Elsevier

Geomorphology

Volume 364, 1 September 2020, 107169
Geomorphology

Computer-assisted terrain sketch mapping that considers the geomorphological features in a loess landform

https://doi.org/10.1016/j.geomorph.2020.107169Get rights and content

Abstract

In geography, a terrain sketch map is necessary to understand the features and internal structures of a landscape due to its ability to depict key information in a geographical scene using as few lines as possible. Previous computer-drawn sketch maps have focused on the artistic effect rather than on depicting terrain features and landform structures, and thus, they differ considerably from hand-drawn sketch maps by geographers or geologists. This study develops a DEM-based method for terrain sketch mapping that considers the typical feature descriptions of a loess landform and the visual hierarchy expression in line with the law of visual perspective. The method was tested with experiments on two landforms of Chinese Loess Plateau, based on digital elevation models (DEM) with a horizontal resolution of 5 m. In the developed method, first, typical terrain features, including the visual outline, shoulder line, gully and flow lines, are extracted from the DEM. Second, the map is divided into three visual levels in accordance with the data extension and viewing point. Then, terrain feature lines are assigned to different visual levels. Finally, the visual outlines in the distant view are generalized following the law of visual perspective. Results are assessed through a questionnaire with specialists (experts) and students (non-experts). The sketch map was able to characterise loess landforms, and is somewhat similar to traditional hand-drawn maps. The generalisation method realises the near and distant view characteristics of a sketch map, which are detailed and simplified, respectively. The results of the questionnaire also showed that our method presents terrain morphology and geographical scene more accurately and reliably than a hand-drawn sketch map.

Introduction

Sketch maps partially project cognitive maps and reflect acquired spatial knowledge on to paper or other media outside the human mind so that we are able to infer spatial knowledge from sketch map information (Wang and Worboys, 2017). A terrain sketch map can effectively describe the topographic features and spatial pattern of a landscape, which are sketched with lines by a geographer after recognising, analysing and summarising the features of a terrain scene to present his/her understanding of a landscape (Половинкин, 1949; Chen, 1958; Liu, 1988). Compared with a digital elevation model (DEM) scene map, a terrain sketch map, as a scientific abstraction and expression of geomorphological features from the perspective view, is characterised by conciseness and profoundness to show the intrinsic features of a landscape. Therefore, terrain sketch map is one of the effective methods that geographers, geologists, and geomorphologists often use to conduct research in Geoscience. In addition, the sketch map also contains important content of geographical cognition by using concise lines besides artificial effect, in which the geographical environment can be easily understood. In this research, we try to reproduce the process of drawing terrain sketch map based on computer to represent this kind of map in order to assist geomorphological teaching and research, as well as to enrich the approaches of terrain exhibition. However, this type of terrain sketch map is becoming increasingly rare because of the difficulty in drawing one, which requires professional knowledge of geomorphology as well as drawing or painting skills. With the rapid development of computer visualisation and digital terrain analysis methods, it is possible automatically to generate computer-assisted terrain sketches based on DEM, depending on geographic cognition theory and the sketch description method. As explained below, previous automated sketch mappings have not satisfactorily represented geomorphology, especially the unique features of the Loess Plateau of China, as well as ignoring visual perspective law under larger scene.

Previous research has generated terrain scene maps based on DEM data and has produced various styles of terrain representations, i.e. watercolour, oil painting, pen and pencil sketch maps. These studies have focused on rendering methods and aesthetic quality, not on reproducing geomorphology.

Artistic terrain maps can be produced using different brushes designed with non-photorealistic rendering (NPR) algorithms. Watercolour or oil paintings can be achieved via NPR on polygons for an entire terrain surface. The styles, i.e. watercolour, sketch-like and pop art paintings, based on DEM rendering can be found in the Atlas of Switzerland 2.0 system (Hurni and Räber, 2004). In addition, Bhattacharjee and Narayanan (2008) proposed a painterly rendering technique for terrain that integrates several brushstrokes to generate oil-like paintings (Fig. 1a).

Chinese ink paintings can be produced by rendering the lines extracted from a DEM with NPR techniques. Li (2004) derived streamlines and silhouettes from a DEM and designed Chinese brushstrokes to simulate surface wrinkles and terrain outlines by rendering these lines, thereby creating a style of Chinese painting (Fig. 1b). Way and Shih (2006) developed Li's algorithm into four new brushstrokes, namely, LotusLeafStroke, RainDropStroke, MiDotStroke and BonelessStroke, which produce different Chinese ink styles paintings.

Pen- or pencil-like paintings are similar to terrain sketch maps. Che Mat and Nordin (2004) extracted silhouette edges based on a DEM from contours and designed the rendering algorithm using OpenGL to obtain a sketch-like effect. Visvalingam and Dowson (1998) generated a terrain sketch-like map through using different brush stokes (P-strokes) to render terrain profiles which are extracted from DEM and rendering them as segments by a designed filter (Fig. 1c). Lesage and Visvalingam (2002) generated terrain sketch maps using edge detection operators on a terrain illumination image from DEM in 3D perspective, and noted that the best result was obtained by combining sketches from luminance maps generated using a headlight and a vertical light. Whelan and Visvalingam (2003) regarded that occluding contours are insufficient for sketching the silhouettes of hills comparing to the map drawn by cartographers. Then he proposed an enhanced silhouettes extraction method to complete the missing part of silhouettes ignored by a lower viewpoint (Fig. 1d).

