Adaptation of SLEUTH cellular automata based urban growth model for high-performance system: This project aims at developing high-performance version of the popular SLEUTH cellular automata model to simulate urban growth in megaregions. Land use change models have seen a growth in use due to the availability of multi-temporal, high resolution images, and computational power. While it is widely adopted and highly regarded, the computational complexity of the brute-force parameter sweep style of execution leads to long run times, which is further exacerbated by large area and high resolution images. In this study, we developed a distributed framework for SLEUTH, called DSLEUTH, for executing and managing subsets of the parameter space on a shared memory machine which is published in SpringSim 2019 (Chaudhuri and Foley, 2019). Currently, we are exploring the model performance and its relationship to urban typology and physiography. Our goal is to adapt the model to effectively harness the power of advance computing techniques and variety of geospatial data, and make it easily accessible to non-technical practitioners and scholars of urbanization. Within that context, our group is working on: conversion of the desktop SLEUTH model from C++ to Python, adaptation of the SLEUTH on Kubernetes platform and National Science Foundation's (NSF) Extreme Science and Engineering Discovery Environment (XSEDE) platform, evaluation of machine learning calibration approach, and development of new visualization approaches to display model results.
COMPLETED RESEARCH PROJECTS
Application of deep learning classification to map land cover in very high resolution imagery - Aquatic Invasive Plants (AIPs) are a global threat to biodiversity, ecosystem degradation, economies and human health. AIPs are known for their rapid and effective adaptation to the new environments. They benefit from the ecosystem changes and habitat disturbances by global climatic changes and anthropogenic impacts. Lythrum salicaria, commonly known as Purple Loosestrife is one of the most important AIPs which has invaded all counties in Wisconsin and has been designated as a deadly threat to the state’s wetlands. Accurately estimating the current extent of AIPs and slowing their future spread has been identified as a top priority for the WI-DNR. Unmanned aerial vehicles (UAV) have been used to acquire images with very high spatial resolution in a variety of studies in agriculture and forestry and can reveal both high spatial and temporal detail on vegetation patterns. The use of computer vision and emerging technologies like Deep Learning (DL) techniques, also known as self-learning artificial intelligence approaches, allow the incorporation of expert knowledge to the automatic processing of these images. In this research we aim to develop a deep learning based semantic image segmentation and classification methodology. Specifically, this research aims to develop an image processing pipeline that will take advantage of the field validated very high-resolution UAV imagery and apply a deep learning based semantic segmentation approach and transfer learning to detect the presence and density of the Purple Loosestrife patches in selected study sites. The results have been shared at the Mississippi River Research Consortium (2022), Ecological Society of America Conference (2022), and published in Remote Sensing journal (2023). The model that was developed is available via Github repository.
The Geography of Long-Term Care: Implications for SSI and Understanding Disparities in Living Arrangements Among Older Adults(Co-PI) - Research reveals that Black older adults are overrepresented in nursing homes and underrepresented in assisted living. It could be that community-based care facilities are more likely to locate in predominantly White areas. Using a national business database and a “racial landscapes” approach to characterizing local demographics, this project explores associations between the locations of LTC facilities by type across predominantly White and non-White areas of the U.S. Across the continental U.S. we find all types of LTC are less abundant in predominantly White than in predominantly non-White areas. Yet, these differences are almost entirely mediated by socioeconomic factors. Estimates based only on metropolitan areas indicate different relationships by facility type. Assisted living without nursing care and nursing homes are more abundant in predominantly White areas than in non-White areas, but the reverse is true for adult day centers and assisted living with nursing care. We find that historical redlining grades mediate these metro estimates. We also compare estimates across states that do and do not offer supplements to federal Supplemental Security Income (SSI) payments and by Medicaid Home and Community-Based Care (HCBS) waiver participation. We find higher abundance of non-nursing home facilities in predominantly non-White communities than in White communities in states that supplement SSI, but no consistent association with the adoption of HCBS waiver programs. Although our main estimates differ across states that do and do not supplement federal SSI payments, the mix of LTC offerings in a county does not appear to be predictive of county-level SSI enrollment. Project Team includes Mary Hamman, UWL -PI, Stephanie Robert, UW Madison -PI, Gargi Chaudhuri, UWL Co-PI, Megan Jenkins Morales UW Madison Co-PI, Milanika Turner, FAMU, Co-PI.
