OTC KPI Dashboard

An interactive dashboard for monitoring the data quality of fCO₂ measurements in Level 2 datasets, as well as the availability of Level 1 Near Real-Time (NRT) data from ICOS ocean stations.

The heatmap displays the monthly percentage of fCO₂ records flagged as good quality (QC = 2) relative to all valid measurements. Dark grey horizontal bars indicate the temporal coverage of the most recent Level 1 NRT data for stations that provide NRT observations. All data are sourced in real time from the ICOS Carbon Portal.

Quick start

# 1. Create a virtual environment
python -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate

# 2. Install dependencies
pip install -r requirements.txt

# 3. Run the dashboard
python app.py

Open http://localhost:8050 in your browser.

You can pre-set the number of months shown by adding ?months=N to the URL, e.g. http://localhost:8050/?months=6.

Production deployment

gunicorn app:server --bind 0.0.0.0:8050 --workers 2 --timeout 300

The --timeout 300 is important because the initial data fetch from ICOS can take a few minutes.

Docker

docker build -t otc-kpi .
docker run -p 8050:8050 -v ~/.icoscp:/root/.icoscp otc-kpi

Configuration

Edit config.py to change:

• CACHE_TTL — how often data is re-fetched from ICOS (default: 6 hours)
• IGNORE_STATIONS — list of station IDs to exclude from the dashboard (e.g. "BE-SOOP-Belgica"). Add or remove entries to control which stations appear in the heatmap.
• COLORMAPS — available colour map options
• DEFAULT_N_MONTHS / DEFAULT_CMAP — initial widget values
• DESCRIPTION — descriptive text shown above the heatmap

Project structure

File	Purpose
app.py	Dash app, layout, callbacks, figure generation
data_fetch.py	ICOS API calls, data processing, incremental disk cache
config.py	Constants, SPARQL queries, settings
assets/style.css	Custom styling (auto-loaded by Dash)
requirements.txt	Python dependencies
Dockerfile	Container deployment
l2_cache.pkl	Auto-generated disk cache for Level 2 processed results

Credits

Developed by the ICOS Carbon Portal with the assistance of Claude (Anthropic).