Forecasting the Weather with Your Raspberry Pi and Bluemix

Share this article

Raspberry Pi and IBM Bluemix Adventures Part Three

IBM Bluemix and Raspberry Pi are a perfect combination when it comes to building a device with incredible potential. In this series, we’ve been exploring that potential and slowly building up what our Raspberry Pi can do. In the first article, we explored how to connect up a Raspberry Pi to IBM Bluemix, IBM Watson and Node-RED and in the second, we gave our Raspberry Pi a voice with IBM Watson’s text to speech. If you haven’t read those just yet, I’d recommend reading them and then heading back here! In this article, we’re going to give our Pi the ability to tell us the weather forecast for our area.

Raspberry Pi and IBM Bluemix Adventures Part Three
Artwork by SitePoint/PatCat, Logo rights: IBM and Raspberry Pi Foundation

This example can be completely standalone from the previous creation, but it will use the same nodes for the text to speech functionality — so you’ll want to keep those around!

Adding the Weather Company Data Service to Bluemix

In order to bring you some pretty valuable weather results, the guys at IBM literally purchased The Weather Company. This means you now have access to all of that worldwide data, forecasts and more! There’s a lot there. To add it into your IBM Bluemix set up, head back over to the Bluemix Services page and type in “weather” to filter your options and find the “Weather Company Data for IBM Bluemix” service:

Finding the weather service in Bluemix

The initial plan available for us to test with is a free plan. It comes with the following restrictions:

The Weather Company Data Free plan allows you to make a maximum of 10 calls to The Weather Company per minute, up to a maximum of 10,000 API calls, for each Bluemix Account.

In the options that pop up, leave your space as “dev” (if that’s the space you’ve been using in the other tutorials), leave the service unbound so you can reuse it in other apps, rename the service if you’d like (I left it as is), leave the credentials as is and click “Create”!

Creating our weather service in Bluemix

In order to be able to use this service from Node-RED (or anywhere really), we’ll need our credentials that prove we’re the ones asking for it. Head to the “Service Credentials” page and copy the username and password details somewhere safe:

Copying our weather service credentials

Accessing our Weather Data in Node-RED

Open up the same flow as we had from the previous tutorials (if you are starting from scratch, you’ll be able to follow along with most of these steps, just know you’ll need to go back to part 2 for details on how to do the text to speech bit).

We drag a new function node into our flow, I put it underneath the other one (you could create a whole new flow for this if you’d like):

Dragging in a new function node

Name that function “hourlyWeatherURL”, or whatever you’d like. Then put in the following JavaScript code:

msg.url = '{latitude}/{longitude}/forecast/hourly/48hour.json?units=m&language=en-US';
return msg;

You’ll notice there are two spots you need to customise there. Add in your own latitude and longitude into the {latitude} and {longitude} fields. If you aren’t sure of your city’s values, NASA has a lovely latitude and longitude finder. For my city, Sydney, it looked like so:

msg.url = '';
return msg;

There is also a units variable you can change if you’d prefer farenheit rather than celsius. units=m is for the metric system (celsius) and units=e is for the imperial system (farenheit). As we use the celsius here in Australia, I used units=m.

Once you’ve put all that in, click “Done”:

Creating our new function

Next up, we want to be able to make a HTTP request to the weather data service that uses this URL we set up in the first node. To do so, we drag the “http request” node in:

Dragging a new http request node

Then, we double click it to open up its settings and do the following:

  • Tick “Use basic authentication” — we’ll need that to include the username and password we copied earlier.
  • Add in the aforementioned username and password from earlier.
  • Select the “a parsed JSON object” as our return value — that way we’ll receive a JSON object that we can further filter through in following nodes.
  • Name it whatever you’d like, I chose “retrieveWeather”.
  • Click “Done”!
Setting up our http request node

In order to bring the URL from our first node, we’ve gotta remember to connect it to our “http request” node:

Connecting our http request module to our function

Then, drag in a “debug” node and connect it to the end. This will let us see logs of what data is coming through to check if it’s working:

Dragging in a debug node and connecting it up

Finally, we need a way of starting this whole process. To do that, drag in a new “inject” node and connect it to the start:

Adding an inject node to the start

Now we’re ready to test it out! Click “Deploy”:

Clicking deploy to test it out

In order to run our flow, we need to click the small square button on our “inject” node with the label “timestamp”. That will run our request for the weather and bring up the data in our debug tab:

Starting to run our flow using the debug node

Excellent! That should all be working as we’d hoped.

