Developing a spatial approach for identifying new recreational opportunities in natural environments Arrowsmith C. School of Mathematical and Geospatial Sciences RMIT University Melbourne, Victoria 3000 Phone: 03 9925 3277 Fax: 03 9663 2517 Email: colin.arrowsmith@rmit.edu.au  Chhetri P. CR-SURF, School of Geography, Planning and Architecture University of Queensland Brisbane, Queensland. Phone: 07 3365 3565, Fax: 07 3365 3561 Email: p.chhetri@uq.edu.au ABSTRACT This paper discusses a GIS-based technique to measure recreational potential for natural tourist destinations. The technique was developed for a study area in western Victoria, Australia, known as the Grampians National Park (GNP), a popular natural tourist destination. In this paper, we have used regression modelling to develop a set of predictors of scenic attractiveness derived from biophysical data collected via questionnaire administered to a group of university students. The derived output was then overlaid with ‘recreational opportunity potential’ generated for the region. In the final stage, a spatial model of ‘recreational potential’ was developed from input maps over the entire geographic area. The study found that the areas with high recreational potential are concentrated around more popular walking tracks in the Wonderland Ranges near Halls Gap in the northeast sector. However, other underdeveloped areas of high recreational potential are suggested as an alternative strategy to ease potential pressure developing around heavily used walking tracks. 1. Introduction Recreation is a social activity that is highly personal. Tourists usually visit particular locations because of some attraction or series of attractions. These attractions are either embedded in the geographic characteristics of localities in terms of their scenic qualities or else are socially constructed as discrete attractions. Australian examples of discrete attractions are Uluru (formerly known as Ayers Rock) and The Twelve Apostles coastal rock stacks in western Victoria. However, levels of attractiveness for these tourist attractions may vary according to an individual’s gender, perception, cultural and ethnic background, socio-economic status, educational level, family situation, health, disability and or age. Therefore, the scope of this paper prevents an in-depth assessment of the myriad of cognitive processes that lead to different appreciation of tourism attractions. The model we discuss in this paper can be used to predict areas with good recreational potential that are unexplored and/or under-utilised. The model is designed to estimate the potential for hiking and therefore ignores the potential for other forms of recreation such as canoeing or rock climbing. ‘Recreation potential’ is the inherent capacity of a location to support recreational activities. Two measures were developed to estimate recreational potential. These include ‘contained scenic attractiveness’ and ‘recreational opportunity’. The ‘contained scenic attractiveness’ is defined as the scenic quality of a geographic space that can be seen by an observer in his or her immediate surrounding. In other words, the observer is within the confines of the biophysical components for that particular geographic space. Examples of these components include trees, understorey, water bodies, slope and relief. ‘Recreational opportunity’, on the other hand, is defined as the degree to which a recreational activity can be undertaken due to certain favourable physical or social conditions. Such favourable conditions often represent the presence of attractions. The potential of an area is measured as the number of opportunities available for contact or interaction from a given point or location. The recreational potential of an area, therefore is the sum total of values that it holds for its contained scenic attractiveness and the number of recreational opportunities accessible within its geographic neighbourhood. 2. The Study Area The Grampians National Park (GNP) represents a system of rugged but attractive series of parallel sandstone ridges with escarpments. The area is geographically situated in western Victoria 260 km to the west of Melbourne and 460 km from Adelaide in South Australia. A map for the Grampians National Park in Australia is shown in figure 1. The GNP offers a series of approximately 80 walking tracks ranging both in length and levels of difficulty (Parks Victoria 2003). These walks traverse through diverse landscapes, characterised by unique vegetation communities at lower reaches followed by a sequence of spectacular scenic views and rocky outcrops at higher elevations.  Figure 1 Location of the Grampians National Park in Australia 3. Data Collection Twenty-five university students hiking along the Pinnacle walking track in the GNP recorded field measurements of attractiveness. Observational data were collected via questionnaires that were completed at a series of points along the track. Biophysical details of the location as well as their scenic attractiveness were recorded on a five point Likert scale. Hiking subjects were divided into two groups, one hiking in one direction on the first day, the second group hiking in reverse direction on the second day to see whether preceding events, or time of day, impacted upon responses given. Details of the surrounding scenery that induced that feeling of attractiveness were also recorded along a series of observation points. This allowed the observer to record biophysical properties of the perceived scenes. Observers were also requested to take photographs of the scenes against which responses about biophysical information were recorded. The survey has amounted to a total of 252 observations for a total number of 13 variables. These variables include: elevation, relative relief, vegetation variety, proximity to water, slope diversity, tree canopy, enclosure, steepness of slope, tree trunk size, tree density, amount of understorey, visual alteration and visual penetration. 