environmental-modeling

Research project

🐢 Interface for Running NetLogo Simulations from R

Infrastructure and tooling for the NextGen Research DataStream (NRDS)

Heterogenous Graph Attention Transformer for high-resolution, distributed spatiotemporal flood prediction. Published at ACM SIGSPATIAL 2025.

Numerical interpolation and approximation of water pollution data to construct continuous spatial pollution maps.

A Python based kernel to perform spatial (environmental) impact assessment

Spatiotemporal Gaussian process modeling for environmental data: non-stationary PDE prior, deep kernels, multi-fidelity fusion, and A-optimal sampling.非稳态 PDE + 核深度学习 + 多保真 Co-Kriging + 主动采样的物理约束克里金方法，用于复杂时空环境建模与预测

Hydrological Model Continuum (HMC) – Open-source distributed hydrological model for simulation of water cycle processes, including runoff, soil moisture, snow, and routing.

Comparing LightGBM, a vanilla neural net, and a Physics-Informed NN for ETo estimation. The PINN embeds FAO-56 physics in its loss; reproducible synthetic data show physics helps most when measurements are scarce or noisy. Ready-to-run code.

Complementary repository with data and code for Wolf & Tollefsen, 2021.

Advanced framework combining 2D contaminant transport simulation (advection–diffusion) with machine learning to classify ecological risk (Low/Medium/High) and generate dashboards, GIFs, and reports.

Ready-to-run Docker image of TerraME 2.0.1 with LuccME 3.1 for land-use change simulation and cellular automata models (headless and GUI).

Particle-based Python demo of synthetic plume transport through a variable wind field using advection, random-walk diffusion, eddy/shear flow, and density visualization.

Code and processed data accompanying the MSc thesis Integrating Dynamical Systems and Machine Learning for Modeling and Predicting Decay Across Wood Treatments, conducted at DTU in collaboration with the Danish Technological Institute.

Data-driven river water quality modeling using MATLAB and HSY algorithm with simulation and optimization techniques

Reproducible hybrid mGFD-machine-learning defect correction for two-dimensional reactive contaminant transport, with trained PyTorch models, held-out validation, and an irregular-geometry audit.

C++ systems development for environmental modeling, 3D visualization, and performance-critical applications. Combining computational efficiency with domain expertise in environmental science and data systems.

Enhanced Rock Weathering Analysis using USGS Alaska Geochemical Database

Synthetic radiological plume modeling and sensor-network analysis pipeline with Gaussian plume transport, detector thresholds, source reconstruction, Monte Carlo uncertainty, and MATLAB/Python figures.

Open, physics-based solver unifying near-, medium- and far-field dispersion of brine and other buoyant effluents in one coupled framework — with a fully worked desalination-outfall case study on synthetic site data.