PySGN: A Python package for constructing synthetic geospatial networks

In previous posts, we have written about the generation of synthetic populations based on real world locations, and how such populations can have various types of networks associated with them. We have also written about network generation techniques in the past and keeping with this line of research, Boyu Wang, Taylor Anderson, Andreas Züfle and myself have a new paper in the Journal of Open Source Software entitled "PySGN: A Python Package for Constructing Synthetic Geospatial Networks"

In this paper we introduce a Python package that can generate geospatial networks which we have called PySGN (Python for Synthetic Geospatial Networks). For readers not familiar with geospatial networks, to quote from the online documentation we have put together:  

Geospatial networks are a type of network where nodes are associated with specific geographic locations. These networks are used to model and analyze spatial relationships and interactions, such as transportation systems, communication networks, and social interactions within geographic constraints. By incorporating spatial data, geospatial networks provide insights into how location influences connectivity and network dynamics.

PySGN generates synthetic geosocial networks that mimic the spatial relationships observed in real‑world networks as it embeds nodes in geographic coordinate space, modifies connection rules to decay with distance, and allows users to incorporate clustering and preferential attachment while respecting spatial constraints. Online we provide examples of creating Geospatial Erdős-Rényi, Watts-Strogatz & Barabási-Albert Networks along with ways to sample points based on a specified bounding box or specific polygon boundaries (examples of which are shown below). 

The package is intended for researchers and practitioners in fields such as urban planning, epidemiology, infrastructure resilience and social science who require robust tools for simulating and analyzing complex geospatial networks. In addition to the paper, we have also made available extensive documentation (along with examples of the various network types) at https://pysgn.readthedocs.io/en/ 

Examples of Geospatial Erdős-Rényi and Watts-Strogatz Networks.

Example of Geospatial Barabási-Albert Network based on different ordering strategies for how nodes are added to the network.

Examples of sampling points based on a bounding box or a specific set of polygons

Full Referece: 

Wang, B., Crooks, A.T., Anderson, T., and Züfle, A. (2026), PySGN: A Python Package for Constructing Synthetic Geospatial Networks. Journal of Open Source Software, 11(119), 9346, https://doi.org/10.21105/joss.09346 (pdf)

A hybrid simulation methodology for identifying and mitigating supply chain disruptions

Durring times of crisis, shocks to supply chains can propagate through the entire economy (e.g., global shortages of critical goods, such as personal protective equipment during COVID-19). At the same time, criminal organizations may disrupt and manipulate licit supply chains for financial gain or political objectives.  Thus there is a strong need for modeling and simulating not only supply chain operations but also malicious actors who may act to disrupt them. 

To explore this we (Abhisekh Rana, Raj Patel, Aman Goswami, Sean Luke, Alok Baveja, Carlotta Domeniconi, Benjamin Melamed, Fred Roberts, Weiwei Chen, Vladimir Menkov, Viswanath Narayan, James Jones, Hamdi Kavak and myself) have a new paper entitled  "Hybrid simulation methodology for identifying and mitigating supply chain disruptions" which was recently published in the Journal of Simulation. 

In the paper we introduce a novel hybrid modeling framework (implemented in MASON) designed to identify vulnerabilities across supply networks. Through the framework, we are able to analyze disruption scenarios  and evaluate mitigation strategies using a pharmaceutical supply chain model (i.e., PharmaSim). As such this paper and proposed framework provides a foundation for simulation-driven planning tools that help organizations anticipate risks and strengthen supply chain resilience.

If this sounds of interest, below we provide the abstract to the paper, some of the figures which show the supply chain we model and the simulation framework along with some results. While at the bottom of the page, you can find the full referece to the paper and a link to it, while the model itself is available at https://github.com/eclab/DES-Supply-Chain-demo. 

Abstract

Global disruptions have shown that shocks to supply chains can quickly ripple through entire economies, highlighting the need to identify vulnerabilities and evaluate mitigation strategies to build resilience. In this paper, we propose a simulation methodology, Hybrid Integrated Supply-Chain Simulation (HISS), to identify and mitigate potential disruptions in supply chains. We demonstrate HISS using a generic pharmaceutical supply chain model including sourcing, outsourcing, production, packaging, and distribution processes, created using MASON’s hybrid modeling capabilities. We classify disruptions from malicious actors and analyze their timing, impact, and scope. The simulation is further extended to modeling mitigation strategies and assessing their efficacy. Extensive optimization allowed us to identify worst-case disruptions and optimized safety stock strategies reduced impacts by a factor of five, while anomaly detection achieved a high recall of 0.966. The modeling approach proposed in this paper provides a basis for planning tools that support resilience and preparedness of supply chains.