All the aforementioned studies aimed to achieve artistic effect rather than topographic features and geographical recognition descriptions. Silhouettes alone cannot depict the topographic features of a landform, such as a ridge, a valley, a saddle and an outlet of a watershed, in the field of geographic landscape.

A great number of researches have been carried out to generate geomorphological maps from DEMs over the past two decades (Dragut and Blaschke, 2006; Iwahashi and Pike, 2007; Minar and Evans, 2008; Romstad and Etzelmüller, 2009; Jasiewicz and Stepinski, 2013; Garcia and Grohmann, 2019). All these maps focus on topographic segmentation or geomorphologic classification in planimetric view. The results are semantic map which can be understood with geomorphological knowledge. Different from those kinds of maps, terrain sketch maps are painted in side-view which consistent with the habit of human being to explore geographic environment in field investigation. The sketch map can be easily understood through highlighting topographic features and ignoring unimportant background elements. Three studies attempted to generate terrain sketch map for geomorphological representation by considering more topographic features (Mower, 2009; Mower, 2011; Qian, 2015). Mower (2009) extracted silhouettes, creases (important topographic features) and form lines (lines depicting overall surface curvature) based on TIN and rendered them in the style of pen-and-ink landscape illustrations. After that, due to the problem of under-generalized surfaces, Mower (2011) proposed a B-spline based method to render the lines extracted from grid cell DEM. The result can represent the local and global topography forms, similar the hand-drawn shading maps (Fig. 1e). However, these sketch maps are still far from the maps drawn by geographers or geologists, who are interested in the landform type and surface material. In order to present more topographic features, Qian (2015) extracted ridge and valley lines in addition to silhouettes to generate a terrain sketch map in the area of the Loess Plateau in China (Fig. 1f). However, this map is insufficient for describing a loess landform because it ignored the typical gully erosion on slope and resulted in the consequence that the landform in the map looks like a hill rather than a loess ridge-hill. To achieve a terrain sketch map similar to the map drawn by geographers, the challenges are the topographic features expression which can help to reflect the geomorphological features effectively and their generalizations in side-view perspective.

The landforms of the Chinese Loess Plateau have been formed through eolian transport and accumulation of loess deposits over bedrock during the Quaternary (Liu, 1985). Soils in the Loess Plateau are brownish yellow, loose and relatively homogenous in texture. The parent material is eolian silt loam, with over 60% silt in the fine fraction. Loess soil has a strong vertical joint structure. Soil cohesion is dependent on calcium carbonate (Zhu, 1991, Shi and Shao, 2000, Wang et al., 2018). For this reason, it is easily eroded by flowing water to form gullies, leading to subsidence and collapse. The surface of the Loess Plateau is cut fragmentally by gully erosion, resulting in three main landforms: loess gullies (both ephemeral and permanent), loess interfluve, and loess karst (Liu, 1964). Apart from loess karst, the vast majority of regions are composed of loess interfluve and gullies. According to the topographic feature, loess interfluves can be subdivided into loess tableland, loess ridge and loess hill. These three landforms are regarded as typical landforms in the Loess Plateau of China (Zhao et al., 2017; Zhu et al., 2018). The main difference of these three landforms is the morphology of loess interfluves and gullies. Therefore, morphologic features can be used as signatures to present different loess landforms effectively.

The general topographic feature lines (such as silhouettes, ridge lines, valley lines and terrain profiles) in terrain sketch map cannot effectively depict the dense features of gully distribution of loess landform. Therefore, some typical features at detailed scale as well as typical topographic lines must be included in the terrain sketch map of loess landform. The shoulder line, as a critical terrain structural line in Loess Plateau, separates the positive and negative terrains attributed to surface and gully erosion, respectively (Yan et al., 2014; Xiong et al., 2014). The shoulder line is also the partition between different land uses. The area above the shoulder line generally has a gentle slope and is usually cultivated for agriculture, while the area below the shoulder line is characterised by a steep slope with intense soil loss, and is usually left as barren land (Chen and Cai, 2006). Therefore, the shoulder line is an indispensable line in terrain sketch map as the boundary of the morphological type, landscape type as well as land use type.

The general topographic feature lines such as silhouettes, ridge line, valley line and terrain profiles in terrain sketch map cannot effectively depict the dense gully distributed features of loess landform. This broken and complex terrain cannot be described by surface rendering methods with these general lines. Therefore, some typical features at detailed scale as well as typical lines must be included in the terrain sketch map of loess landform.

In summary, the existing method cannot accurately represent the loess landscape at the semi-detailed landscape scale. Terrain sketches in geomorphology are large scenes, not just for one or a few hills. Over larger areas as shown in a side-view perspective, the visual perspective covering longer views must be considered. In previous studies sketch maps were made only for relatively small areas, without considering visual hierarchy. In a larger 3D scenes the objects are detailed or clear in near view but simplified or conciseness in distant view. Thus, visual hierarchy division and line generalisation should be considered when the area for sketch mapping is relatively large.

