GeoLifeCLEF 2024

Species presence prediction based on occurrences data and high-resolution remote sensing images

Motivation

Predicting species presence in an area is vital for ecology and biodiversity conservation. These predictions inform decisions about threatened species, land use planning, protected areas, and eco-friendly agriculture. However, species distribution is influenced by complex local factors that are hard to measure, such as population interactions, landscape connectivity, habitat history, and biases in data collection. Traditional ecological models struggle to account for these factors, leading to coarse-scale resolutions. Additionally, many species are rarely observed due to sampling biases. GeoLifeCLEF aims to evaluate models on an unprecedented scale, covering thousands of species, with a spatial resolution of about 10 meters, and utilizing millions of occurrence data.

Data collection

The 2023 edition of GeoLifeCLEF revealed significant room for dataset improvement. It emphasized the need for more standardized data to enhance predictions in diverse contexts. Therefore, the 2024 edition will introduce new presence-absence data from different parts of Europe thanks to partners from the European Vegetation Archive (EVA) and the network of a large-scale European project on biodiversity monitoring (MAMBO, Horizon EU program). Test sites will be balanced better to represent the diversity of European habitats and regions and the final test set is expected to be composed of several tens of thousands of presence-absence data, as for the validation set. Like 2023, the training data will consist of 5 million non-standardized occurrences from GBIF spanning 38 European countries and over ten thousand plant species. Explanatory variables will include 2023's data with high-resolution remote sensing (e.g., Sentinel-2 RGB-NIR, multi-band Landsat time-series, ASTER elevation raster) and coarser resolution environmental data (e.g., Chelsa climate, SoilGrids, MODIS land use, human footprint, ISRIC soil salinity raster).

Task description

Given a test set of geolocation and year combinations (plot) and given the corresponding high-resolution remote sensing images and environmental covariates, the goal of the task will be to return for each plot the set of species that were inventoried at that location and time by botanical experts over a small area (about 100m²). The test set will include only locations for which an exhaustive plant species inventory is available (i.e., in the form or presence/absence data).

How to participate ?

Coming soon!

Credit

This project has received funding from the European Union’s Horizon research and innovation program under grant agreement No 101060639 (MAMBO project) and No 101060693 (GUARDEN project).