Short-term local meteorological prediction through search for analogues

The short term local weather forecast is investigated through related conditions for the meteorological station of the National Technical University of Athens, which lies at the base of Mount Hymettus. The forecast was for four meteorological variables: temperature, humidity, wind speed and rainfall. The time steps tested were: ten minutes, one hour and one day, with a forecast horizon of up to ten time steps. Two models were developed where the first was considering each meteorological variable separately while the second combines the four meteorological variables for parallel export prognosis. The investigation was done by creating a program that was aimed at the implementation and control of the process and the potential (in the future), disclosure of a website. In conclusion, the multivariate model did not give any advantage to the results so eventually the simple model qualified. The method provides a satisfying forecast for the meteorological variables of temperature, humidity and wind speed for the entire forecasting horizon for the time steps of ten minutes and one hour, while for the time step of a day results of up to a fourth day are considered satisfactory. Finally, the method is considered not suitable for predicting rainfall.