Data4Land carries out preliminary analysis of input raster datasets, requests to the REST APIs to retrieve vector datasets from OpenStreetMap and World Database on Protected Areas to enrich land-use/land-cover maps, harmonisation and rasterisation of vector datasets.

Data4Land also extracts biodiversity stressors based on the user configuration to enriches landscape impedance. The processed outputs may be used for subsequent analysis of habitat connectivity in Graphab, MiraMon or other software.

Further information can be found in Deliverable 6.2, the following softwate paper (Kriukov et al, 2025) and AD4GD blog site.

The developed processing Graphab workflow is formed by a few main components, including Java application, Wrapping main Python script and underlying job, the configuration files defining ecological parameters and commands scheduled to run for each case study and auxiliary Python scripts to harmonise processing outputs, calculate stats, visualise trends, and to access, read, and upload data from/to MinIO object storage.

The Graphab application, along with common commands to calculate habitat connectivity indices, has been implemented on Aston University’s HPC cluster, Nandi, to accelerate processing and improve performance. Read more on Deliverable 6.2 and on GitHub.

AD4GD has developed MiraMon ICT (Index of Terrestrial Connectivity), a stand-alone plugin tool for the MiraMon GIS Software. The original computation of the ICT index was performed pixel by pixel using a large number of sequential batch processing comments, making it difficult to manipulate and execute. The new MiraMon ICT calculates the ICT index more efficiently and significally reducing the execution time.

Dive into the details at Deliverable 6.2 and GitHub.

This tool is used to extract available data for potential subspecies of interest from GBIF, IUCN and ancillary user-defined sources, and enrich it with spatial raster datasets in several steps. Registration is required to access the relevant data from some of the external services like the Digital Observatory for Protected Areas. These were made accessible via our collaboration with the EU’s Knowledge Centre for Biodiversity based at the EU Joint Research Centre (JRC). GBIF2IUCN uses datasets like the species lists, custom conservation information and raster samples and produces enriched tables and ocurrence datacubes. Read more about it in Deliverable 6.2.

AD4GD has tested machine learning (ML) models to predict future Functional Landscape Connectivity using existing land-use/land-cover (LULC) data, comparing results to those obtained from the MiraMon software. A trained AI model was employed to reduce the computational cost and processing time required to translate LULC maps into ICT values. While models trained on insufficiently large datasets produced inconclusive results, the outcomes for the Forest Connectivity Index were sufficiently robust, achieving a drasctic reduction in computation time, from weeks to minutes. This suggests that, with adequate training data, AI models can serve as a viable and highly efficient alternative to tradicional approaches when evaluating the implications of modificating the landscape. 

An in-depth study of accuracy and time take for the models is depiected in Deliverable 5.2.