Michael Ferrari

We try to utilise as much data as possible.

This is not your father’s weather forecast.

Weather presenters are basically presenting information from basic weather models. After a few days the weather information is basically useless.

When you’re trying to base financial decisions on short term weather forecast, it’s very [bad/difficult].

We offer a completely different approach.

The world is warming, it’s a trend so won’t be constant and everywhere. Using traditional methods you can’t predict this properly.

We can do weather forecasts up to 10 months out using ~400 sensors around the world.

We can’t predict everything. Down to the daily level we can produce a granular forecast. We use multiple sensor networks.

Usually seasonal forecast that your tax dollars pay for say “there’s equal chances of everything happening”. We offer a more granular level showing, e.g. daily temperature changes. A lot of this information is supplemented by the sensor datas constantly.

Last year there was a landmark paper published Craig Venter recreated the darwin journey and sampled sea water at regular locations across the globe. We put the information into a geo-spatial context.

In evolutionary biology there’s always been a saying that “everything is everywhere”, took samples from around the globe. the thought before the study was that genes would be similar, but study showed that repeatability of some genomes was very specific. Caribbean vs Indian Ocean for instance.

Paradigm shift for future sampling studies. We now have a great dataset to base this stuff on.

Some of this will be addressed in further talks. We’re not at the point where we can prevent weather disasters but with realtime monitoring we can plan and react better.

Global Weather Visualization: Utilizing Sensor Networks to Monetize Realtime Data

Technorati tags: weather, sensors, where, where2.0, where2008