Keywords: Hybrid simulation, supply chains modeling, resilience, optimization, evolutionary computation. 

Visual representation of pharmaceutical supply chain (PSC), which was used to code PharmaSim

Time series of daily production flow through the active pharmaceutical ingredient (API) Production node (resilience triangles are shown in red and the number of units on the vertical axis is in millions).

Overview of the software components and their interactions.

Sample time series of numbers of packaged units with anomalies due to (left) a disruption and due to (right) normal fluctuations (the number of units on the vertical axis is in millions).

Full reference:

Rana, A., Patel, R., Goswami, A., Luke, S., Baveja, A., Domeniconi, C., Melamed, B., Roberts, F., Chen, W., Crooks, A.T., Menkov, V., Narayan, V., Jones, J. and Kavak, H. (2026). A hybrid simulation methodology for identifying and mitigating supply chain disruptions. Journal of Simulation, 1–22. https://doi.org/10.1080/17477778.2026.2628944 (pdf)

Driving Anxiety and Visual Attention in Young Drivers

Over the last summer I participated in a research experience for undergraduate at the University at Buffalo (UB) hosted by the Geologic and Climate Hazards Center. In this program students spent several weeks at UB working with faculty on a diverse set of projects ranging from understanding snow events over the great lakes, forest die off to utilizing crowdsourced data to study dust events. One of the projects I was involved with resulted in a poster being presented at the 105th Transportation Research Board (TRB) Annual Meeting entitled "Driving Anxiety and Visual Attention in Young Drivers: A Driving Simulator Study". 

In this study Phoebe Schrag worked alongside Austin Angulo, Irina Benedykc, Gongda Yu, Daisha Cardenas, Hayden Radel (from UB's Transportation Research and Visualization Laboratory (TRAVL)) and myself to explore driving anxiety of young drivers between the ages of 18 and 25. Using eye-tracking data from a high-fidelity virtual reality (VR) driving simulator we explored the effects of self-reported driving anxiety on visual attention, decision-making, and cognitive load. We found that driving anxiety can impair situational awareness. If this sounds of interest and you want to find out more, below you can read the abstract to the poster along with a link to the actual poster itself. 

Abstract: 

Motor vehicle crashes have remained the second leading cause of death among adolescents and young adults in the United States. Although high crash rates are commonly attributed to inexperience, risk-taking behavior, and underdeveloped executive functions, the role of emotional factors such as driving anxiety remain under-explored. Driving anxiety, which is characterized by persistent fear or worry while driving, may have a significant impact on young or novice drivers due to their limited experience and developing emotional regulation abilities. However, existing research has relied heavily on adult samples, self-report measures, or clinical cases, rarely incorporating real-time behavioral data from young adults. This study addresses these gaps by using eye-tracking in a high-fidelity virtual reality (VR) driving simulator to objectively evaluate the effects of self-reported driving anxiety on visual attention, decision-making, and cognitive load. Thirty-one licensed drivers aged 18–25 were classified into anxiety and non-anxiety groups using a questionnaire with reference to the Driving Cognitions Questionnaire. Participants completed five mixed urban-rural scenarios (two dynamic, two static, and one repeated dynamic) while wearing a Varjo XR3 headset with iMotions eye-tracking monitoring. Key eye-tracking metrics (e.g., dwell time proportion, fixation duration and saccade count) were analyzed using scenario-specific Welch’s t-tests (α = 0.05). The results showed that anxious drivers had significantly fewer saccadic movements in high-demand scenarios, indicating reduced scanning and increased cognitive load. These findings demonstrate how driving anxiety can impair situational awareness and suggest that targeted psychological interventions could improve attentional control. This work informs emotionally adaptive driver training for young drivers.

KEYWORDS: Driving Anxiety, Eye-Tracking, Visual Attention, Young Drivers, Cognitive Load, VR Driving Simulation.

