Met Office YCN Brief

For this pro­ject I set out to ex­plore dif­fer­ent modes of data vi­su­al­i­sa­tion and sci­ence com­mu­ni­ca­tion in the con­text of weather data. The UK Met Office makes the ma­jor­ity of its data pub­licly avail­able, in­clud­ing cur­rent weather ob­ser­va­tions, weather fore­casts and his­tor­i­cal records reach­ing back hun­dreds of years.

Initially I worked on a chat­bot that could serve as an in­tu­itive in­ter­face to the Met Office’s datasets. A chat­bot is a nat­ural-lan­guage in­ter­face that ex­ists within ex­ist­ing mes­sag­ing ap­pli­ca­tions such as Whatsapp, SMS or Facebook Messenger. I pro­pose this as a more per­sonal, light­weight al­ter­na­tive to tra­di­tional weather ap­pli­ca­tions.

The ro­bot is built on the Microsoft Bot Framework, which uses ma­chine learn­ing to un­der­stand nat­ural lan­guage. Location-based in­puts such as How’s the weather in Manchester” are parsed us­ing the Google Maps API be­fore the ap­pro­pri­ate weather in­for­ma­tion is re­quested from the Met Office API. The re­sult­ing data is then con­verted back into nat­ural lan­guage and pre­sented to the user. The source code is avail­able on Github.

Next, I fo­cused on a dataset con­tain­ing his­tor­i­cal weather ob­ser­va­tions go­ing back as far as 1853. I made a se­ries of at­tempts to vi­su­alise the roughly 50.000 data points it con­tains.

Eventually I found a suc­cess­ful way to show av­er­age tem­per­a­tures as colours on a spec­trum. This was the first ap­proach I dis­cov­ered that made global warm­ing eas­ily vis­i­ble.

All vi­su­al­i­sa­tions were built us­ing d3.js. The source code is avail­able on Github.

I have some thoughts on de­sign com­pe­ti­tions.