Evaluating Loss Functions and Learning Data Pre-Processing for Climate Downscaling Deep Learning Models
This work addresses the problem of adapting computer vision models to climate science for researchers, but it is incremental as it focuses on specific optimizations rather than a new paradigm.
The study tackled the challenge of applying deep learning to climate downscaling by evaluating loss functions and learning non-linear data pre-processing, finding that L2 loss outperforms L1 loss for imbalanced data like precipitation and that an automated pre-processing method improves model accuracy.
Deep learning models have gained popularity in climate science, following their success in computer vision and other domains. For instance, researchers are increasingly employing deep learning techniques for downscaling climate data, drawing inspiration from image super-resolution models. However, there are notable differences between image data and climate data. While image data typically falls within a specific range (e.g., [0, 255]) and exhibits a relatively uniform or normal distribution, climate data can possess arbitrary value ranges and highly uneven distributions, such as precipitation data. This non-uniform distribution presents challenges when attempting to directly apply existing computer vision models to climate science tasks. Few studies have addressed this issue thus far. In this study, we explore the effects of loss functions and non-linear data pre-processing methods for deep learning models in the context of climate downscaling. We employ a climate downscaling experiment as an example to evaluate these factors. Our findings reveal that L1 loss and L2 loss perform similarly on some more balanced data like temperature data while for some imbalanced data like precipitation data, L2 loss performs significantly better than L1 loss. Additionally, we propose an approach to automatically learn the non-linear pre-processing function, which further enhances model accuracy and achieves the best results.