Home

Sataako?

The Sataako? -service issues warnings to users of when it will rain at their location within the next hour in the Helsinki metropolitan area. The service will use optic flow to interpolate the movement of precipitation systems during the next 60 minutes based on data gathered through FMI open data.

Data

Our service uses the Finnish Meteorological Institute (FMI) open data, namely radar data. Prior to data-analysis, the data needs to be converted to numpy arrays and cropped. We have limited access to data, so it can't be fetched on every request and must be cached.

Data analysis

To be able to give a rain forecast (nowcast), we need to do time extrapolation of radar rainrate fields. This is done using an optic flow algorithm (http://scholarpedia.org/article/Optic_flow). In addition, we also need to do some smoothing of the extrapolated rainrate fields. Given the user coordinates, we analyze if precipitation systems will move above the user location. From the extrapolated precipitation field we calculate the precipitation accumulation in millimeters in the user location. We use this information to issue warnings to the user.

Communication of results

The warnings will be issued as Telegram messages, using Telegram bot API. The user will be informed if there is a risk of rain within the next 60 minutes. The warnings also give information about precipitation accumulation. User can view images/video of weather maps in the metropolitan area and forecasts of rain movement. Operationalization The users get information rainfall only if there is a risk of rain. A use case would be a commuter thinking if they should go by bike or take the bus. If there is no warnings then they can safely choose the bike option.