Jayanta Biswas
Spatial Geographer & GeoAI Researcher bridging knowledge-guided deep learning, human mobility downscaling, and spatial epidemiology.
GeoHealth Dynamics Research (GeoHDR) Lab & CAGIS, UNC Charlotte
Advisor: Dr. Yao Li · NSF Award #2451156
McEniry Building, Charlotte, NC, USA
Downscaling Human Mobility (5.5 km → 100 m) for Precision Malaria Modeling in Zambia
About & Research Focus
I am a PhD student in Geography and Graduate Research Assistant at the University of North Carolina at Charlotte, affiliated with the GeoHealth Dynamics Research (GeoHDR) Lab and the Center for Applied Geographic Information Science (CAGIS) under the mentorship of Dr. Yao Li.
My doctoral research focuses on developing knowledge-guided GeoAI architectures—specifically U-Net deep ensembles with mass-conserving (pycnophylactic) loss constraints—to downscale Meta human mobility arrays from a coarse ~5.5 km grid to an ultra-fine 100 m resolution across Zambia. This work, funded by NSF Award #2451156, integrates 38 covariate layers in Google Earth Engine and mobile-phone Call Detail Records (CDR) to pinpoint Human-Vector Contact Zones (HVCZ) for precision malaria transmission containment.
Prior to UNC Charlotte, I completed my M.Sc. in Earth Sciences at The University of Memphis, working with Dr. Angela Antipova on a major TDOT/FHWA study evaluating induced travel demand and long-term land development in Tennessee using 2SLS instrumental variable regression. My foundational training includes both Bachelor's and Master's degrees in Urban and Rural Planning from Khulna University, Bangladesh.
GeoAI & Knowledge-Guided Deep Learning
PyTorch U-Net ensembles, pycnophylactic mass-conserving physics loss, spatial block cross-validation, and uncertainty quantification.
Multi-Source Remote Sensing & GEE
Planetary-scale raster processing on Google Earth Engine, Sentinel-2 time series, Landsat, WorldPop, and environmental ecological indexing (RSEI).
Spatial Epidemiology & Public Health
Modeling disease vectors, transmission contact zones, and disease vulnerability using population-weighted mobility and Voronoi catchments.
Featured Research & GeoAI Projects
Fine-Scale Human Mobility Downscaling for Precision Malaria Modeling
Deep ensemble U-Net downscaling Meta mobility from ~5.5 km to 100 m across Zambia using 38 covariate layers, enforcing pycnophylactic mass conservation and benchmarked against mobile Call Detail Records.
Improved Remote Sensing Ecological Index for Urban Sustainability
Evaluated decadal environmental quality across South Asian cities, benchmarking Rockefeller Foundation 100 Resilient Cities (Pune, Surat) against non-resilient cities (Chattogram, Nagpur).
GeoAI Ensembles for Wildfire Susceptibility Prediction
Comparative assessment of deep learning vs. tree-based machine learning ensembles across California, evaluating computational efficiency and realistic spatial hazard surfaces.
Applying Induced Travel Study in Urban Areas in Tennessee
Analyzed multi-decade land growth (1985–2023) and induced VMT elasticity across individual road facilities using 2SLS instrumental variable regression. Published in Land (2025) and official USDOT/NTL Technical Report RES2024-02.
Remote Sensing of Coastal Tourism on Coral Island Ecosystem
Applied multi-temporal Sentinel-2 imagery (2018–2024) to quantify unregulated tourist infrastructure encroachment on Saint Martin Island with 94.6%–98.9% accuracy.
Interactive Web GIS & Cartographic Laboratory Showcase
Explores Charlotte land use shifts, elevation terrain analysis in Bangladesh, waterbody dynamics, and complete laboratory visualization assignments.
