The port weather application provides local short-term weather forecast, which is very useful for safe sea transportation, especially during rescue operations contributing to the rescue of people at risk in the sea.

The application operates using continuous measurements from a meteorological station on a pilot vessel. The meteorological station records automatically all the measurements of its sensors.
A Machine Learning algorithm embedded on hardware provides wind speed forecast and warnings.


The Lincoln Platform uses the i-captain black box in order to collect the data from the weather station.

  • The data is collected online from an on board weather station.
  • The data is used as input to an advanced Machine Learning Algorithm, which provides Wind Speed prediction for 1-2 hours ahead.


The weather application forecasts the speed of the wind, minutes/hours ahead, based on previously recorded values and other measured parameters.

It utilizes a small memory footprint Machine Learning (ML) algorithm, which runs in a low power, low cpu frequency, low memory microcontroller.

It produces warnings reports and provides the forecasting values to the i-captain black box user.

The method has been successfully tested in real data from a tug boat trip from Chios island (Greece) to Athens (Greece).

Portweather Introduction

LINCOLN: Portweather tutorial 1 by CTI

Portweather tutorial 2: The LINCOLN IoT platform

Portweather tutorial 3: The Portweather application

Portweather IoT Platform

Portweather Application