In research on global land-use dynamics, a long-standing gap in global change and sustainability-related sciences can be addressed by developing consistent data products on the recent global dynamics in multiple land-use variables. When integrating and harmonizing heterogeneous data sources through advanced statistical modelling techniques, several quality aspects, such as untertainties in the datasets, play a major role.
Time series of global land-use data (e.g. crop, livestock, forest management) are crucial for studying climate change, biodiversity loss, food security and many other contemporary issues. Although relevant to a large number of environmental disciplines, the time series of global land use data (developed in a project as LUCKINet) give rise to a set of challenges relating to data curation, quality assurance and the provision of a consistent data basis for environmental research. Among the challenges, the widespread and unknown gaps and uncertainties in primary observational data, integrated through statistical and computational tools, hamper the quality of the derived gridded global land use maps.
In the GeoKur project, we aim to advance best practices for land-use data quality assurance. We contribute to the validation and continuous quality-improvement of global land-use data by detecting, quantifying and ultimately reducing uncertainties in various sources of primary data inputs into land-use models. Major focus is placed on remotely sensed data on land cover, with an additional focus placed on statistical and point data on land use.
This project thereby supports the development of a next generation of uncertainty-conscious gridded data-products. Moreover, the results from this project will enable targeted efforts to address identified uncertainties in global primary information on land-use and land-cover changes.
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 of 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.
To this end, we develop modelling frameworks addressing different data-quality issues in point grounded data, gridded census and remotely sensed data.
Examples of the work include:
These uncertainty-conscious data models and products will enable more robust applications of land-use data in various scientific fields and policy processes.
The data integration and analysis steps are developed as updatable pipelines to enable dynamic quality-assurance through retrospective updates. Additionally, we are contributing to the development of a standardized ontology, software tools, and best practices for robust land-use data integration and quality-assurance.
S. Ehrmann, C. Meyer, R. Seppelt: Harmonise and integrate heterogeneous areal data with the R package arealDB. in Environmental Modelling and Software, DOI: http://dx.doi.org/10.1016/j.envsoft.2020.104799, August 2020.
We use cookies on our website. Some of them are essential for the operation of the site, while others help us to improve this site and the user experience (tracking cookies). You can decide for yourself whether you want to allow cookies or not. Please note that if you reject them, you may not be able to use all the functionalities of the site.