noaa-cdo-api 0.0.3

A NOAA NCEI API wrapper for weather data

NOAA Climate Data Online API Client

An asynchronous Python client for the NOAA National Centers for Environmental Information (NCEI) Climate Data Online (CDO) Web Services API v2. Features automatic rate limiting, connection pooling, and comprehensive type safety.

Features

• ⚡ Asynchronous API: Built with aiohttp for high-performance async I/O
• 🚦 Automatic Rate Limiting: Enforces NOAA's limits (5 req/sec, 10,000 req/day)
• 🔄 Connection Pooling: Efficient TCP connection reuse
• 📝 Type Safety: Full type hints and runtime validation
• 🎨 Beautiful Documentation: Color-formatted docstrings with pdoc
• 🛡️ Resource Management: Proper async context management
• 📊 Complete Coverage: All documented NOAA CDO v2 endpoints supported

Installation

pip install noaa-cdo-api

API Documentation

Full API documentation with colored formatting is available at https://fxf8.github.io/noaa-cdo-api/.

Quick Start

import asyncio
from noaa_cdo_api import NOAAClient, Extent

async def main():
    # Best Practice: Use async context manager for automatic cleanup
    async with NOAAClient(token="YOUR_TOKEN_HERE") as client:
        # Query available datasets
        datasets = await client.get_datasets(limit=10)

        # Query stations in a geographic region
        stations = await client.get_stations(
            extent=Extent(40.0, -80.0, 45.0, -75.0), # latitude_min, longitude_min, latitude_max, longitude_max
            datasetid="GHCND",
            limit=5
        )

        # Get climate data with unit conversion
        data = await client.get_data(
            datasetid="GHCND",
            startdate="2022-01-01",
            enddate="2022-01-31",
            stationid="GHCND:USW00094728",
            units="metric",
            limit=100,
        )

if __name__ == "__main__":
    asyncio.run(main())

Important Implementation Notes

Event Loop Management

# ❌ BAD: Creating multiple event loops
client1 = await NOAAClient(token="TOKEN1")
client2 = await NOAAClient(token="TOKEN2")

results = [*asyncio.run(client1.get_datasets(...)), *asyncio.run(client2.get_datasets(...))]

# ✅ GOOD: Share the same event loop (note that rate limits apply **per token**)
async with NOAAClient(token="TOKEN1") as client1, \
         NOAAClient(token="TOKEN2") as client2:
    await asyncio.gather(
        client1.get_datasets(),
        client2.get_datasets()
    )

Resource Management

# ❌ Less ideal but functional: Manual cleanup
client = NOAAClient(token="TOKEN")
try:
    await client.get_datasets()
finally:
    client.close()  # Might miss resources (note: close does not need to be awaited)

# ✅ Better: Use async context manager
async with NOAAClient(token="TOKEN") as client:
    await client.get_datasets()

Rate Limiting

# ✅ Ideal
async def parallel_with():
    async with NOAAClient(token="TOKEN") as client:
        tasks = [client.get_datasets() for _ in range(20)]
        return await asyncio.gather(*tasks)  # Rate limits respected


# Works too since returns are asynchronous
async def parallel_separate():
    tasks = []
    for i in range(20):
        client = NOAAClient(token="TOKEN")  # Each has separate limiter
        tasks.append(client.get_datasets())
    return await asyncio.gather(*tasks)  # May exceed limits

Tips

1. Connection Pooling

   • Reuse the same client instance
   • Default connection limit is 10
   • Adjust with tcp_connector_limit parameter

2. Pagination

   • Use limit and offset for large result sets
   • Process data in chunks for memory efficiency

3. Data Volume

   • Limit date ranges (1 year for daily, 10 years for monthly)
   • Use specific station IDs when possible
   • Set includemetadata=False if not needed

4. Caching

   • Cache frequently accessed metadata
   • Implement local caching for historical data

Available Endpoints

• /datasets: Query available datasets
• /datacategories: Query data categories
• /datatypes: Query data types
• /locationcategories: Query location categories
• /locations: Query locations
• /stations: Query weather stations
• /data: Query actual climate data

Type Safety

The library provides comprehensive type checking through:

• TypedDict schemas for all parameters
• Runtime validation of parameter values
• Proper enum types for constrained fields

Example with type checking:

from noaa_cdo_api import parameter_schemas

params: parameter_schemas.StationsParameters = {
    "extent": "42.0,-90.0,40.0,-88.0",
    "datasetid": "GHCND",
    "limit": 100
}

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• NOAA's National Centers for Environmental Information (NCEI)
• The aiohttp team for their excellent HTTP client
• Contributors to the project

Getting Help

• Open an issue
• Read the docs
• NOAA CDO API Documentation