Monday, January 9, 2012

Weatherswarm: Crowdsourced Weather Data

It's a seasonably cold but snowless morning in Vermont, and the computer models can't decide if the weather this Thursday will feature a warm rain, an ice storm, a blizzard, a mix of all three, or cold and dry conditions.  The computer models used to forecast weather are among the most powerful weather tools we have available to us, despite their fickle nature.  They aren't the only thing that computer-based technology brings to the weather forecasting table, however.  This week I've also been checking out a very different new weather tool, but one which could be the next big thing in weather data collection, or at least a vital aid to National Weather Service offices stricken by ridiculous funding cuts.

(above: the Middlebury River has plenty of ice, but it looks out of place without any snow.  The weather forecast is indecisive as to whether or not we'll pick some up this week)

WeatherSwarm is a beta (still in development) version of what could become a huge distributed system of realtime amateur weather reporters.  Think iNaturalist, but for weather.  The site includes a map that displays realtime reports of weather phenomena - drawn from Twitter or entered directly into the site.  Even more exciting, there is a mobile app in development that will allow weather to be reported directly from the field.  It's currently being tested (I'm helping!) but hopefully will be available to the general public soon.

Like all crowdsourced citizen science data, what this data may lack in precision it could potentially make up for in quantity and timeliness.  I don't see this as a replacement for trained weather spotters such as CoCoRaHS - but it could be an aid to these groups, and perhaps NWS would be interested in integrating this sort of technology into their reporting network.  (I hope some day i am settled in a location and financially secure enough to set up a web-linked weather station and become an 'official' spotter...)  With photographs and GPS data available through smartphones, many weather phenomena such as hail, blizzards, etc, could be reported by amateurs.  I could also see a tie-in with hydrological events such as floods, droughts, and ice jams.

As one of my friends pointed out recently, many of us are carrying around little censors that can collect a phenomenal amount of data (he pointed out that his smartphone has a BAROMETER!  Why does mine not?)  As computers become smaller and smaller, and the technology spreads, there is the potential for roving vehicles such as supertanker ships, big rigs and USPS trucks to also act as localized, precise sensors of weather conditions and other important data (I am not the one who came up with this idea, and I don't remember where I heard it, but I think it was someone from the CMU Human-Computer Interface Institute, which is doing some reeeeallllly neat stuff with this technology).  One of the reasons the computer models we use to forecast weather are at times unreliable is simply due to lack of data.  As recording devices become more powerful, we can feed more data into the models, which in turn will provide better forecasts.  At this point crowdsourced data isn't at the level it can be inputted into these models, but I think we will get there.

This is the future!  Isn't it much more awesome than sitting in air-traffic?

The main challenge to crowdsourced data gathering like this is that the quality is directly proportional to the number of people using the service.  In rural areas such as Vermont, there may not be very many people able or interested in participating, and in cities, where weather stations and reports are already abundant, the service is less necessary.  I'm hoping this takes off.  Give the website a try (be patient since it is new and still being developed) and when the app is ready for public use I'll post that info in this blog.  Meanwhile, go try iNaturalist if you haven't already.  I can't wait to get more plant distribution data on there to play with!


  1. Nice post. Very entertaining and full of information.

  2. You should join the Weathermob. On iPhone. Weather love.

  3. I'll check WeatherMob out. It looks interesting but doesn't seem to have a map like weatherswarm.

    I haven't really hard much from weatherswarm lately. anyone still using it?