pyobs-weather¶

pyobs-weather is an aggregator for data from several weather stations. Rules can be defined for when weather is defined to be “good”. It provides both a web frontend and an API for access.

Installation¶

While it is definitely possible to run pyobs-weather without Docker, we highly recommend it for its simplicity.

First, build the image:

cd https://github.com/pyobs/pyobs-weather.git
cd pyobs-weather
docker build . -t pyobs-weather

pyobs-weather requires a database for storing its data, redis for task brokering, celery for providing those tasks, and nginx for serving static files. Easiest way to deploy everything is using docker-compose.

A typical docker-compose.yml looks like this:

version: '3'

services:
  db:
    image: postgres:11
    volumes:
      - pgdata:/var/lib/postgresql/data
    restart: always

  weather:
    image: pyobs-weather
    volumes:
      - ./local_settings.py:/archive/pyobs_archive/local_settings.py
      - static:/weather/static
    depends_on:
      - db
    restart: always
    command: bash -c "python manage.py collectstatic --no-input && python manage.py makemigrations && python manage.py migrate && gunicorn --workers=3 pyobs_weather.wsgi -b 0.0.0.0:8000"

  redis:
    image: redis
    restart: always

  celery:
    image: pyobs-weather
    command: celery -A pyobs_weather worker --beat --scheduler django --loglevel=info
    depends_on:
      - db
      - redis
    restart: always

  nginx:
    image: nginx
    volumes:
      - ./nginx.conf:/etc/nginx/conf.d/default.conf
      - static:/static/static
    ports:
      - 8002:80
    restart: always

volumes:
  pgdata:
  static:

In this example, nginx needs a configuration file nginx.conf in the same directory, which might look like this:

server {
    listen 80;
    server_name  127.0.0.1;

    location / {
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $host;
        proxy_redirect off;
        if (!-f $request_filename) {
            proxy_pass http://weather:8000;
            break;
        }
    }

    location /static/ {
        root /static;
    }
}

And pyobs-weather itself needs a configuration file called local_settings.py. Here is the file for MONET/S as an example:

# disable debug
DEBUG = False

# we're reverse proxying, so only localhost is allowed to access
ALLOWED_HOSTS = ['localhost']

# weather settings
OBSERVER_NAME = 'MONET/S @ SAAO'
OBSERVER_LOCATION = {'longitude': 20.810808, 'latitude': -32.375823, 'elevation': 1798.}
WINDOW_TITLE = 'Weather at ' + OBSERVER_NAME

With all three files in one directory, you can easily do:

docker-compose up -d

and the whole system should be up and running within a minute.

Finally, you need to get into the container and init the database (name of container may vary):

docker exec -it weather_weather_1 bash
./manage.py initweather

While at it, you can also create a super user:

./manage.py createsuperuser

The web frontend should now be accessible via web browser at http://localhost:8002/ and the admin panel at http://localhost:8002/admin.

Configuration¶

Setting pyobs-weather up mainly consists of adding weather stations and evaluators for their sensors in the admin panel. Three stations have automatically been added by the initweather weather script during the installation:

Current: Always returns the averaged latest values from all weather stations.
Average: Returns 5-minutes averages for all values
Observer: An “astronomical observer” that mainly calculates the position of the sun.

Adding a station¶

When adding a new station, one has to provide a unique code for it and a name. Then a Python class must be given and optionally JSON encoded arguments that are passed to its constructor. Finally, for regular polling of the station a crontab or an interval must be specified.

See a list of available weather station classes in the Stations below. New station types can easily be added by creating new Python classes.

Adding evaluators¶

Adding a new station automatically adds its sensors to the system. In order to device, whether the current value of a sensors is good or bad, evaluators need to be attached to them. For a list of available evaluators see Evaluators. New evaluators can simply be added by creating new Python classes.

For adding an evaluator, you just click on a sensor and select the evaluator from the list of evaluators.

API Reference¶

Sensor Types¶

pyobs-weather assumes that every weather station returns only one value for a given type of sensor, which makes it easier to combine values from different stations. For instance, if multiple stations give a value for “humid”, the system can easily combine them to create an average humidity.

Sensor types are defined within the Python classes for the stations, and can be extended by new stations easily. The current list of sensor types is:

temp: Temperature [°C]
humid: Relative humidity [%]
press: Pressure [hPa]
winddir: Wind direction [°E of N’]
windspeed: Wind speed [km/h’]
particles: Particle count [ppcm]
rain: Raining [0/1]
skytemp: Relative sky temperature [°C]
sunalt: Solar altitude [°]

Stations¶

A station in pyobs-weather is a single weather station, from which it polls data.

