can be ratified by combination with cognitive data obtained from respondents in areas of differing vegetation and cultural characteristics (LaGro, 1991).

How do we relate the cognitive and digital elements within a GIS environment?

Discussion has shown how absolute values may be extracted from digital information. It may then be possible to relate these values to cognitive components. To parameterise such relationships the questionnaire used provides the foundation for the interpretation of both physiographical and specific feature components within the data and specifically targets the appreciation of landscape components that can either be currently analysed or may form an element of newly proposed functionality. The cognitive data generated is expected to complement the physiographical characteristics discussed above. The respondents to the questionnaire are requested to interpret the viewed landscape according to its feature composition, attractors, detractors and potential user qualities. The subjectively defined criteria obtained are then combined with the digital data in an attempt to extrapolate the operational functions, as illustrated in Table 1.

Cognitive Data	Digital Data	Operational Function
the most satisfying place to view the landscape from	physiographic and relief assessment	satisfaction index
viewing the landscape with land above or below the viewpoint	line of sight and viewshed viewing angle	drama index
interpreting the similarities of the view to one of three photos	viewshape depending on combination of 2D & 3D data	visibility and pictorial control element
comparing the view to a series of digitally produced images	texture and pattern index of landsurface elements	coherence and aesthetic value
relating viewing position to a 2D illustration of the landscape	2D & 3D planimetric comparisons and relief element	perspective and scale measurement
measuring the impact of linear components within the view	linear features such as skyline and intermediate horizons	linear feature measurement
assessing view quality over considering entire viewshed	viewing angle and variable angles of relief within the view	view quality measurement
identification of visual attractors and detractors	specific feature dominance values and harmony component	specific feature contribution
inclusion intrigue and discovery as viewer is drawn into the view	view-shadow and ratio of horizons to visible area	mystery index

Table 1: The combination of cognitive and digital data to create GIS functionality

The relationships between cognitive and digital elements are examined by comparing the data obtained from the questionnaire and the digital information stored in landcover and DEM coverages. For example, it is believed that associations between the viewer position (Litton,1968) and the expected viewer satisfaction may be illustrated, and that the ‘aesthetic experience’ may be determined from a combination of the texture and pattern of the land cover information as and the digital plan form of the viewshed. When incorporating distance and scale variables, it is anticipated that if the viewer can accurately interpret the viewshed characteristics from a viewpoint, the sense of satisfaction and appreciation of the wider landscape will be enhanced. It is also the contention of this paper that there are relationships between the number and shape of the horizons present within a viewscape and the pleasure experienced by the viewer, as well as many other possible combinations of variables that may be interpreted.

A fourth category of landscape value assessment that has not been explored is the interest factor. This criteria is probably the most subjective of the landscape components and hence the hardest to generate absolute results for GIS inclusion. However, in the context of the above approach to landscape evaluation, it is possible to generate ‘interaction measures’ from the questionnaire data by identifying feature classes of interest (Veal, 1974; Daniel and Boster, 1976). For example, a rock climber will be drawn to a view of a cliff from the base whereas a hang-glider will have more interest in the same feature from the top; an archaeologist’s eye will be drawn to specific cultural features irrespective of other features when a hiker might be more interested in the location for its views. Measures of viewer interest can be addressed in the same way as specific features, with a weighting index depending on other elements such as access, facilities, and habitat.

This paper has illustrated some of the possible associations that may be explored between physiographical, feature and cognitive components in the discussion of the GIS environment. It remains to be seen which of the suggested components discussed above lead to enhance currently available GIS functionality. It is the conclusion of this paper that deterministic digital analysis in GIS can become more accurate by adopting an increasingly flexible approach to cognitive criteria in an effort to accommodate the subjective decision making processes that are currently overlooked within concept of Geographic Information systems.

