Improve the mapping of terraced landscapes in Europe...
Fundamental to the study of terraces is the detailed mapping of walls, benches, water features and the underlying topography. Recent research has underlined the effectiveness of LiDAR data in the modelling of terraced slopes (Sofia et al. 2014; Tarolli et al. 2014a/b). In particular, Sofia et al. (2014) proposed an approach to automatically label terraced sites and differentiate them from natural landforms and Tarolli et al. (2014a) has shown how, by using topographic indices derived from LiDAR and statistical indicators, it is possible to extract terraced walls. The capability of LiDAR technology to derive a high-resolution (∼1 x 0.1 m) DTM from the bare ground data, by filtering vegetation from raw LiDAR data, underlines the effectiveness of this methodology in mapping abandoned and vegetated terraces. This approach can be used for a first and rapid assessment of the location of terraces, particularly in abandoned systems that might require management and renovation planning. For more on the development of these methods, see our publication in volume 23 of Developments in Earth Surface Processes (Cucchiaro et al. 2020a).
The first step of this project objective is to identify and label terraced sites, even when abandoned and covered by vegetation, and the Sofia et al. (2014) procedure will be applied with this precise aim. Once terraced sites have been labelled and identified, the use of topographic attributes and statistical thresholds as proposed by Tarolli et al. (2014a) will allow the extraction of the terraces. Segmentation based approaches are expected to perform well, especially for the terrace benches and walls, and will result in a delineation of the individual terrace benches which, in turn, will be used to be compared to regional terrace characteristics and terraces state of ‘health’. Information gathered this way about terraces walls can also be applied to stability modelling. Recognising and characterising terraced areas can offer multi-temporal insights into critical issues concerning land degradation and development. The proposed metrics can also identify terraced landscapes for areas where no information is available, such as abandoned terraced sites in remote areas. LiDAR will be supplemented by aerial photogrammetry from an unmanned aerial vehicle (UAV) (eg. quadcopter) and TLS for wall surveys. Terrace depth will be estimated using ground penetrating radar (GPR) which we use extensively at Southampton. The first results of terrace extraction have been published in Remote Sensing (Cucchiaro et al. 2020b).