Publications & Technical Reports
Improved remote sensing ecological index for monitoring urban sustainability: 100 resilient cities vs. non-resilient cities in South Asia
Jayanta Biswas, Niloy Biswas
This study develops an improved Remote Sensing Ecological Index (RSEI) to evaluate urban ecological environment quality across four South Asian cities, comparing designated 100 Resilient Cities (Pune and Surat) with non-100RC counterparts (Chattogram and Nagpur) to examine how resilience frameworks impact environmental sustainability over decadal timescales.
Applying Induced Travel Study in Urban Areas in Tennessee
Angela Antipova, Sabya Mishra, Aman Agrawal, Jayanta Biswas
Investigates the causal impacts of highway capacity additions (road widening, new lane construction) on travel behavior and induced vehicle miles traveled (VMT) across urban corridors in Tennessee. By developing disaggregated facility-level instrumental variables, this study overcomes the area-wide aggregation bias of prior state-level research, providing empirical elasticity estimates to inform TDOT project screening, capital allocation, and sustainable regional planning.
Mapping wildfire dynamics: GeoAI-driven comparative analysis of deep and machine learning ensembles for susceptibility prediction in California
Niloy Biswas, Jayanta Biswas, Imtiaz Ul Shahid, Mahmudul Hasan Sabuj
Investigates wildfire susceptibility across California through a rigorous comparative benchmarking of machine learning ensembles and deep neural networks. Evaluates both quantitative predictive accuracy and the spatial realism of hazard surfaces under computational constraints.
Land development 1985–2023 as a function of road improvement, employment, and mobility: A case study of Tennessee
Jayanta Biswas, Anzhelika Antipova
Quantifies the spatial relationships between highway capacity additions, regional employment density, and urban land transformation across Tennessee from 1985 to 2023. Spatial regression models demonstrate that employment access acts as a widespread catalyst for urban expansion, whereas mobility (VMT) exhibits concentrated effects along higher-speed arterials.
Urban environmental degradation in Bangladesh: Insights from a weighted entropy ecological index
Jayanta Biswas
Introduces an objective entropy-weighted formulation of the Remote Sensing Ecological Index (RSEI) to mitigate the subjectivity of PCA-based weighting and better account for moisture and water regimes across 12 major cities in Bangladesh, providing evidence for targeted urban environmental mitigation.
Tourism and environmental change in Saint Martin Island, Bangladesh: Insights from remote sensing data
Jayanta Biswas, Tanmoy Malaker, Taposh Mollick
Utilizes high-resolution Sentinel-2 multispectral time series (2018–2024) to evaluate the environmental footprint of uncontrolled tourism development on Bangladesh's sole coral reef island, reaching classification accuracies of 94.6%–98.9% and documenting critical loss of vegetation and beach habitat.
Mapping and monitoring land use land cover dynamics employing Google Earth Engine and machine learning algorithms on Chattogram, Bangladesh
Jayanta Biswas, Md. Abu Jobaer, Salman F. Haque, Md. S. Islam Shozib, Zamil A. Limon
Leverages cloud computing on Google Earth Engine to train and evaluate supervised machine learning classifiers (Random Forest, SVM, Classification and Regression Trees) across multitemporal Landsat data, charting over two decades of rapid urbanization and vegetative fragmentation in Chattogram.
Spatiotemporal change detection of land use land cover (LULC) in Fashiakhali Wildlife Sanctuary (FKWS) impact area, Bangladesh employing remote sensing
M. S. Hossain, M. A. H. Khan, T. V. Oluwajuwon, Jayanta Biswas, S. M. Rubaiot Abdullah, M. S. S. I. Tanvir, S. Munira, M. N. A. Chowdhury
Employs satellite image processing and GIS-based change detection to analyze habitat encroachment, agricultural conversion, and forest fragmentation in the peripheral buffer zone of Fashiakhali Wildlife Sanctuary over a 30-year period.