Time-series analysis of South Asian Megacities (PI) - This project focuses on mapping nonlinear land transitions in and around urban areas and its effect on local micro-climatic systems. Our results are published in the Environment and Planning B (2022) where we used time-series analysis with MODIS data in Delhi Metropolitan Area, India to map land changes within urban area. The results reveal different non-linear trajectories that, until recently, could only be observed via long-term surveys and/or local knowledge. We plan to expand our work in other rapidly urbanizing South Asian cities to better understand the phenomenon.
Land use and land cover change in Ganges Brahmaputra Delta Region(PI): The project is aimed at furthering our understanding of the impact of climate change and land use policy on land use and land cover change. In particular, detailed land use maps derived from satellite imageries will be integrated in time and space to answer questions about how different policies alter the sequencing of land use changes in the same region and under the same driving forces but under different implementation agencies (Indian and Bangladeshi governments). The goal is to examine and forecast these non-linear feedback spatially. The project will have three phases: (i) Data fusion for long term land use change trajectories; (ii) Land use change modeling under multiple scenarios; and (iii) Coupling forecast results with socio-economic and policy data. Some output from this on-going project includes, poster presentations at NASA LULC meeting 2012 in Coimbatore, India (Mishra and Chaudhuri, 2012), International Cartographic Conference 2013 in Dresden, Germany (Chaudhuri, 2013), and published paper on LULC change patterns in Applied Geography (Chaudhuri and Mishra, 2016). The last paper published in Computer, Environment and Urban Systems (Chaudhuri and Clarke, 2019) under this project focused on application of SLEUTH in large highly heterogeneous region region with medium spatial resolution via distributed modeling, and effective simulation of drivers of urbanization.
Crime Analysis(PI):Extending some student research interest, has lead to some collaborative work in crime analysis. The focus of crime analysis research has been mostly on application of geocomputational technique in pattern analysis and predictive modeling. This has lead to one student co-authored book chapter on Liquor Law violation in La Crosse during Oktoberfest (Chaudhuri et al, 2014), and some more undergraduate student research work. Further, the project aims at spatio-temporal visualization of time series of crime data, which has resulted in regional conference presentation and a manuscript which is currently under preparation.
Data analysis and visualization of Asian Carp in Mississippi river (Co-PI): This project aims at spatial and temporal visualization analysis of Illinois River Asian carp catch data, based on results from previous data exploration of the Brandon Road and Dresden Island pools' catch data. An online visualization tool for exploring clustering analysis results will be developed for incorporation with the Illinois River Catch database for integrated data exploration capabilities. This work will continue to support Asian carp removal and other management actions in the Illinois River Waterway.
DOCTORAL DISSERTATION RESEARCH
Title: Impact of policy on interaction between land use – road network and its change over time
Advisor: Dr. Keith C. Clarke, Dept. of Geography, University of California Santa Barbara
Abstract: Dr. Chaudhuri’s PhD dissertation research, under the supervision of Prof. Keith Clarke, focused on capturing and simulating the impact of differential policies and checkered historical background on urbanization of twin cities in trans-border region of Italy and Slovenia. She used SLEUTH, cellular automata based land use change model and multi-temporal remotely sensed satellite images to capture and simulate the historical growth of the twin cities and analyzed the impact of differential policies and historical background on urbanization via scenario forecasting. This chapter of her dissertation was published as a peer-reviewed paper in Journal of Land Use Science (Chaudhuri and Clarke, 2013). The dissertation further explored the influence of checkered history on the interaction between changes in road networks and land use. She used network analysis measures and spatial statistical analysis to capture the dynamic relationship between land use and road networks over multiple time periods and at varying spatial scales. The results of this part are published in Environment and Planning B (Chaudhuri and Clarke, 2015).The dissertation proposal won University of California Transportation Center Dissertation Research Grant in 2009. In addition to application of SLEUTH to model urbanization, Dr. Chaudhuri is interested in different aspects of model development and improvisation itself. For the same study area, she tested the effect of temporal distribution of the