Mapping Wilderness by Stephen J. Carver & Steffen Fritz

Author:Stephen J. Carver & Steffen Fritz
Language: eng
Format: epub
Publisher: Springer Netherlands, Dordrecht

Figure 6.7 shows that there are large differences primarily in areas of agriculture where there are higher values of human impact using the crowdsourced approach compared to the human footprint. Since the human footprint is based partly on gridded population density, there may be many areas where density is low in rural areas but Google Earth shows a different picture. The presence of road networks and borders also show up as visible differences between the two images. Moreover, areas of Africa, China and the Middle East all have a higher human impact according to Sanderson et al. (2002) compared to the crowdsourced map. Google images suggest landscapes that are less influenced by humans than that which results from a more top-down methodology. Some of these differences may arise because of the fact that the two approaches may actually be measuring slightly different concepts, e.g. assessing human impact via Google Earth considers the direct impact of manmade features on the Earth’s surface but it does not directly take remoteness from anthropogenic features into account, as with more traditional wilderness mapping.

Each approach to mapping human impact clearly has advantages and disadvantages. In the case of the crowdsourced approach, the data can be collected very easily but this approach relies on only a sample of data, where some areas may need a much denser representation to accurately reflect human impact. Scaling up this approach to produce a truly representative map of human impact may require far more data than have currently been collected. Moreover, the data have been collected at a resolution of 1 square km. As this resolution contains many heterogeneous pixels of mixed land cover types, determination of human impact is complicated. Increasing the resolution of the sample may improve the ease with which human impact can be identified although this will result in a trade-off in terms of how much data can be collected.

There are also issues which arise regarding the quality of crowdsourced data. Although control points of known human impact have been used to determine overall quality, this applies only to a very small number of points. More systematic methods of bias correction and more interactive and ongoing crowd training need to be incorporated into future crowdsourcing campaigns. The temporal element of Google Earth images is another issue, where images are available from different time periods. However, we have started to collect this information so that we can use it to filter out data from old images or provide some indication of certainty based on the currency of the data.

In contrast the map of human footprint was created using global datasets that have comprehensive spatial coverage. Yet Sanderson et al. (2002) clearly acknowledge that there are potential problems with these datasets. Land cover data have been shown to have high spatial uncertainties and accuracies that even today are still only between 65 and 75 % (Fritz et al. 2011; Gong et al. 2013). There are also issues with all of the other datasets used, as pointed out by Sanderson et al.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.