Vector fields as a framework for modelling the mobility of commodities

Area of Science:

• Spatial analysis
• Network science
• Epidemiology

Background:

• Commodity trade data is often sparse, incomplete, and costly to collect, hindering comprehensive analysis of global supply chains.
• Traditional origin-destination (OD) matrices omit unobserved locations, creating data gaps and limiting characterization of movement patterns in underrepresented areas.
• Accurate modeling of commodity flows, like cattle trade, is crucial for disease surveillance (e.g., foot-and-mouth disease) and economic stability.

Purpose of the Study:

• To develop and validate a vector-field-based method for modeling commodity flows using incomplete origin-destination (OD) data.
• To enhance the characterization of spatial and temporal patterns in commodity movements, particularly in underrepresented regions.
• To improve disease transmission modeling and surveillance by accurately inferring flow patterns from sparse datasets.

Main Methods:

• Transforming OD matrices into continuous vector fields to represent spatial flow patterns comprehensively.
• Employing interpolation techniques to effectively handle incomplete and sparse commodity trade datasets.
• Applying the vector-field framework to cattle trade data from Minas Gerais, Brazil, to analyze movement dynamics.

Main Results:

• The vector-field method captures spatial flow patterns and mobility directions more effectively than traditional network approaches.
• The approach demonstrates robustness with incomplete data; over one-third of areas maintained flow direction accuracy (<15 degrees deviation) even when 500 out of 853 areas were removed.
• Analysis of cattle trade data revealed distinct spatial clusters, temporal flow patterns, and seasonal variations linked to breeding, slaughter, and market demands.

Conclusions:

• The developed vector-field framework offers a robust solution for inferring commodity flow patterns from incomplete and sparse datasets, outperforming traditional methods.
• This approach significantly enhances the capacity for spatial analysis in underrepresented areas, crucial for applications like disease spread modeling.
• The study advances the understanding of large-scale commodity trade dynamics and provides a valuable tool for improving animal health surveillance and economic preparedness.