4. Digital data Spatial data were obtained from a number of sources including the Victorian State Government Department of Natural Resources and Environment (now known as the Department of Sustainability and Environment) and Parks Victoria. Most spatial data used in the modelling were recorded in ratio scales with the exception of vegetation. Vegetation could only be ascertained as nominal data classified into so called ecological vegetation classes (EVC). Triangulated Irregular Networks (TIN) were created from two-dimensional contour lines obtained from the 1: 25,000 digital topographic maps. A Digital Elevation Model (DEM) of 100 X 100 metre cells was generated from the TIN using ArcView 3.2a. Other required information such as slope (maximum slope across a TIN facet), aspect (the bearing of this maximum slope) and elevation were derived from the DEM. 5. Estimating Recreational Potential for the Grampians National Park Recreational potential for the study area was calculated using measures of both scenic and recreational opportunity potential. ‘Scenic attractiveness potential’ was generated as a continuous surface where each cell across the study area holds a modelled value of scenic quality. Recreational opportunities, on the other hand, were stored as a series of discrete points and then were converted into a continuous grid surface. The final recreational potential for the study area was estimated from outputs derived for both the scenic attractiveness potential and the recreational opportunity potential grids. The following five-step method describes the procedure adopted for estimating recreational potential. 5.1 Step 1: Measuring the contained scenic attractiveness potential Data of scenic attractiveness evaluations collected as a part of the survey were used to develop the model of contained scenic attractiveness potential. A step-wise multiple regression analysis was applied to build the model. The scenic attractiveness score was entered as a dependent variable, while the recorded biophysical characteristics of landscape, for example, slope diversity, vegetation variety, and amount of understorey, were used as independent variables. Several assumption tests, including missing values, outliers and multicollinearity were undertaken before conducting further analyses. Based on calculated pair-wise correlations, the problems associated with singularity and multi-collinearity could be ignored. Five predictors of scenic attractiveness (table 1) were identified in the model, which explain sixty percent of the variance in the dataset. These model predictors include elevation, relative relief, vegetation variety, proximity of water and slope diversity as perceived by the surveyed participants. They also show high t statistics and are significant at 0.05. The coefficients of the independent variables are listed in table 1 in the column labelled B. A regression equation has been developed using these coefficients. Whilst the variables of enclosure, visual alteration and trunk size have negative coefficients indicating an increase in their values causes a decrease in the scenic value of the view, they were not included in the final equation. These variables explained less than the remaining 30% of variance in the data. MODEL SUMMARY 
5.2 Step 2: Generation of geographic overlays Data used in the modelling to date, were subjective evaluations collected by students using questionnaires. Such data hold no explicit geographic properties. In order to build a model of scenic potential across a larger geographic space, objective datasets measuring characteristics of the environment itself were needed. Objective datasets were quantified or approximated from digital data. Some indicators were directly derived from the spatial and biophysical characteristics - such as distance, accessibility, and proportion of open space to total area - while others are more complicated (for example, relative relief). Objective datasets for five parameters identified in the linear model were generated so that the geographic variations in the scenic attractiveness could be mapped. These objective datasets were derived as follows: Elevation: Elevation is the terrain height above mean sea level. It was derived from two-dimensional contour lines obtained from the 1:25 000 digital topographic data. The mean height calculated for the GNP is 293.83 metres with a standard deviation of 130.74 metres. The maximum value of height in the park is Mt William at an elevation of 1160 metres. Slope diversity: Slope steepness can be measured as a degree of angle in the digital elevation model. Slope steepness is derived for each cell in the grid from the digital elevation model. The values ranged from 3.53o on gentle slopes to 50.22o in near cliff-like situations, with a standard deviation of 6.32 o. From the slope steepness grid, slope diversity was calculated for every cell using a ‘neighbourhood