Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
5741880 | Ecological Informatics | 2017 | 13 Pages |
Highlightâ¢The absence of temporal spatial data hinders ecological modelling in West Africa.â¢Neutral landscape models offer new dimensions in data scarce regions.â¢Landscape generation through SG4GISCAME presents a flexible alternative tool.â¢We present a variant of the Turing test approach to validate model outcomes.
Despite its popular adoption and use, neutral landscape models have been unexplored in data scarce areas of the Sudanian Savanna region where its application could serve as inputs for spatial ecosystem service assessment. Thus, the need for an easy to use tool to produce landscape patterns similar to real landscapes in this area is imminent. In this article, we aimed at introducing SG4GISCAME as a tool to meet this purpose by exploring it capabilities to generate landscapes similar to real agricultural landscapes of the Vea catchment area in Ghana in three steps. We used Voronoi tessellation polygons to develop the image patterns. The resulting artificial patterns were subsequently evaluated through a visual and landscape structural metric comparison between the simulated and real landscapes. Finally, we used a modified Turing Test to test the credibility of SG4GISCAME model output through expert pattern identification cues. The results show that SG4GISCAME can successfully generate agricultural landscape mosaics similar to real landscape under different parameters and user specifications. We attribute this to the tools' intuitive and interactive user interface. Statistical test outcomes of the modified Turing Test suggested that geographic information systems and remote sensing map experts found marked pattern similarities between real and synthesis maps, resulting in challenges in identifying real maps from synthetic ones. Our approach could be replicated in other landscapes of West Africa to provide a substitute for unavailable or expensive spatial data and to test the hypothetical relationship between patchy landscape structure and ecosystem service provision through modelling.