List of weather station classes

Evaluators¶

Evaluators take sensors and evaluate their values.

List of evaluators

REST API Reference¶

All API endpoints return JSON data.

Current weather¶

The /api/current/ endpoint returns the current values for all sensor types, as shown on the web page in the left column:

$ http https://weather.monet.saao.ac.za/api/current/

{
    "good": false,
    "sensors": {
        "dewpoint": {"good": null, "value": -5.22835},
        "humid": {"good": true, "value": 15.754024},
        "press": {"good": null, "value": 824.888119047619},
        "rain": {"good": true, "value": 0.0},
        "skytemp": {"good": false, "value": -19.43},
        "sunalt": {"good": false, "value": 71.0194850756101},
        "temp": {"good": null, "value": 25.7139337142857},
        "winddir": {"good": null, "value": 298.59207},
        "windspeed": {"good": true, "value": 32.2115857142857}
    },
    "time": "2020-02-13T10:44:29.302Z"
}

Single station/sensor¶

A list of weather stations can be retrieved via /api/stations/:

$ http https://weather.monet.saao.ac.za/api/stations/

[
    {"code": "average", "name": "Average values"},
    {"code": "current", "name": "Current values"},
    {"code": "observer", "name": "Observer"},
    {"code": "lco", "name": "LCO"},
    {"code": "salt", "name": "SALT"},
    {"code": "suth", "name": "Sutherland Weather"},
    {"code": "monet_cur", "name": "MONET current"},
    {"code": "monet", "name": "MONET"}
]

Using one of the codes returned by /api/stations/, all its respective sensor values can be requested from /api/stations/<station>/:

$ http https://weather.monet.saao.ac.za/api/stations/monet/

{
    "code": "monet",
    "name": "MONET",
    "sensors": [
        {
            "code": "temp",
            "name": "Temperature",
            "time": "2020-02-13T10:45:00Z",
            "value": 25.3448275862069
        },
        {
            "code": "press",
            "name": "Pressure",
            "time": "2020-02-13T10:45:00Z",
            "value": 824.786206896552
        },
        {
            "code": "winddir",
            "name": "Wind direction",
            "time": "2020-02-13T10:45:00Z",
            "value": 285.51724137931
        },
        {
            "code": "humid",
            "name": "Relative humidity",
            "time": "2020-02-13T10:45:00Z",
            "value": 16.8931034482759
        },
        {
            "code": "rain",
            "name": "Raining",
            "time": "2020-02-13T10:45:00Z",
            "value": 0.0
        },
        {
            "code": "windspeed",
            "name": "Wind speed",
            "time": "2020-02-13T10:45:00Z",
            "value": 35.0193103448276
        }
    ]
}

Which also works just for single values using the code of a sensor at /api/stations/<station>/<sensor>/:

$ http https://weather.monet.saao.ac.za/api/stations/monet/windspeed/

{
    "code": "windspeed",
    "good": true,
    "name": "Wind speed",
    "since": "2020-02-05T19:00:06.769Z",
    "time": "2020-02-13T10:50:00Z",
    "unit": "km/h",
    "value": 36.2358620689655
}

History¶

A list of sensor types, for which historic data is available, can be fetched using the /api/history/ endpoint:

$ http https://weather.monet.saao.ac.za/api/history/

[
    "humid",
    "skytemp",
    "dewpoint",
    "windspeed",
    "rain",
    "press",
    "temp",
    "winddir"
]

Finally, the actual history for one of these sensor types (for all stations) is available via /api/history/<type>/:

$ http https://weather.monet.saao.ac.za/api/history/temp/

[
    {
        "code": "average",
        "color": "rgba(0, 0, 0, 0.1)",
        "data": [
            {"time": "2020-02-13T10:50:00.011Z", "value": 25.7011098488768},
            {"time": "2020-02-13T10:45:00.012Z", "value": 25.6358858386895},
            [...]
            {"time": "2020-02-12T10:55:00.015Z","value": 28.482568678953}
        ],
        "name": "Average values"
    },
    {
        "code": "lco",
        "color": "rgba(0, 0, 0, 0.1)",
        "data": [
            {"time": "2020-02-13T10:54:01Z", "value": 27.0},
            {"time": "2020-02-13T10:53:00Z", "value": 26.9},
            [...]
            {"time": "2020-02-12T10:55:00Z", "value": 28.2758620689655}
        ],
        "name": "MONET"
    }
    [...]
]