input data in uncertainties of simulation outputs. This study was published as a peer-reviewed extended abstract in proceedings of Geocomputation 2011 (Chaudhuri and Clarke, 2011). The same study was further expanded and different map comparison techniques were used to measure disagreement between simulated and observed data. The results were published as a peer-reviewed paper in Transactions in GIS (Chaudhuri and Clarke, 2014). Being an open-source and easily adaptable land use change simulation model, SLEUTH has been used extensively over the past 15 years to model urban growth in various parts of the world. As a result of her interest in SLEUTH modeling and its applications, Dr. Chaudhuri developed an online repository where she gathered and tracked various developments and improvisation of the SLEUTH model. This led to a co-authored paper on different new developments of SLEUTH, which was published in International Journal of Environmental Resources Research (Chaudhuri and Clarke, 2013). In AAG 2012, Dr. Chaudhuri co-organized a symposium (2 sessions and a panel discussion) with Dr. Claire Jantz of Shippenburg University on the developments of SLEUTH where they invited several speakers from around the world to present their ongoing work with the model - http://www.ncgia.ucsb.edu/projects/gig/Repository/AAG2012_Symposium.html
MASTERS THESIS RESEARCH
Title: Transportation Infrastructure of Kolkata, India (2000 - 2005)
Advisor: Dr. Apurba R. Ghosh, Dept. of Geography, University of Calcutta
Abstract: The transport system of Kolkata is a mix of modern mass rapid transport and the old transport modalities like the rickshaws. Kolkata is connected to the rest of India by the National Highways, the extensive network of the Indian Railways, and also by air. Major trafficking of the North-East India takes place via Kolkata. The thesis looked at 4 major modes of transportation in Kolkata today: (i) road transport: the present condition of the road transport, available modes of road transportation, future developments of the road infrastructure, such as: widening of roads, making new ways, flyovers, road over bridges, with special emphasis has been given on tram services were discussed; (ii) water transport: a comparative study of the state government launch services and the co-operative launch services were been shown. Discussions on cargo ships, water traffic, and dying condition of the Calcutta Port along with the future developmental projects that are being undertaken to relift its condition, were discussed; (iii) air transport: discussion on air connectivity of Kolkata and its future development projects are being discussed. As well as map showing the air linkages between Kolkata and the other parts of the country is being shown along with the cartograms illustrating the numerical data of the present status of the air transport in the city are presented; (iv) railways:discussion on railways were further divided into: Circular Railways, Metro Railways, Local Railways joining the sub-urban areas and Express trains joining the city with other parts of the country. Quantitative analysis and future developments in each of them were discussed. In the conclusion, the future development of the transport network of Kolkata is being discussed about; along with the implementation of the new projects and those which are yet to be implemented and how this can further be enhanced to make the transport condition of the city better.
PROJECTS WORKED AS RESEARCH ASSISTANT
Interactive Campus Map, University of California, Santa Barbara (Supervisor: Dr. Keith Clarke)
As a graduate student herself, Dr. Chaudhuri worked with undergraduate and graduate interns to create and update the resourceful interactive map via ArcGIS server on UCSB’s campus map project
Spatial Database for Off-shore Energy and Mineral Resources Infrastructure, Pacific West Coast Region, University of California, Santa Barbara (Supervisor: Dr. Michael Goodchild)
Travel behavior simulation of southern California residents, University of California, Santa Barbara (Supervisor: Dr. Kostas Goulias)
Besides Dr. Chaudhuri’s primary research, during her graduate studies, she has had the opportunity to work on two projects with outstanding groups of people. The first experience was as a research assistant to Prof. Kostas Goulias for his project on travel behavior simulation of southern California residents. In this project, funded by Southern California Association of Government (SCAG), she worked on areal interpolation of multivariate transportation data from multiple sources and at varying hierarchical levels (such as, Parcel layer, Block Administrative layer, Traffic Analysis Zone layer, and County layer) with misaligned boundaries. ArcGIS was used with a customized python script to solve the problem whose objective was to use this interactive information in activity based travel models.
Photo details: The photo above was taken by from Basilique Notre-Dame de la Garde overlooking the beautiful city of Marseille, France.