HYPERSPECTRAL IMAGE CLASSIFICATION USING RESIDUAL 2D AND 3D CONVOLUTIONAL NEURAL NETWORK JOINT ATTENTION MODEL

Climate models are essential for understanding future climate change, but their outputs are often too coarse and inaccurate for local use. We developed a simple, transparent method that first corrects model errors and then adds realistic spatial detail using machine learning. Tested on global climate simulations, the method improved rainfall patterns while keeping overall totals realistic, making the climate data more useful for research and planning.