In this research field, we address a long-standing gap in Global Change and Sustainability-related sciences by developing consistent gridded data products on the recent global dynamics in multiple land-use variables. To this end, we integrate and harmonize heterogeneous data sources through advanced statistical modelling techniques, while addressing various data-quality issues throughout the data life cycle.
This work rests on the shoulders of an international collaborative network, which we coordinate at iDiv. The founding of the Land Use Change Knowledge Integration Network (LUCKINet, www.luckinet.org) was motivated by a shared vision of free, quality-assured, highly detailed and interoperable data products that depict global dynamics in different facets of land use over the past decades. Through different projects, the network develops a series of data products and tools that contribute to longer-lasting, collaborative, and open-science solutions for the continuous monitoring and analysis of global land-use dynamics. This will enable manifold applications in science, policy, and management for which currently available data products are woefully insufficient. The Macroecology and Society lab contributes to the LUCKINet program with several focused research projects (see project descriptions at the end of this page). All data products will be made openly available using FAIR principles.
LUCKINet emphasizes data quality. The accuracy of any modelled data product crucially hinges on the availability of detailed, high quality, and unbiased primary data that are available for model training and validation. Therefore, we invest substantially in the mobilization and integration of primary observational data on land-use from hundreds of disparate sources. We also take great care assessing the uncertainties in those input data and making them visible in the final gridded land-use maps.
LUCKINet emphasizes applicability. Gridded land-use data are vitally needed in many different fields, but different applications place very distinct demands on such data. We cater for these different needs by developing alternative versions of data products that are optimized for a range of typical downstream applications. Our computational routines can, for example, produce gridded data that are consistent with the official countrywide statistics published by national agencies and are thus ready to be used in global assessments to support international policy-making processes. On the other hand, applications tailored to environmental resource managers can provide highest-possible accuracy of the individual pixel values in their focal region by including local expert knowledge and richer covariate information, even if the results deviate from official numbers. Yet other applications make minimal use of model covariates and assumptions to support causal analysis between variables without the risk of circular reasoning.
LUCKINet builds on the power of collaboration. We fully embrace the need to address more challenges than would be feasible for a single research group over the course of a few years. Our vision further acknowledges that not all aspects can be addressed at the same speed or in parallel. Using technologies that support new forms of collaboration, we can address many of these challenges via distributed efforts. We develop workflows that capture standardized metadata that allow tracing data provenance, to also enable future projects to easily pick up where we left off. The open and modular design of our computational workflows facilitates both successive improvements and retrospective updates of the developed data products, as our collaborative network grows and better input data become available.
Our aim is to engage a growing network of contributors from the wider community of land system scientists, remote sensing experts, and geoinformaticians in this endeavor. Together, we will successively increase the list of covered land-use variables, consolidate and improve the employed techniques and, of course, we will collectively make use of these data!
Current projects contributing to this research field at iDiv:
Gridded data products on global land-use dynamics
Detailed global-scale information on spatiotemporal dynamics in crop, livestock and forest production systems are needed to address various environmental and socio-economic issues, and to monitor progress towards multiple Sustainable Development Goals. Despite their importance, gridded data products for those land-use variables are rarely available as historical time-series, are not mutually consistent, and generally suffer from low precision and accuracy. The lack of data consistency and interoperability hampers the joint analysis of dynamics in crop, livestock and forest systems, and their interdependencies.
In this project, we develop global, fully interoperable, gridded data products that depict status and recent historical changes in multiple important land-use variables at high resolutions. These products will provide a first detailed view into the global land-use dynamics over the last few decades, and form the basis for subsequent continuous updates and further enhancements.
To achieve this, we develop and employ new modelling frameworks that integrate data- and theory-driven elements to jointly predict the dynamics of different land-use variables.
We invest substantially in mobilizing and integrating subnational agricultural and forest census statistics from multiple different sources to ensure quality and completeness of the used land-use data sources and to generate gridded data products of highest-possible accuracy. New informatics tools developed in this project support the harmonization and integration of these heterogeneous data sources.
Project lead: Dr Steffen Ehrmann
Uncertainties in global land-use and land-cover datasets
Detailed information on global land use and land cover is crucial for studying climate change, biodiversity loss, food security, and many other contemporary issues. The integration of heterogeneous data sources through advanced statistical and computational tools opens exciting avenues for mapping land-use dynamics globally at fine resolutions. However, due to widespread but unknown gaps and uncertainties in primary observational data, the derived gridded maps inherit tremendous biases that challenge their downstream application.
In this project, we aim to advance the validation and continuous quality-improvement of global land-use and land-cover data. We will detect, quantify, map, and ultimately reduce uncertainties in various sources of primary data on land-use and land-cover changes, such as remotely sensed environmental data, agricultural and forestry census data, and in-situ point observations. We will also assemble an open database with ground-truthing information from existing databases and through a data validation campaign in the citizen-science platform Geo-WIKI.
This project will support the development of a next generation of uncertainty-conscious gridded data-products, including those developed in the MAS group for agricultural and forestry systems and biodiversity habitats. Moreover, the results from this project will enable targeted efforts to close gaps in global primary information on land-use and land-cover changes.
This project also contributes to the development of standardized vocabularies and best practices for land-use data and data-quality assurance. The mobilized data and derived maps will be published using FAIR principles.
Project lead: Caterina Barrasso
Addressing data uncertainties in spatiotemporal statistical modelling of land use dynamics
New developments in mapping global land-use dynamics at high resolution offer exciting avenues for studying issues in climate, biodiversity, food security, and many other research fields. However, the gridded maps inherit tremendous uncertainties from their primary data inputs, which challenges the downstream application.
In this project, we develop statistical and computational tools to address multiple sources of data uncertainty in the spatiotemporal statistical models that are used to develop gridded space-time data cubes on land use and other variables. Our goal is to develop a framework that allows propagating the uncertainties in all data inputs throughout the modelling workflow, to make these visible at pixel-level in the final gridded products. We will implement these tools to improve newly developed time-series products on multiple land-use variables. These uncertainty-conscious data products will enable more robust applications of land-use data in various scientific fields and policy processes.
This project also contributes to the development of standardized vocabularies and best practices for quality assurance in modelled data products.
Project lead: Dr Marina Jimenez
Shedding lights on land-use and biodiversity changes in former USSR and Arabic-speaking countries
Documenting global land-use dynamics and their ecological consequences are central topics in environmental change and sustainability. However, global scientific databases documenting these dynamics are biased by vast data gaps across Asian former-USSR and Middle-Eastern countries. Many datasets indeed exist for these countries, but remain hidden from the eyes of ‘global’ scientific assessments in Russian- and Arabic-language literature and databases. A key reason are persisting intercultural, communication, and collaboration barriers between scientists and practitioners from 'Western' and non-'Western' countries.
This project addresses this long-standing data divide. Led by two scientists with backgrounds in former-USSR and Arabic-speaking countries, and collaborating with regional partners, we will establish contacts to data holding institutions and scientists. Through these contacts, we will seek access to offline databases and non-digital sources in libraries of universities, NGOs, statistical offices, and other institutions. In parallel, we will perform comprehensive online searches for datasets published in Russian- and Arabic-language scientific and grey literature.
We will harmonize and integrate all mobilized databases and newly digitized data into global scientific databases using FAIR principles. Focusing on selected areas in Central Asia, the Caucasus, Siberia, North Africa and the Middle East, we hope to develop a wide collaborative network across these regions.