Model skill performance dashboards

This package purpose is to generate static HTML pages used for the dissemination of model skill assessment reports of global surge models.

Features

Regional Statistics Dashboard

• Multi-tab Dashboard: Visual exploration by ocean region including info, mesh, and metric views.
• Key Metrics:
  • Root Mean Square Error (RMSE)
  • Correlation Coefficient (CR)
  • Bias
  • Kling Gupta Efficiency (KGE)
  • Lambda Index
  • Peak Error metrics (R1, R3)
  • More can be added easily
• Visual Components:
  • Histograms
  • Taylor diagrams
  • Scatter plots
  • Spider charts
  • Bathymetric and region maps
• Export Capabilities:
  • Export all tabs to a single PDF file.
  • Save interactive dashboard as standalone HTML.

Historical Storm Events comparison

Visual Components:

• Scatter plots for storm events over 2022-2024
• model vs obs scatter
• time series

Tidal Dashboard

Visual Components:

• Comparitive histograms for main tide constituents
• Interactive map with callback to switch stations
• Model vs obs tidal time series
• Score indicator

Getting Started

1. Install dependencies

We use Poetry for dependency management.

poetry install

2. Launch the Dashboard

python -mpanel serve app.py

• TS data is fetched automatically in data/models/ and data/obs/ and processed (if not already done) using the package.
• Once processed, stats files are saved in data/stats/*.parquet
• Dashboard will fetch meshes in data/meshes/*.gr3 and generate graphics in data/images/*.png
• Regional report will be ready in the browser shortly after

Reproduce

• Create the data folder and subfolders. By default, the application uses data from the data/ folder. To test with a different folder (e.g., data_demo/), set the data_dir as initial varible in the dashboard:

instance = RegionalDashboard(data_dir="data_swl")

follow the directory structure below:

Directory Structure

├── app.py                      # Example app for launching the dashboard
├── auto_reports                # Core functions
│   ├── graphs/                 # All visual components
│   ├── _io.py                  # Data loading and preprocessing
│   ├── _markdown.py            # Markdown content for documentation
│   ├── _proj.py                # [Not used] spilhaus projected mesh maps
│   ├── _render.py              # Plot style and layout settings
│   ├── _stats.py               # Statistical computations
│   ├── _storms.py              # Storms definition
│   └── _tide.py                # (De)Tide functions
├── dashboards                  # Thematic dashboards
│   ├── regional_dashboard.py
│   ├── storms_dashboard.py
│   └── tidal_dashboard.py
├── data_demo
│   ├── models                  # Folder containing "{model}/{station}.parquet" files
│   │   ├── ...
│   │   └── stofs2d
│   │       ├── ...
│   │       └── naha.parquet
│   ├── obs                     # Folder containing "{station}_{sensor}.parquet" observation files
│   │   ├── ...
│   │   └── naha_rad.parquet
│   └── stats                   # Folder containing stats files used in dashboards
│       ├── ...
│       ├── stofs2d_eva.parquet
│       ├── stofs2d.parquet
│       └── stofs2d_tide.parquet
├── LICENSE
├── poetry.lock                 # Project dependencies
├── pyproject.toml              # Project dependencies
├── README.md                   # You're here
└── thumbnails                  # Images for README and landing page
    ├── dashboard_tide.png
    ├── global_metrics_twl.png
    └── storm_events_twl.png