Land-use change and forest cover depletion in Bhawal National Park, Gazipur, Bangladesh from 2005 to 2020
Md. Mahfuz Ur Rahman, Tanmoy Dey, Jayanta Biswas
Quantifies the depletion of over 2,284 hectares (~49%) of canopy cover in Bhawal National Park between 2005 and 2020 using Landsat-derived NDVI and land-use transition matrices, identifying industrialization and illegal settlement expansion as core drivers.
Conferences & Presentations
A deep learning framework for fusing multi-modal environmental data to downscale human mobility for precision malaria modeling in Zambia
2026 UCGIS Symposium
Jayanta Biswas†, Noah Kroll†, Amy Wesolowski, Arun Sharma, Shengjie Lai, Andrew Tatem, Yao Li
Presents a novel GeoAI deep learning architecture (U-Net deep ensemble) that fuses 38 environmental and infrastructural covariates from Google Earth Engine with annual WorldPop surfaces to downscale aggregate Meta mobility counts (~5.5 km) to a fine 100 m resolution across Zambia under pycnophylactic mass conservation constraints, enabling accurate Human-Vector Contact Zone (HVCZ) malaria risk mapping.
A spatial perspective on urban land growth: Infrastructure, employment, mobility, and land use change in Tennessee
80th Annual Meeting of the Southeastern Division of the American Association of Geographers (SEDAAG 2025)
Jayanta Biswas
Examines how highway capacity projects, spatial employment density clusters, and vehicular mobility jointly drove parcel-level urban growth across Tennessee between 1985 and 2023 using spatial autoregressive econometric modeling.
Exploring the relationship between road improvements, employment, and land development in Tennessee
2nd Annual Memphis Area Geographic Information Council (MAGIC) Collegiate Geospatial Showcase
Jayanta Biswas, Anzhelika Antipova
First-place award-winning presentation demonstrating spatial modeling workflows to isolate the localized versus regional developmental impacts of road improvements in Tennessee, accounting for accessibility differentials and employment centers.
Exploring the road improvement impact on induced travel in Tennessee
Graduate Student Research Forum
Jayanta Biswas, Anzhelika Antipova
Investigates facility-level induced travel demand across Tennessee highway corridors using 2SLS instrumental variable regression, highlighting policy implications for state transportation funding and capacity screening criteria.
Induced travel: An illusion in traffic demand management
Institute for Intelligent Systems (IIS) Seminar Series
Jayanta Biswas
Critical evaluation of historical traffic demand management paradigms, analyzing whether highway capacity expansions induce demand or merely absorb latent regional mobility needs.
GitHub Repositories
human-mobility-downscaling
Deep learning framework utilizing U-Net ensembles to downscale 5.5 km Facebook/Meta Activity Statistics Mobility rasters to 100 m human mobility surfaces under mass-conserving pycnophylactic constraints.
zambia_mobility
Interactive web mapping dashboard and visualization interface displaying fine-scale (100 m) human mobility layers and malaria risk zones across Zambia.
100RCImproved-RSEI
R workflows and statistical routines implementing the Improved Remote Sensing Ecological Index (RSEI) for comparative urban sustainability assessment in South Asian cities.
Urban-Degradation-using-RSEI
Jupyter notebook implementation of the Weighted Entropy Remote Sensing Ecological Index for quantifying urban environmental degradation across 12 cities in Bangladesh.
TourismRemoteSensing
Satellite image processing and multi-temporal Sentinel-2 classification scripts assessing tourism impact and environmental degradation on Saint Martin Island, Bangladesh.
Geospatial_Workflows
Modular end-to-end geospatial workflows covering spatial machine learning ensembles, wildfire susceptibility modeling, raster processing, and automated spatial feature engineering.
Google-Earth-Engine
Collection of Google Earth Engine (GEE) JavaScript and Python API scripts for satellite imagery classification, indices computation (NDVI, NDWI, NDBI), and LULC dynamics.
charlotte-landuse
Interactive geospatial web map application exploring urban land use zoning and development patterns in Charlotte, North Carolina.
