Evaluating Loss Functions and Learning Data Pre-Processing for Climate Downscaling Deep Learning Models

Xingying Huang

2023-06-20

Abstract: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.

Machine Learning,Atmospheric and Oceanic Physics

What problem does this paper attempt to address?

The paper aims to address the challenges faced when applying deep learning models to climate downscaling tasks, particularly focusing on the significant differences between climate data and image data. Specifically, the paper focuses on the following aspects:

1. **Evaluating Loss Functions**: The study explores the performance of different loss functions (such as L1 loss and L2 loss) in climate downscaling tasks and analyzes the impact of these loss functions on different types of climate data (such as precipitation data and temperature data).

2. **Nonlinear Data Preprocessing Methods**: Given the distribution characteristics of climate data differ from image data, the paper investigates several nonlinear data preprocessing techniques (e.g., gamma correction) and evaluates whether these techniques help improve model performance.

3. **Automatically Learning Nonlinear Preprocessing Functions**: A learnable nonlinear preprocessing scheme is proposed, enabling the model to automatically find the optimal data preprocessing method to adapt to different types of climate data.

Through the above research, the main goal of the paper is to identify the best practice methods suitable for climate downscaling tasks, particularly in selecting appropriate loss functions and data preprocessing techniques, thereby enhancing the application effectiveness of deep learning models in this field.