The current study aims to generate a computer-assisted terrain sketch map that is similar to the hand-drawn maps made by geographers or geomorphologists. The specific objectives are: (1) to extract feature lines that present the typical characteristics of loess landforms, (2) to realise visual hierarchy division that is in line with the law of visual perspective (3) to generalise the visual outlines in distant view to improve aesthetics, (4) to evaluate the representational success of the sketch maps produced from DEM at different horizontal resolutions (grid cell sizes) and with different generalisation thresholds.

Section snippets

Areas

The area covered by this study is the core area of Loess Plateau. Two representative areas of the loess plateau, namely, Yijun and Wuqi, were selected by considering the typicality of their geomorphic types and the availability of data. These areas are typical representations of a loess tableland and a loess ridge. The specific information of the areas is provided in Table 1 and Fig. 2.

Data

The data of two areas are from the National Bureau of Surveying and Mapping of China. They were generated by

Basic concept

In hand-drawn sketch mapping, a geographer determines which objects should be depicted after observing and summarising the geographical environment surroundings and then simplifies these objects with lines (Chen, 1958; Liu, 1988). Therefore, three aspects should be considered when using a computer to draw a terrain sketch map. Firstly, the typical landform features must be clarified and then abstracted into lines. Secondly, visual hierarchy should be considered. We found that two or three level

Produced sketch maps

Four type of maps, namely, the original DEM data map, the visual outline map, the map on which visual hierarchy division and visual outline generalisation were not performed and the final terrain sketch map, were drawn and compared as shown in Fig. 8, Fig. 9.

Figs. 8a and 9a are the DEMs in 3D view. In these maps, features are easily recognised as hills and fluvial landforms rather than loess ridges and tablelands, because the typical features of dense deep-cut gullies cannot be clearly

Conclusions

Previous studies on terrain sketch maps have focused on rendering methods to obtain artistic effects rather than depict landform features and consider visual laws. In this study, the important feature lines added to the terrain sketch map were able to effectively represent the typical landform characteristics in the loess plateau of China. Thus, these computer-generated sketch maps are similar to the maps drawn by geographers or geologists. In addition, the law of perspective was considered, so

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We are grateful for the financial support provided by the National Natural Science Foundation China (grant numbers 41771415, 41930102, 41571383, 41971333) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (number 164320H116).

The authors express their gratitude to the journal editor and reviewers, whose thoughtful suggestions played an important role in improving the quality of this paper. Many thanks are also given to Wen DAI, JinWei LI and Xue YANG for

References (46)

  • R.B. Shruthi et al.

    Object-based gully feature extraction using high spatial resolution imagery

    Geomorphology

    (2011)
  • M. Visvalingam et al.

    Algorithms for sketching surfaces

    Comput. Graph.

    (1998)
  • H. Wang et al.

    Soil erodibility influenced by natural restoration time of abandoned farmland on the Loess Plateau of China

    Geoderma

    (2018)
  • J.L.G. Balboa et al.

    Generalization-oriented road line classification by means of an artificial neural network

    Geoinformatica

    (2008)
  • S. Bhattacharjee et al.

    Real-time painterly rendering of terrains

  • J.E. Bresenham

    Algorithm for computer control of a digital plotter

    IBM Syst. J.

    (1965)
  • S.P. Chen

    Sketch Method of Landscape

    (1958)
  • H. Chen et al.

    Impact of hillslope vegetation restoration on gully erosion induced sediment yield

    Sci. China Ser. D Earth Sci.

    (2006)
  • L. Fei et al.

    A three-dimensional Douglas–Peucker algorithm and its application to automated generalization of DEMs

    Int. J. Geogr. Inf. Sci.

    (2009)
  • H. Hu et al.

    Colony optimization algorithm applied to simplification of closed linear objects. Geoinformatics

  • L. Hurni et al.

    Atlas of Switzerland 2.0

  • T. Isenberg et al.

    Stylizing silhouettes at interactive rates: from silhouette edges to silhouette strokes

    Computer Graphics Forum, Wiley Online Library.

    (2002)
  • Y.L. Li

    An Interactive Navigation System of 3D Chinese Landscape Paintings

    (2004)
  • Cited by (4)

    • Large-scale spatial variability in loess landforms and their evolution, Luohe River Basin, Chinese Loess Plateau

      2022, Geomorphology
      Citation Excerpt :

      This aspect is the focus of this study. Early descriptions of loess landform variability were artistic (e.g., in paintings), highly visual, but not quantitative (Xiong et al., 2018; Cheng et al., 2020). With the benefit of rapid developments in remote surveying methods (e.g., unmanned aerial vehicle, lidar and InSAR), high precision topographic data has become more accessible and has facilitated more quantitative approaches (Lane et al., 1993; Hu et al., 2020; He et al., 2021; Xiong et al., 2021).

    View full text