The msg.payload that is returned contains an array of forecasts for the upcoming hours. We take the first item in the array to get the current hour. The msg.payload looks a bit like so overall:

  "metadata": {
  "forecasts": [
      "class": "fod_long_range_hourly",
      "expire_time_gmt": 1472786395,
      "fcst_valid": 1472785200,
      "fcst_valid_local": "2016-09-02T13:00:00+1000",
      "num": 1,
      "day_ind": "D",
      "temp": 16,
      "dewpt": 14,
      "hi": 16,
      "wc": 16,
      "feels_like": 16,
      "icon_extd": 1200,
      "wxman": "wx2500",
      "icon_code": 12,
      "dow": "Friday",
      "phrase_12char": "Rain",
      "phrase_22char": "Rain",
      "phrase_32char": "Rain",
      "subphrase_pt1": "Rain",
      "subphrase_pt2": "",
      "subphrase_pt3": "",
      "pop": 93,
      "precip_type": "rain",
      "qpf": 0.96,
      "snow_qpf": 0.0,
      "rh": 92,
      "wspd": 8,
      "wdir": 17,
      "wdir_cardinal": "NNE",
      "gust": null,
      "clds": 100,
      "vis": 6.0,
      "mslp": 1006.8,
      "uv_index_raw": 0.83,
      "uv_index": 1,
      "uv_warning": 0,
      "uv_desc": "Low",
      "golf_index": 2,
      "golf_category": "Very Poor",
      "severity": 1
    ... (the next hours go here)

Now let’s move beyond the basic debug log and get our Pi to tell us this data through its previously learnt ability to speak!

Reminding our Pi How to Talk

To get our Pi to speak, just like it does with the CPU temperature from the previous article in the series, we just need to copy the last seven nodes from that tutorial. You can copy them by dragging and selecting them all, then pressing Ctrl+C (on Windows) or Cmd+C (on Mac):

Selecting and copying the speech nodes from our previous tutorial

Paste them using Ctrl+V (on Windows) or Cmd+V (on Mac), and place them neatly underneath our new weather flow. Rename the first function “Weather Text” and connect it to our “retrieveWeather” node so that it also takes the output from this node:

Renaming the node to Weather Text

Then double click the “Weather Text” function. Within this, we can set up what we would like our Pi to say. To start with, lets put this JavaScript code into it:

msg.payload = "The current temperature is " + 
              msg.payload.forecasts[0].temp +
              " degrees celsius";
return msg;

This will tell our Pi that we’d like the output to be structured to return “The current temperature is X degrees celsius” (if you’re using fahrenheit, feel free to change that here).

Now, you can skip this particular step if you’d like the Pi to continue telling you its CPU temperature, however I personally find the weather a bit more interesting and don’t want to hear about its CPU anymore. If you’re like me, double click the older “timestamp” node from the other flow. Take away the repeat setting and untick “Inject once at start?”. This way, that flow will only run if we click and tell it to run. Then click “Done”:

Take away the repeat and inject once of our other flow to stop it running

Now, we can set our weather flow to run automatically instead. Double click the new “timestamp” node and set it to repeat as frequently as you’d like (I chose every 30 minutes). Click “Inject once at start?” so that it runs automatically when you deploy the flow as well. Then click “Done” as usual:

Set our other node to repeat every 30 minutes

Deploy those changes and we’ll have an automatic notification of the current temperature nicely spoken to us!

Deploying our finished version

What Else Can We Do?

Wondering what other data is available from the service? You can find all of the possibilities that the weather API can provide on the Weather Company Data For IBM Bluemix APIs page. We are using the “48-Hour Hourly Forecast by Geocode” API in this example. To make our example a bit more advanced, let’s add data on whether it’ll rain in an hour like so (the next hour is just the second value in the array returned):

msg.payload = "The current temperature is " + 
              msg.payload.forecasts[0].temp +
              " degrees celsius. " +
              "There is a " +
              msg.payload.forecasts[1].pop + 
              " percent chance of " +
              msg.payload.forecasts[1].precip_type +
              " in an hour.";
return msg;

We can even go a step further and add info on whether the temperature (when you add humidity and wind) will actually feel like that temperature. The API has a field for that and we can access it like so:

var tempToSay = msg.payload.forecasts[0].temp != msg.payload.forecasts[0].feels_like ? 
        "The current temperature is " + msg.payload.forecasts[0].temp + 
        " degrees celsius, but it will feel more like " + 
        msg.payload.forecasts[0].feels_like + ". " : 
        "The current temperature is " + msg.payload.forecasts[0].temp + 
        " degrees celsius. ";

msg.payload = tempToSay +
              "There is a " +
              msg.payload.forecasts[1].pop + 
              " percent chance of " +
              msg.payload.forecasts[1].precip_type +
              " in an hour.";
return msg;

Basically, the code above only mentions the msg.payload.forecasts[0].feels_like value if it is different to the msg.payload.forecasts[0].temp value.


