Program Structure

Module 1: Why HPC for Environmental Science?

• Why environmental simulations are compute-heavy: resolution, time steps, ensembles.
• Model types: weather/climate, air quality, hydrology, ocean, land surface (overview).
• Scaling challenges: memory, I/O, and runtime constraints.
• HPC workflow mindset: prepare → run → validate → analyze → iterate.

Module 2: HPC Foundations (Cluster Basics + Linux Workflow)

• Cluster architecture: login node vs compute nodes, cores, memory, GPUs (overview).
• Linux essentials for HPC: directories, permissions, SSH, modules, environment variables.
• Storage and file systems: scratch vs home, quotas, and data organization.
• Best practices: naming, run folders, logs, and checkpointing.

Module 3: Parallel Computing Concepts (MPI, OpenMP, GPUs)

• Shared vs distributed memory: why it changes how programs run.
• MPI concept: message passing and multi-node scaling.
• OpenMP concept: multi-threading on a single node.
• GPU overview: when acceleration helps (and when it doesn’t).

Module 4: Job Scheduling & Resource Management

• Schedulers overview: SLURM/PBS/LSF concepts and queue etiquette.
• Writing job scripts: CPU, memory, walltime, modules, and output logs.
• Monitoring jobs: status checks, failed runs, and restart strategies.
• Efficient usage: right-sizing resources to avoid slow queues or wasted cores.

Module 5: Environmental Simulation Workflows (Model-Agnostic)

• Input preparation: grids, boundary conditions, initial conditions (conceptual).
• Spin-up and stability: why models need warm-up time.
• Running ensembles: uncertainty, scenario comparisons, and sensitivity runs.
• Validation approach: comparing outputs with observations (basic workflow).

Module 6: Climate & Weather Modeling on HPC (Concept + Run Pipeline)

• Resolution and time step trade-offs: accuracy vs runtime.
• Downscaling concepts: regional modeling and nested grids (overview).
• Typical outputs: NetCDF, gridded fields, time series extraction.
• Operational workflows: running repeated cycles and managing output volumes.

Module 7: Air Quality, Dispersion & Environmental Exposure Simulations

• Dispersion modeling concepts: emissions, chemistry, transport (overview).
• Urban-scale challenges: street canyons, high-resolution grids, data needs.
• Computational bottlenecks: chemistry solvers, I/O, and parallel scaling issues.
• Exposure mapping pipeline: gridded outputs → population overlays (overview).

Module 8: Hydrology, Flood & Watershed Modeling

• Watershed modeling basics: rainfall-runoff and routing concepts.
• Flood simulation overview: terrain, boundary conditions, and time stepping.
• Coupling with weather data: precipitation uncertainty and scenario runs.
• Outputs: hydrographs, inundation maps, and risk metrics.

Module 9: Post-Processing, Visualization & Big Output Handling

• Working with large files: chunking, compression, and efficient reading.
• Common formats: NetCDF/HDF5 (overview) and metadata discipline.
• Visualization pipeline: maps, time series, anomalies, and scenario comparisons.
• Automating analysis: scripts, reproducible notebooks, and summary reports.

Module 10: Performance Tuning & Reproducible HPC Science

• Performance metrics: strong scaling vs weak scaling (concept).
• Bottlenecks: CPU, memory bandwidth, network, and I/O.
• Profiling overview: what to measure before “optimizing.”
• Reproducibility: environment capture, versioning, containers concept (overview).

Final Project

• Create an HPC Environmental Simulation Run Plan + Results Report.
• Choose one scenario: air quality episode, flood event, heatwave simulation, or watershed analysis.
• Include: inputs, compute plan (nodes/cores/time), job script outline, output management plan, validation approach, and key plots/metrics.
• Example projects: city air dispersion run plan, flood inundation scenario set, climate downscaling workflow blueprint, ensemble-based watershed risk analysis.