References

Appleton, J. (1975) The experience of landscape. London: J Wiley and Sons
Bishop, I.D., Hulse, D.W., (1994) Prediction of scenic beauty using mapped data and geographic information systems. Landscape and Urban planning 30, 59-70
Cayley, A. (1859). On contour and slope lines. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science. XVIII, 264-268.
Countryside Commission (1986) Wildlife & Countryside Act 1986, Conservation maps of National Parks: CC no.6.
Craik, K.H. (1972) Appraising the objectivity of landscape dimensions, in Krutilla (ed) Natural environments: studies in theoretical and applied analysis, 292-308, Resources for the future inc.
Crofts, RS. (1975) The landscape component approach to landscape evaluation Transactions of Institute of British Geographers 66, 124-129.
Davidson D. A., Watson A. I., Selman P. H. (1993) An evaluation of GIS as an aid to the planning of proposed developments in rural areas. In Mather, P.M. (ed) Geographical Information Handling: Research and Applications, 251-259, London:Wiley
Daniel, T.C., Boster,R.S. (1976) Measuring Landscape Aesthetics: The scenic beauty estimation method. USDA Forest Service Research paper RM-167
De Floriani L, Marzano P, Puppo E. (1994) Line-of-sight communication on terrain models. International Journal of Geographical Information Systems 8, 329-342
Evans, I.S. (1979) An integrated system of terrain analysis and slope mapping. Final report on grant DA-ERO-591-73-G0040, University of Durham, England.
Evans, I.S. (1980) An integrated system of terrain analysis and slope mapping. Zeitschrift fur Geomorphologie, Suppl-Bd 36,.274- 295.
Fisher P F. (1991) First experiments in viewshed uncertainty: The accuracy of the viewshed area. Photogrammetric Engineering and Remote Sensing 57, 1321-1327
Fisher P F. (1993) Algorithm and implementation uncertainty in viewshed analysis. International Journal of Geographical Information Systems 7, 331-374
Fisher P F. (1994) Probable and fuzzy models of the viewshed operation. In Worboys, M. (ed) Innovations in GIS 1, 161-175, London:Taylor & Francis
Fisher P.F. (1995) An exploration of probable viewsheds in landscape planning. Environment and Planning B: Planning and Design 22 (4), 527-546.
Gimblett H.R., Itami R.M., Fitzgibbon J.E. (1985) Mystery in an Information processing model of Landscape preference. Landscape Journal 4 (2), 87-95
Gobster P H., Chenoweth R E. (1989) The dimensions of aesthetic preference: a quantitative analysis. Journal of Environmental Management 29, 47-72
Hadrian D.R., Bishop I.D., and Mitcheltree R. (1988) Automated mapping of visual impacts in utility corridors, Landscape and Urban Planning 16, 261-282
Higuchi T. (1983) The visual and spatial structure of landscapes [eg.p72] (translated by Terry C.S.), Tokoyo: Gihodo Pub.Co.Ltd.
Hodgson, M.E. (1995) What cell size does the computed slope / aspect angle represent ? Photogrammetric Engineering and Remote Sensing, 61(5), 513-517.
Jankowski P. (1995) Integrating GIS and mulitiple criteria decision- making. International journal of GIS 9 (3), 251-273
Kaplan S., and Kaplan R. (1982). Cognition and Environment: Functioning in an Uncertain World, New York: Praeger
Kliskey A.D., and Kearsley G.W. (1993) Mapping multiple perceptions of wilderness in southern New Zealand. Applied Geography 13 203-223
LaGro, J. (1991) Assessing patch shape in Landscape Mosaics. Photogrammetric, Engineering and Remote Sensing 57(3) 285-293
Lange, E. (1994) Integration of computerized visual simulation and visual assessment in environmental planning. Landscape and Urban planning 30 99-112
Lee J. (1991) Analyses of visibility sites on topographic surfaces. International Journal of Geographical Information Systems 5, 413-429
Lee J. (1994) Visibility dominance and topographic features on digital elevation models. Photogrammetric Engineering and Remote Sensing 60, 451-456.
Leopold L,B. (1969) Quantitative comparison of some aesthetic factors among rivers. Geological survey Circ. 630 1-16 USDI G.S.Washington D.C.
Litton R.B.Jr. (1974) Visual vulnerability of forest landscapes Journal of Forestry 72 (7), 392-397
Mark, D.M. (1975) Geomorphometric parameters: a review and classification, Geografiska Annaler 57 A, 165-177.
Meisels, A., Raizman, S. and Karnieli, A. (1995) Skeletonizing a DEM into a drainage network, Computers and Geosciences, 21 (1), 187-196.
Miller D.R., Morrice,J.G., Whitworth,P.L., Aspinall,R.J. (1994) The use of GIS for the analysis of scenery in the Cairngorm Mountains in Heywood and Price (eds) GIS in Mountainous Regions, London: Taylor & Francis
Orland B. (1992) Data Visualization techniques in environmental management: a research, development and application plan. Landscape and Urban Planning 21, 241-244.
Pike, R.J. (1988) The geometric signature: Quantifying landslide terrain types from digital elevation models, Mathematical Geology, 20 (5), 491- 511.
Polidori, L. (1995) Fractal-based eevaluation of relief mapping techniques, in Wilkinson, G, Kanellopoulos, I. and Mégier, J. (eds) Fractals in Geosciences and Remote Sensing, Joint Research Centre, Report EUR 16092 EN.277-297.
Potter D.R., Wagar J.A. (1971) Techniques for inventorying manmade impact in roadway environments USDA FSRP PNW-121 12p
Preece R.A., (1991) Designs upon the Landscape. London: Belhaven Press
Robinson, V.B. (1988) Some implications of fuzzy set theory applied to geographic databases. Comput., Environ. and Urban systems 12
Schafer E.L.Jnr, Hamilton J.F. Jnr, Scmidt E.A. (1969) Natural Landscape preferences: a predictive model. Journal of Leisure Res. 1(1) 1-19
Unwin, K. (1975)The relationship of observer and landscape in landscape evaluation Transactions of Institute of British Geographers 66 130- 134.
Veal, A.J. (1974) Environmental Perception and Recreation.A review and annotated bibliography
Wood, J.D. (1996a) The geomorphological characterisation of Digital Elevation Models, PhD Thesis, University of Leicester, UK.
Wood, J.D. (1996b) Scale-based characterisation of Digital Elevation Models, in Parker, D. (ed) Innovations in GIS 3, Ch.14, London: Taylor and Francis