With that, we’ve got a Raspberry Pi that can tell us the weather forecast in all sorts of ways. There are a lot more bits of information in there from wind data to snow accumulation! There’s even a value representing the weather conditions for playing golf, which seems bizarrely specific. The power is now in your hands!

If you’re craving more functionality for your Raspberry Pi with Bluemix, never fear! There’s more to come in this series as we continue to explore more possibilities with the Bluemix platform. I’m quite enjoying the process and have grand aspirations here!

Have you built anything exciting with Bluemix and Raspberry Pi? Or do you have anything you’d love to make using Bluemix that you’d like us to cover? Leave a note in the comments!

Frequently Asked Questions (FAQs) about Forecasting Weather with IBM Watson, Bluemix, and Raspberry Pi

How does IBM Watson help in weather forecasting?

IBM Watson is a powerful AI tool that can analyze vast amounts of data and provide insights. In the context of weather forecasting, IBM Watson can process and analyze data from various sources such as satellites, weather stations, and sensors. It uses machine learning algorithms to predict weather patterns and conditions. This can be particularly useful for predicting severe weather events and helping communities prepare for them.

What is the role of Raspberry Pi in this weather forecasting setup?

Raspberry Pi acts as the hardware component in this setup. It can be connected to various weather sensors to collect data such as temperature, humidity, pressure, and wind speed. This data is then sent to IBM Watson for analysis and prediction. Raspberry Pi is a cost-effective and versatile tool that can be used to build a personal weather station.

How can I set up my own weather station using Raspberry Pi and IBM Watson?

Setting up your own weather station involves several steps. First, you need to connect your Raspberry Pi to the necessary weather sensors. Then, you need to set up IBM Watson and Bluemix to receive and analyze the data from these sensors. Detailed instructions for setting up the hardware and software components can be found in the article.

What kind of weather sensors can I use with Raspberry Pi?

Raspberry Pi is compatible with a wide range of weather sensors. Some of the most commonly used sensors include temperature and humidity sensors, pressure sensors, and wind speed sensors. The choice of sensors depends on what kind of weather data you want to collect.

How accurate are the weather predictions made by IBM Watson?

The accuracy of weather predictions made by IBM Watson depends on several factors, including the quality and quantity of data it receives. However, IBM Watson uses advanced machine learning algorithms and has access to a vast amount of data, which generally results in highly accurate predictions.

Can I use this setup to predict severe weather events?

Yes, this setup can be used to predict severe weather events. IBM Watson can analyze patterns in the data and predict potential severe weather events such as storms, hurricanes, and heatwaves. However, it’s important to note that while this system can provide valuable insights, it should not be the sole source of information for severe weather predictions.

How does IBM Watson’s weather forecasting compare to other methods?

IBM Watson’s weather forecasting has several advantages over traditional methods. It can process and analyze vast amounts of data from various sources, which can lead to more accurate predictions. Additionally, it can predict weather patterns and conditions in real-time, which can be particularly useful for predicting severe weather events.

Can I use other AI tools instead of IBM Watson for weather forecasting?

Yes, there are other AI tools that can be used for weather forecasting. However, IBM Watson is one of the most advanced and widely used tools in this field. It offers a range of features and capabilities that make it a popular choice for weather forecasting.

What is Bluemix and how does it fit into this setup?

Bluemix is IBM’s cloud platform that provides a range of services including AI, data analytics, and IoT. In this setup, Bluemix is used to host the IBM Watson service and to store and analyze the data collected by the Raspberry Pi.

Can I use this setup for other applications besides weather forecasting?

Yes, this setup can be used for a range of other applications. For example, it can be used for environmental monitoring, agricultural applications, and even home automation. The possibilities are virtually endless, thanks to the versatility of Raspberry Pi and the powerful capabilities of IBM Watson and Bluemix.

Patrick CatanzaritiPatrick Catanzariti
View Author

PatCat is the founder of Dev Diner, a site that explores developing for emerging tech such as virtual and augmented reality, the Internet of Things, artificial intelligence and wearables. He is a SitePoint contributing editor for emerging tech, an instructor at SitePoint Premium and O'Reilly, a Meta Pioneer and freelance developer who loves every opportunity to tinker with something new in a tech demo.

bluemixibm watsonnode-redpatrickcRaspberry Pi
Share this article
Read Next
Get the freshest news and resources for developers, designers and digital creators in your inbox each week
Loading form