Abstract
I Introduction
Landforms and terrain features represent Earth surface evolution in relation to endogenous agents (such as geomorphological, geological and hydrological processes) and exogenous processes connected with past to modern climates, ecology and, for more recent times, human activities (Evans, 2012; Goudie and Viles, 2016; James et al., 2013; Tarolli et al., 2019). Thus, an accurate representation of landform in a given area plays an essential role in revealing the formation mechanisms and indicating the landform evolution process. The Geomorpho90 m dataset (Global High-Resolution Geomorphometric Layers) by Amatulli et al. (2020) contributes to this demanding task. It presents the most recent and updated global set of landscape variables, describing the spatial properties (dimensions, slope, curvature, relief) of landforms (morphometry) for the whole globe.
II Structure of the dataset
Starting from the MERIT Hydro Digital Elevation Model (DEM) (Yamazaki et al., 2019), the dataset builds on the authors’ previous work (Amatulli et al., 2018), and it provides a comprehensive, enhanced quantitative suite of land-surface metrics at 90, 100 and 250 m resolution. The dataset comprises 26 layers, organized into three categories of metrics, able to describe the rate of change (Figure 1(a), (d)), the heterogeneity (Figure 1(b), (d)) and geomorphology (Figure 1(c), (e)) of the landscape globally, for a variety of environments.
Examples of layers included in the dataset. (a) Iceland and Greenland: rate of change of terrain in (slope in (d)); (b) Madagascar: heterogeneity of the landscape (maximum multiscale roughness in (e)); (c) Philippines: geomorphology of volcanoes (geomorphons in (f)).
1 Rate of change
The set of variables is the most extensive one, and it provides landscape measures describing the rate of change of the elevation. It comprises 11 layers of first derivative, and five layers of second derivatives.
The first set mainly deals with spatial and frequency distributions of altitude and slope, and it comprises primary attributes calculated directly from the DEM and secondary compound attributes that derive from combinations of primary attributes (slope and upstream contributing area). All of the layers in this set are evaluated, comparing each pixel to its surroundings at a local scale (scale-dependent parameters). The scale-dependent parameters are always carried out considering a cell and its nearest eight neighbors (elevation values of a 3 × 3 cell neighborhood around the processing or center cell).
With the primary attributes, the dataset allows us to quantify: slope and aspect; the north–south or east–west gradient in various forms (aspect sine and cosine, northness and eastness); surface variation with respect to latitude or longitude (east–west and north–south first-order partial derivative); and the relief as a set of convergent (channels) and divergent areas (ridges).
The set of secondary compound attributes allows the investigation and quantification of topographic control on hydrological processes (compound topographic index) and the description of potential flow erosion at any given point in a landscape (stream power index).
The second set of attributes allows the systematic investigation of land curvature, and it also allows: the identification of concavity and convexity in the direction of, or perpendicular to, the slope (profile, tangential curvature); and the description of the rate of change of curvature with respect to latitude, longitude or both (east–west and north–south second-order partial derivative).
2 Heterogeneity
Under the ‘heterogeneity’ umbrella, the authors propose a set of nine variables that describe the variability of elevation values by comparing each pixel to its surroundings at a local scale (scale-dependent parameters), or multiple scales (multiscale parameters).
For the scale-dependent values, the dataset provides: elevation summary statistics (elevation standard deviation); average absolute differences in elevation (terrain ruggedness index); range of elevation (roughness); the dispersion of vectors orthogonal to the terrain surface (vector ruggedness measure); and difference with the average elevation (topographic position index).
Two multiscale measures are provided, to improve the ability to interrogate information about topographically dependent relationships and processes over a range of spatial scales, and they describe: the maximum local elevation z-score, and the scale at which it exists (maximum multiscale deviation, scale of the maximum multiscale deviation); and the maximum surface roughness, and the scale at which the maximum is reached (maximum multiscale roughness, scale of the maximum roughness).
3 Geomorphology
The layer offered represents a comprehensive and exhaustive set of all possible morphological terrain types (geomorphons) (Jasiewicz and Stepinski, 2013). The classification includes standard elements of the landscape, as well as unfamiliar forms rarely found in natural terrestrial surfaces. This layer enables analysis that requires information on terrain attributes as well as landform types.
III Remarks
The quantitative portrait of the Earth’s surface, in a diverse set of environments, and at a range of spatiotemporal scales, is a vital element to determine presently operating processes (Sofia, 2020).
Users interested in this dataset should always keep in mind two significant aspects. Firstly, since most of the parameters are evaluated on a 3 × 3 cell basis, the variable used to characterize the topography of a particular location is dependent on the spatial resolution used in the calculation. The various indices at different resolutions might be capturing different processes and patterns, for the same location.
Secondly, they should consider parameter redundancy. Many of the proposed variables express much the same attribute of the land surface. An example is that of landscape heterogeneity (Figure 2), as shown by the terrain ruggedness index (Figure 2(a)), Topographic position index (Figure 2(b)), maximum multiscale roughness (Figure 2(c)), roughness (Figure 2(d)) and elevation standard deviation (Figure 2(e)). It is evident that despite looking at the same landscape variability, each synthetic roughness parameter is sensitive to different actual topographic roughness, and, thus, either one may work better than others for specific conditions.
Madagascar landscape heterogeneity, as depicted by (a) terrain ruggedness index (TRI), (b) topographic position index (TPI), (c) maximum multiscale roughness (Max Rough), (d) roughness (Rough) and (e) elevation standard deviation (ElStd). Each topographic parameter has been classified into four classes (high, medium–high, medium–low and low) by identifying breakpoints that best group similar values and maximize the differences between classes (natural breaks). Cross-correlation between the various parameters is shown in (g), with values ranging from 0 (no correlation) to 1 (collinear variables).
To guard against the downfall of redundancy, statistical testing should be carefully performed (e.g. cross-correlation; Figure 2(g)), because including all nearly redundant variables can overemphasize their contribution. An index that performs best for all landscapes likely does not exist; rather, the best method is generally site-specific, and there is nothing mathematically right (or wrong) about such a procedure. Users should make a judgment call based on their analytical objectives and knowledge of geomorphology, to help relate individual variables to actual processes in the landscape.
Given its completeness, the dataset will enhance a wide range of geoscience applications, including geomorphology, geology, hydrology and soil science, as well as climate and carbon emissions studies. The proposed layers can offer a preliminary investigation method in land management projects and geological risk assessment and zoning. They further provide geomorphological baseline data that could be used for first-order information about the geographical distribution of biodiversity, facilitating investigations about ecological systems and biology, and the processes that underlie them.
Footnotes
Declaration of conflicting interests
The author(s) has no conflicts of interest to declare.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.