Pattern-based identification and mapping of landscape types using multi-thematic data

ABSTRACT

Categorical maps of landscape types (LTs) are useful abstractions that simplify spatial and thematic complexity of natural landscapes, thus facilitating land resources management. A local landscape arises from a fusion of patterns of natural themes (such as land cover, landforms, etc.), which makes an unsupervised identification and mapping of LTs difficult. This paper introduces the integrated co-occurrence matrix (INCOMA) – a signature for numerical representation of multi-thematic categorical patterns. INCOMA enables an unsupervised identification and mapping of LTs. The region is tessellated into a large number of local landscapes – patterns of themes over small square-shaped neighborhoods. With local landscapes represented by INCOMA signatures and with dissimilarities between local landscapes calculated using the Jensen-Shannon Divergence ($JSD$), LTs can be identified and mapped using standard clustering or segmentation techniques. Resultant LTs are typically heterogeneous with respect to categories of contributing themes reflecting the human perception of a landscape. LTs calculated by INCOMA are more faithful abstractions of actual landscapes than LTs obtained by the current method of choice – the map overlay. The concept of INCOMA is described, and its application is demonstrated by an unsupervised mapping of LT zones in Europe based on combined patterns of land cover and landforms.

KEYWORDS:

• Spatial patterns
• pattern similarity
• regionalization
• thematic maps
• global categorical datasets

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data and code availability statement

Three global categorical raster datasets used in this study were derived from the following resources: http://maps.elie.ucl.ac.be/CCI/viewer/, https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-land-cover, and https://rmgsc.cr.usgs.gov/outgoing/ecosystems/Global/.

The data that support the findings of this study are available at https://doi.org/10.6084/m9.figshare.13379228.v1.

The R package allowing to create integrated co-occurrence matrices and integrated co-occurrence histograms is available at https://github.com/Nowosad/comat. The R package allowing for identification of landscape types (LTs) is available at https://github.com/Nowosad/motif.

Additional information

Funding

This work was partially supported by the University of Cincinnati Space Exploration Institute.

Notes on contributors

Jakub Nowosad

Jakub Nowosad in an Assistant Professor in the Institute of Geoecology and Geoinformation at Adam Mickiewicz University in Poznan, Poland. His main research is focused on developing and applying spatial methods to broaden our understanding of processes and patterns in the environment. It includes developing, collaborating on, and improving geocomputational methods and software. Jakub is also a co-author of the Geocomputation with R book.

Tomasz F. Stepinski

Tomasz Stepinski is the Thomas Jefferson Chair Professor of Space Exploration at the University of Cincinnati and a Director of Space Informatics Lab. His recent area of research is a development of automated tools for intelligent and intuitive exploration of very large Earth and planetary datasets. He led the team who developed the GeoPAT2 – a toolbox for pattern-based spatial analysis. He is also interested in computational approaches to geodemographics, racial segregation and diversity.