function’. Neighbourhood functions are spatial functions that alter the value of individual cells based on the values of adjacent or nearby cells (Heywood et al. 1998). The neighbourhood function used for determining slope diversity assigned a new value to the source cell calculated from the standard deviation of height values located in the ‘middle ground zone’ between the inner and outer limits (200 to 500 metres) as defined by the ‘kernel’. Relative relief: Relative relief is defined in this study as a variation of vertical elevation measured in range between the maximum and minimum height in the neighbouring areas. The function involves ascribing a value to a location according to the difference in height of the surrounding area in the output layer. Water proximity: A ‘distance’ function was used to calculate the distances of every cell in the study area to its nearest water feature. The derived distance values were reclassified into seven categories so that visual and non-visual effects on hikers could be incorporated. These categories are less than 25 metres, 25-50 metres, 50-75 metres, 75-100 metres, 100-125 metres, 125-150 metres and more than 150 metres. Vegetation variety: Ecological vegetation class (EVC) data acquired at 1: 25,000 were used to measure vegetative variety. There are 253 EVC classes in the Grampians area. The calculated vegetative variety refers to the number of vegetation communities within the surrounding area of 100 metres. The area was given a score from 1, which represents the lowest number of vegetation communities, through to a value of 7, which represents the most varied. In order to generate the model of scenic attractiveness potential, derived maps of elevation, slope diversity, relative relief, proximity to water and vegetation variety were used as input layers in the model. The data for these layers were converted to standard scores by subtracting the mean and dividing by the standard deviation for each variable (Hair et al. 1995). This measure converts each raw data score into a standardised value with a mean of 0 and a standard deviation of 1. Returning to the linear model developed earlier (table 1), the scenic attractiveness potential was calculated by multiplying each of the cell values of standardised input layers by their respective coefficients and then added into a single layer called ‘Scenic Attractiveness Potential’ using the following linear equation. Scenic attractiveness = .782 + .316 (elevation) + .235 (relative relief) + .148 (vegetation variety) + 0.09 (water proximity) + .133 (slope diversity) The scenic potential model as shown in figure 2 demonstrates the distribution of walking tracks and levels of scenic potential. Most tracks traverse through highly potential scenic areas that show the appropriateness of the existing walking tracks. However other areas of high potential can also be seen across the map. This map enabled the spatial distribution of attractions to be pinpointed and to be contained within what could be considered a reasonably concentrated location (Arrowsmith 2003). Tourists are unlikely to consider individual attractions in their own right, but rather a collection of attractions. For example, in an analysis of waterfalls as a tourism resource, Hudson (1998: 959) notes that visitors come mainly for other reasons than visiting the attraction of a waterfall per se. The output map enabled areas in the vicinity of Halls Gap considered widely to be highly attractive to be identified, and comprise those walks that are currently heavily used.  Figure 2: Map showing scenic potential along existing walking tracks in the Wonderland Ranges 5.3 Step 3: Developing a spatial database of recreational opportunities Recreational opportunities in the region were stored, attributed and classified whereby the quantity, quality and extent of the resource base were given importance in the construction of GIS database (Hall and Page 2002). A total of 190 recreational opportunities were identified, attributed and stored as point features in the GIS database. The point feature is a simple method of constructing a geographic database, as each point only comprises a set of coordinates and some pertinent attribute values (Bailey and Gatrell 1995). For this reason, all attractions were converted to point features. For walking tracks, access points along roads or car parks were identified as points, based on the assumption that visitors could have access to walking tracks only at an access point. Point features were classified as nature-based, recreation-based, cultural and/or historic-based and infrastructure-based opportunities. Nature-based opportunities included scenic lookouts, waterfalls, walking tracks and unique geomorphic features. Recreational-based opportunities comprised rock climbing areas, boat launch sites, picnic grounds and barbeque sites. Significant buildings, monuments, ruins, aboriginal artworks and cave paintings were considered as recreational opportunities for their cultural and historic significance. Finally infrastructure opportunities included accommodation, information centres, horse hire venues, golf courses, caravan parks, and boating facilities. However, these defined categories of recreational opportunity were not necessarily mutually exclusive. For example, a number of more significant lookouts and walking tracks co-existed with unique geomorphic features. These features were incorporated as separate entities in the database. The spatial distribution of recreational opportunities shows that there were fewer recreational opportunities identified in the Southern Grampians. Furthermore, opportunities in the Southern Grampians were generally related to historic and/or cultural interest. 