Spatial network surrogates for disentangling complex system structure from spatial embedding of nodes.

Area of Science:

• Network Science
• Spatial Analysis
• Complex Systems

Background:

• Networks embedded in metric spaces are increasingly studied.
• The impact of spatial embedding on network properties is often overlooked.
• Understanding spatial networks requires accounting for node positions.

Purpose of the Study:

• To develop null models for spatial networks that preserve spatial statistics.
• To quantify the influence of spatial embedding on network characteristics.
• To differentiate intrinsic network structure from spatial effects.

Main Methods:

• Proposed a hierarchy of null models for generating random spatial network surrogates.
• Preserved global and local statistics related to node embedding in a metric space.
• Compared real spatial networks with generated surrogates using null models.

Main Results:

• The proposed null models better capture macroscopic properties of real-world spatial networks compared to standard models.
• The framework quantifies how much network characteristics are predetermined by spatial embedding.
• Networks were categorized based on the performance of the proposed null models.

Conclusions:

• The developed framework effectively disentangles complex system structure from spatial embedding.
• This approach is applicable to diverse real-world spatial networks across various scientific fields.
• It provides a robust method for analyzing networks where spatial location is a key factor.