Combining vector and raster data in regionalization: A unified framework for delineating spatial unit boundaries for socio-environmental systems analyses

Abstract

Regionalization has emerged as a crucial research area for the past 50 years, including aggregating smaller areas into larger, contiguous, and/or homogeneous regions. Spatial optimization techniques are advantageous for solving regionalization problems, yet their nondeterministic polynomial-time (NP) hard nature leads to computational complexity and time consumption, especially with extensive datasets. Although regionalization studies play a pivotal role in defining boundaries for multi-scalar analysis and modeling complex socio-environmental systems (SES), current approaches lack integrated consideration of raster and vector data. We introduce a unified, structured framework integrating Geographic Information Systems (GIS), image segmentation, and regionalization to identify and characterize socio-environmental units accounting for various data models and types. We use a public geodatabase of the Rio Grande/Bravo basin, an SES covering diverse cultures, ecosystems, and economies, to demonstrate the functionality of our newly developed method, which effectively identifies spatial units for subsequent SES analysis and modeling. The delineation process accounts for various factors, including administrative boundaries, estimated total quantities, compactness, spatial contiguity, and similarity in socio-environmental characteristics. To make this work reproducible, replicable, and expandable, we developed the approach entirely based on open-source Python packages. Our method is easily transferable to other research using various data formats and spatial scales to delineate spatial unit boundaries effectively and efficiently.