A mathematical programming approach for downscaling multi-layered multi-constraint land-use models

Abstract

Land-use and land-cover change (LULCC) models are important tools for environmental policy planning. LULCC models are frequently constrained to the generation of projections at a specific resolution. However, subsequent studies or models may require finer resolutions. In this work, a downscaling method for LULCC models is proposed that uses a mathematical programming approach to disaggregate the multiple layers of the land-use change projections while respecting a series of constraints. The method is calibrated and validated with MapBiomas data for the years 2000 and 2018 converted for the GLOBIOM-Brazil model, successfully predicting land-use at a finer resolution. Also, as proof of concept, the calibrated model is also applied for GLOBIOM-Brazil projections for 2050. This paper advances the state-of-the-art by proposing and testing a downscaling method using a mathematical programming approach with spatial effects, that operates on multi-layered land-use projections with a range of constraints while allowing flexibility on the number and type of the specific layers and constraints.

Keywords:

• Downscaling
• land-use
• land-cover
• mathematical programming

Author contributions

The downscaling methodology, including the calibration, validation and application steps was proposed, developed, and tested by Rafael G. Ramos. The GLOBIOM-Brazil simulations used in the application step were conducted by Marluce da Cruz Scarabello and Aline Soterroni. Processing of the MapBiomas and the GLOBIOM-Brazil datasets into compatible formats with one another was done by Rafael G. Ramos and Wanderson Costa. The Protected Areas dataset was prepared and pre-processed into an adequate raster format by Pedro R. Andrade. Feedbacks on the methodology and results were provided by Marluce da Cruz Scarabello, Wanderson Costa, Pedro R. Andrade, Aline Soterroni, and Fernando Ramos. The manuscript was written by Rafael G. Ramos, with feedbacks and edits from all the other co-authors. Fernando Ramos was the project coordinator, helping in the planning and integration of this research with the other parts of the RESTORE + project.

Data and codes availability statement

Data and code available at the following link: https://doi.org/10.6084/m9.figshare.21896445

Notes

1 Although land-use and land-cover are conceptually different, for the purposes of the proposed method they are equivalent. Therefore, for simplicity and readability, we will simply use land-use to denote both cases.

Additional information

Funding

This work was supported by the RESTORE + project, which is part of the International Climate Initiative (IKI), supported by the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU).

Notes on contributors

Rafael G. Ramos

Rafael G. Ramos is a Research Affiliate at Brazil’s National Institute for Space Research (INPE) and at the Department of Geography at University of California Santa Barbara. He holds a PhD in Geography from the University of California Santa Barbara. His research focuses on geospatial analysis, particularly on methods for improving the mapping of human and environmental datasets, quantitative methods in human geography, and crime analysis.

Marluce da Cruz Scarabello

Marluce da Cruz Scarabello holds a Ph.D. in Applied Computing from Brazil’s National Institute for Space Research (INPE). She has experience in land use modelling for quantitative climate change policy evaluation in Brazil.

Wanderson Costa

Wanderson Costa is Research Assistant at the National Institute of Space Research (INPE), Brazil, participating in the development of land use models such as the FABLE Calculator Brazil and the GLOBIOM-Brazil. His research areas include land use change modelling, remote sensing, image processing, multi-temporal analysis, geoprocessing, and data mining algorithms.

Pedro R. Andrade

Pedro R. Andrade is Senior Technologist at the National Institute for Space Research (INPE), Brazil. He has experience with social simulation, geoinformatics and environmental modeling, participating in the development of tools such as AdaptaBrasil and TerraME, and the R packages sits and GeoBR.

Aline Soterroni

Aline Soterroni is a Research Fellow at the University of Oxford. She has experience with scenarios modelling for environmental and climate policy evaluation with a particular focus on Brazil. She in the main developer of GLOBIOM-Brazil.

Fernando M. Ramos

Fernando M. Ramos is a Senior Researcher (ret.) at Brazil’s National Institute for Space Research (INPE) in São José dos Campos. He has written over 100 academic papers and book chapters on topics such as scientific computing, time series analysis, optimization, modelling, and simulation.