This project implements a Self-Organizing Map (SOM) algorithm to analyze weather patterns over New Zealand. The SOM identifies spatial patterns in the input data and groups them into clusters.
Self-organising maps (SOM), also known as Kohonen maps are an artificial neural network, that transforms complex, nonlinear high-dimensional data into simpler forms. This project implements a Self-Organizing Map (SOM) algorithm to analyze weather patterns over New Zealand. The SOM identifies spatial patterns in the input data and performs unsupervised clustering of data This code preprocesses MSLP data, trains the SOM, and visualizes the resulting patterns. It serves as a tool for understanding the mainly occurring synoptic states.
This code uses the MINISOM Python package. MiniSom is a minimalistic implementation of Self-Organizing Maps (SOM) built using NumPy. Here we specifically use the weather@home model simulations to create SOMs of commonly occurring synoptic states in the New Zealand domain.