Full Reference:

Schrag, P., Yu, G., Cardenas, D., Radel, H., Angulo, A., Crooks, A.T. and Benedyk, I. (2026), Driving Anxiety and Visual Attention in Young Drivers: A Driving Simulator Study, 105th Transportation Research Board (TRB) Annual Meeting, 11th – 15th January, Washington DC. (poster pdf)

Not just numbers: Understanding cities through their words

In the past we have written how one can use social media or newspapers to study the world around us. Keeping with this theme of using text we (Xinyu Fu, Catherine Brinkley, Thomas Sanchez, Chaosu Li and myself) have a new editorial entitled "Not just numbers: Understanding cities through their words" which accompanies a special issue in Environment and Planning B entitled "Leveraging Natural Language Processing for Urban Analytics" 

The editorial discusses how researchers can use natural language processing  (NLP) methods to get a sense of a diverse range of issues impacting cities. To quote from the editorial, these range: 

 "from  analyzing housing development from council planning applications (Lin et al., 2025), revealing visitor perceptions of famous attractions or passengers’ perceptions on transit service quality from social media (Luo et al., 2025; Ma et al., 2025), defining the meaning of urban imageability based on online review (Zhu et al., 2025), understanding the spatial implications of the digital economy (Occhini et al., 2025), and extracting policies from official government reports (Wang et al., 2025)."

These papers, along with the data they used, and findings are summarized in the table below, and as such demonstrate how one can move beyond purely quantitative data and methods to study cities. If this sounds of interest, please feel free to read our editorial along with the papers in the special issue. 

Full reference: 

Fu, X., Brinkley, C., Sanchez, T.W., Li, C. and Crooks, A.T. (2026), Not Just Numbers: Understanding Cities through their Words, Environment and Planning B, 53(1): 3-10. (pdf)

Creating and Assessing an Unconventional Global Database of Dust Storms Utilizing Generative AI

In the past we have written about how one can use social media to monitor dust storms along with how multi-modal large language models (MLLMs) can be used to analyze images. At the recent American Geophysical Union (AGU) Fall Meeting we (Sage Keidel, Stuart Evans and myself) brought these two strands of research together in a poster entitled "Creating and Assessing an Unconventional Global Database of Dust Storms Utilizing Generative AI."

In this work we showcase how MLLMs are providing new opportunities and accessible methods for information extraction from imagery data using geo-located images from Flickr which have a dust keyword tag associated with it from multiple languages (e.g., Arabic, English, Spanish).  We run these images through ChatGPT, which classifies them as dust storms or not and compare this classification with human classifed images. If this sounds of interest, below you can read the abstract, see the poster along with a selection of images that have been labeled as as dust storm or not and ChatGPTs confidence in its classification. While the dust storm database itself can be found here

Abstract:

Complete observations of dust events are difficult, as dust’s spatial and temporal variability means satellites may miss dust due to overpass time or cloud coverage, while ground stations may miss dust due to not being in the plume. As a result, an unknown number of dust events go unrecorded in traditional datasets. Dust’s importance both for atmospheric processes and as a health and travel hazard makes detecting dust events whenever possible important, and in particular, studies of the health impacts of dust are limited by detailed exposure information. 

In recent years, social media platforms have emerged as a valuable source of unconventional data to study events such as earthquakes and flooding around the world. However, one challenge with respect to using such data is classifying and labeling it (i.e., is it a dust storm or not?). While it is relatively simple to classify textural data through natural language processing, it is not the case with imagery data. Traditionally, classifying imagery data was a complex computer vision task. However, recent advancements in generative artificial intelligence (AI) especially multi-modal large language models (MLLMs) are opening up new opportunities and offering accessible methods for information extraction from imagery data. Therefore, in this study we collected geotagged Flickr images referencing dust from around the globe from multiple languages (e.g., English, Spanish, Arabic) and use generative AI (i.e., ChatGPT) to classify the images as dust storms or not. Furthermore, we compare a sample of these classified images from ChatGPT with human classified images to assess its accuracy in classification. Our results suggest that ChatGPT can relatively accurately detect dust storms from Flickr images and thus helps us create an unconventional global database of dust storm events that might otherwise go unobserved from more traditional datasets.

Workflow

Poster

Dust storm database (click here to go to it)

Full Referece: 

Keidel, S., Evans S. and Crooks, A.T. (2025), Creating and Assessing an Unconventional Global Database of Dust Storms Utilizing Generative AI, American Geophysical Union (AGU) Fall Meeting, 15th–19th December, New Orleans, LA. (pdf of poster).