Frequenz Weather API Client

Introduction

Supported Platforms

Contributing

Usage

Installation

pip install frequenz-client-weather

Available Features

Initialize the client

from frequenz.client.weather import Client
from frequenz.client.base.channel import ChannelOptions, KeepAliveOptions, SslOptions
from datetime import timedelta
client = Client(
 service_address,
 channel_defaults=ChannelOptions(
 ssl=SslOptions(
 enabled=False,
 ),
 keep_alive=KeepAliveOptions(
 enabled=True,
 timeout=timedelta(minutes=5),
 interval=timedelta(seconds=20),
 ),
 ),
)

Get historical weather forecast

from datetime import datetime
import pandas as pd
from frequenz.client.weather._types import ForecastFeature, Location
# Define a list of locations, features and a time range to request historical forecasts for
locations = [Location(latitude=46.2276, longitude=15.2137, country_code="DE")]
features = [ForecastFeature.TEMPERATURE_2_METRE, ForecastFeature.V_WIND_COMPONENT_10_METRE]
start = datetime(2024, 1, 1)
end = datetime(2024, 1, 31)
forecast_iterator = await client.stream_historical_forecast(
 features=features, locations=locations, start=start, end=end
)
# Collect and flatten forecasts
flat_forecasts = [f.flatten() async for f in forecast_iterator]
forecast_records = [record for batch in flat_forecasts for record in batch]
# E.g. convert to DataFrame and sort
forecast_df = pd.DataFrame(forecast_records).sort_values(["create_time", "valid_time", "latitude", "longitude"])
print(forecast_df)

Get live weather forecast

import pandas as pd
from frequenz.client.weather._types import ForecastFeature, Location
# Define a list of locations and features to request live forecasts for
locations = [Location(latitude=46.2276, longitude=15.2137, country_code="DE")]
features = [ForecastFeature.TEMPERATURE_2_METRE, ForecastFeature.V_WIND_COMPONENT_10_METRE]
# Returns a Receiver object that can be iterated over asynchronously
stream = await client.stream_live_forecast(
 locations=locations,
 features=features,
)
# Process incoming forecasts as they arrive
async for forecast in stream:
 # The to_ndarray_vlf method converts the forecast data to a 3D numpy array,
 # where the dimensions correspond to validity_ts, location, feature
 # The method can also take filters for validity_ts, locations and features
 # E.g. filter the forecast for wind features
 wind_forecast = forecast.to_ndarray_vlf(features=[ForecastFeature.V_WIND_COMPONENT_10_METRE])
 print(wind_forecast)

Command Line Interface

Get historical weather forecast

weather-cli \
 --url <service-address> \
 --location "40,15" \
 --feature U_WIND_COMPONENT_100_METRE \
 --start 2024年03月14日 \
 --end 2024年03月15日 \
 --mode historical

Get live weather forecast

weather-cli \
 --url <service-address> \
 --location "40, 15" \
 --feature TEMPERATURE_2_METRE \
 --mode live