Page 38 - A bird’s-eye view of recreation - Rogier Pouwels
P. 38

 A bird's-eye view of recreation
If our rules of thumb are to be useful, it is essential that they are applicable in a range of nature areas. Such areas should therefore have similar features to our study area. The most important features of our study area are its size (a few thousand hectares), the large path network with multiple car parks, the fact that it is a cultural landscape, common in western Europe, and the presence of just a few specific attractions. The steep distance decline curve has also been found in several other studies, suggesting that it is a generic description of the correlation between visitor densities and distance to car park (Yang and Diez-Roux 2012, Tratalos et al. 2013, Prins et al. 2014). One way of testing the validity of the algorithm for use in other areas is to compare the average trip length of visitors in the New Forest found in this study with other studies in similar areas. Such a comparison shows that the average trip length is in the same order of magnitude. Meijles et al. (2014) reported 4.8 km in a mixed forest and heathland area in the Netherlands, Taczanowska et al. (2008) reported 5.2 km in an urban forest park in Austria and Zhai et al. (2018) reported 3.4 and 3.8 km in two urban forest parks in China. Shorter lengths were reported by Sharp et al. (2008): 2.2 km for dog walkers and 2.4 km for walkers in the Dorset heaths (UK) and 2.5 km for dog walkers and 2.6 km for walkers in the Thames basin heaths (UK). In small nature areas the results might be less useful as the average trip length and maximum distance visitors penetrate into the area might be lower; Hornigold et al. (2016) uses 400 m as a typical distance covered by visitors entering nature areas in the UK.
2.5.3 Dealing with GPS data
Due to the large numbers of tracks and car parks where visitors have been monitored in the area we consider the dataset to be a good reflection of visitor behaviour and visitor densities in the New Forest. Using GPS devices for monitoring purposes always has limitations due to the accuracy of the locations stored by the GPS device. Especially in woodlands, single data points may lie some distance from the path network (Piedallu and Gégout 2005). Lack of accuracy can lead to errors in the dataset and we found that error handling is a time consuming part of the research (Meijles et al. 2014). Communication errors or breakdowns between the GPS device and satellites, usually for short periods, meant that some parts of the routes taken by visitors were missing. We used the travelling salesman algorithm (Appendix 2) to fill these gaps, but as the algorithm always chooses the shortest distance over the path network, some of the selected paths may not actually have been used. A relatively small part of the routes followed (15%) were constructed by the algorithm and we are confident that most of the paths were selected correctly as the visual check in step four of the data preparation
36






























































































   36   37   38   39   40