5.4 Step 4: Measuring the potential of recreational opportunity Recreation opportunity potential of a location has been estimated using a ‘neighbourhood operation’. One of the advantages of this operation is that it combines both distance and density. The operation converts discrete and point-based feature data into a continuous surface with a value assigned for each cell. Both the number and size of the opportunities accessible within a commutable distance can be measured in the operation. A suitable range can be defined as the limit beyond which people are less likely to travel to visit a particular location. We have ascertained this limit via a survey of 120 tourists travelling throughout the GNP, by asking them to select an option (out of a total of five choices) that represents their willingness to visit a particular location. These choices include: less than 100 metres; more than 100 metres but less than 200 metres; more than 200 metres but less than 500 metres; more than 500 but less than 1 km and greater than 1 km. The majority of respondents had selected the option of ‘more than 200 metres but less than 500 metres’. Therefore, the mid value of 350 metres has been selected as a desirable limit indicating the willingness of tourists to visit to an opportunity in the Grampians National Park. The selected radius was used as ‘kernel’ that moved over the entire region. Weights assigned to the opportunities falling within the defined radius were counted for each grid cell. Weights used in this study were also determined by asking people to rank attractions according to their importance (for detail see Arrowsmith 2003). Using the “FOCALSum” function in the software, the sum total of the importance scores of recreation opportunities present in the scanning neighbourhood of 350 metres area was calculated. These values were then assigned to each focal cell, situated at the centre of the kernel. The procedure was then replicated for each cell across the entire grid. The following equation was used to generate Recreational Opportunity Potential (ROP): ROP = FOCALSum (Opportunity Scores, KERNEL, 350m) The output map (figure 3) shows areas of high recreational opportunities that contain cells where the number of opportunities falling within a 350 metres radius is high. These areas reveal high concentrations, or clustering, of recreational opportunities. Figure 4 also shows that areas of high concentration of recreational opportunities are around the Wonderland Ranges in the northeast section of GNP. The walking tracks in the Wonderland are situated around Halls Gap, a township that provides most of the tourism infrastructure support for the GNP. The areas around the ‘Pinnacle’ and the ‘Grand Canyon’ in the north, are found to have high recreational potential and consist of a variety of recreational opportunities.  Figure 3: Recreational opportunity potential around the Wonderland Ranges  Figure 4: Map showing areas of high potential around existing walks 5.5 Step 5: Modelling recreational potential for the GNP The scenic and recreational opportunity potential maps were combined to generate the final recreational potential model. These two grids were summed to create the new layer of recreation potential. The output enables areas with high recreational potential to be determined. These areas can then be examined to see whether or not they are currently under-visited. For example, figure 4 shows the areas around the existing walking tracks in the Wonderland Ranges that have high potential. This may indicate to managers the need for proactive control of visitor flows. 6. Conclusions The estimated recreational potential has allowed the development of a quantitative and predictive spatial database of recreation resources. Scenic attractiveness and recreational opportunity were investigated to develop the model of recreational potential for the Grampians National Park (GNP) in western Victoria, Australia. The recreational potential model when mapped shows variations in the distribution of recreational resources. Results show that recreational resources are not equitably distributed and are found to cluster in areas around Halls Gap in the Wonderland Ranges to the north of the GNP. The map also shows other areas that exhibit high recreational potential that have not yet been harnessed for recreational purposes. It has been suggested that promotion of less popular areas by developing strategies based on market segmentation can disperse visitors to other areas of under-utilised recreational opportunities. Tourism destinations need to be tailored to suit the requirements and preferences of tourists. Further studies to incorporate collective perceptions and preferences of other market segments (for example, particular tourist groups) need to be addressed. The study has developed a spatial model that can be used to disperse the economic growth generated by tourism over other areas within the GNP. This study recommends that the areas of high recreational potential should be integrated into a single ‘tourist regional system’ (both Northern and Southern Grampians) so that visitor experiences can be enhanced and biophysical impacts can be reduced. REFERENCES
|