Weather data is out there, and there are tons of ways to access it, but ultimately there seems to be a large polarization between 'ease of access' and 'data intensiveness'. Weather tools that err on the side of accessibility seem to display only the barebones measurements, whereas data intensive sites seem to just be outputting unintuitive and unappealing tables from an SQL database call.
seeks to address by offering a sensible middle ground, a quick and easy to use weather lookup that encompasses past, present, and future, for any location the user pleases in the US and associated territories.
What it does
Everything mentioned before, plus the ability to send a text message containing the upcoming weekly forecast temperature and characterization data. Example picture of automated text message below.
How we built it
Node.JS server hosted on IBM Bluemix, historical database from NOAA requested and approved, sent to us over email. Processed, cleaned, filtered, and analyzed in Spark for Bluemix, and then imported into an SQL Database hosted on Bluemix. Historical data displayed using AMcharts JS library (https://www.amcharts.com/). Current and 10day forecast supplied through the Weather Insights API module in Bluemix, called directly to our back end. Front end then calls the data from the back end in order to hide the API credentials.
Challenges we ran into
- There were a few bugs with the Spark Kernel in IBM Bluemix, it would be busy while no jobs were processing from time to time.
- In Bluemix's Spark module, the same code, executed at two different points in time, would sometimes return an error and sometimes not (without the potential of broken API calls etc).
Accomplishments that we're proud of
What we learned
Weather is something that is tantamount to all of our lives, and controls for about 1/3 of the world's GDP. By making data more easily accesible to even the average person, it can account for national cost savings or reduction in wasted energy/resources.