Potential-Accessibility-Analysis
Python and Colab routines implementing potential accessibility measures, travel-time matrix generation, and public transit network modeling for metro lines.
osrm-backend
High-performance routing engine in C++ designed for OpenStreetMap road networks to compute multi-destination matrix routing and shortest path accessibility.
Python-For-GeospatialAnalysis
Comprehensive solutions and code templates for spatial problems using Python libraries (GeoPandas, Shapely, PySAL, Rasterio).
PythonMaps
Curated examples of publication-grade thematic cartography, choropleth maps, and spatial visualizations in Python.
Map-Diary
Creative geospatial visualizations and artistic city maps capturing urban morphology and landscape patterns.
DemoInteractiveMap1
Interactive Web GIS map demo featuring vector layering, custom basemaps, and popup infowindows.
DemoInteractiveMap2
Advanced Web GIS interactive mapping application with spatial filtering and spatial query capabilities.
mountain-bangladesh
Digital elevation and topographic visualization of the Chittagong Hill Tracts and mountainous regions of Bangladesh.
Ecology
Scripts and workflows for landscape ecology metrics, vegetation dynamics, and habitat fragmentation.
Coursera_Capstone
IBM Applied Data Science Capstone project utilizing Foursquare location API and machine learning clustering algorithms for urban venue recommendation.
Mechine-Learning-with-Python-coursera
Supervised machine learning algorithms evaluation and benchmark classifiers (KNN, Decision Tree, SVM, Logistic Regression) with scikit-learn.
biswasjayanta.github.io
Personal GitHub Pages academic and professional portfolio website showcasing research projects, publications, interactive maps, and curriculum vitae.
Education
Ph.D. in Geography
University of North Carolina at Charlotte
Advisor: Dr. Yao Li · Research: Mass-conserving GeoAI downscaling for precision malaria modeling in Zambia (NSF Award #2451156).
M.Sc. in Earth Sciences
The University of Memphis, USA
Thesis: Road Improvement Impact on Induced Travel · Advisor: Dr. Angela Antipova · Outstanding Earth Scientist Nominee.
Graduate Certificate in GIS
The University of Memphis, USA
Master of Urban and Rural Planning (MURP)
Khulna University, Bangladesh
Thesis: First and Last Mile Mobility Operations for MRT-6 Transit Connection, Dhaka.
Bachelor of Urban and Rural Planning (BURP)
Khulna University, Bangladesh
Research Appointments
Graduate Research Assistant
GeoHealth Dynamics Research (GeoHDR) Lab · UNC Charlotte
PI: Dr. Yao Li · Built mass-conserving U-Net ensemble downscaling Meta mobility from 5.5 km to 100 m with 38 covariate layers on GEE (NSF Award #2451156).
Graduate Research Assistant
Department of Earth Sciences · The University of Memphis
PI: Dr. Angela Antipova · Estimated induced travel elasticity for TDOT/FHWA RES2024-02; published in Land (2025) and USDOT/NTL Technical Report RES2024-02.
Research Associate
Planning and Development Research Center (PDRC)
Led feasibility analyses combining spatial modeling and socioeconomic profiling; three master plans approved by city authorities.
Research Assistant
UNDP & SIDA Program on Environment & Climate Change
Conducted extensive household surveys and focus groups on coastal food security and climate adaptation in Bangladesh.
Technical Skills Matrix
- 🏆 First Place Winner, 2nd Annual MAGIC Collegiate Geospatial Showcase (2025)
- 🎓 GSA Award, The University of Memphis (2025)
- 🌟 Nominee, Outstanding Earth Scientist Award (2025)
- 🎖️ NST Fellowship, Govt. of Bangladesh (2022)
Connect & Collaborate
I am always eager to discuss collaborative research in GeoAI, human mobility downscaling, satellite remote sensing, and spatial epidemiology.
McEniry Building · Department of Earth, Environmental, and Geographical Sciences (EEGS)
University of North Carolina at Charlotte · 9201 University City Blvd, Charlotte, NC 28223