Weather is a chaotic physical system that is very hard (computationally and in accuracy) to calculate. Because of its complexity we choose the Dead Sea as a relatively isolated environment as a simple case for this complicated problem.
By using the measurements from a meteorological station of the Geological Survey, and the past ECMWF forecasts for the area as our dataset, we tried to use the Transformers (ML) architecture to predict in a better accuracy the winds (directions and amplitude) for a specific station.
Transformers originally are used in the NLP field, therefore to adapt the architecture to continues time series we had to change the ‘Classic’ structure for both encoder and decoder subunits in the Transformer.
In addition to achieve the best learning process we had to learn some meteorological attributes of weather measuring and estimation process and to adjust the model structure, loss function and more.
Finally, we will compare our performance to the real results, and suggest future research directions.