Landbench 1.0: A Benchmark Dataset and Evaluation Metrics for Data-Driven Land Surface Variables Prediction

Abstract

The advancements in deep learning methods have presented new opportunities and challenges for predicting land surface variables (LSVs) due to their similarity with computer sciences tasks. However, the lack of a benchmark dataset hampers fair comparisons of different data-driven deep learning models for LSVs predictions. Hence, we propose a LSVs benchmark dataset and prediction toolbox to boost research in data-driven LSVs modeling and improve the consistency of data-driven deep learning models for LSVs. LSVs benchmark dataset contains a large number of hydrology-related variables, such as soil moisture, runoff, etc., which can verify the simulation of hydrological processes. Various global data from ERA5-land, ERA5 reanalysis, SoilGrid, SMSC, and MODIS datasets have been pre-processed into daily data at 0.5-, 1-, 2-, and 4-degree resolutions to facilitate their use in data-driven models. Simple statistical metrics, i.e., the root mean squared error and correlation coefficient, are chosen to evaluate the performance of different DL models, including convolutional neural network, long short-term memory and convolution long short-term memory models, with lead times of 1 and 5 days. A processed-based model serves as a physic baseline, and soil moisture and surface sensible heat fluxes are taken as the target variables. The developed LandBench toolbox with Pytorch can facilitate the reimplementation of existing methods, the development of novel predictive models, and the use of unified evaluation metrics. The toolbox also includes address mapping technology for high-resolution global predictions with limited computing resources. We hope LandBench will promote collaboration between computer scientists and Earth system scientists to develop effective predictive models for LSVs with ease. The DOI link for the dataset is https://doi.org/10.11888/Atmos.tpdc.300294 (Li et al. 2023), and the toolbox can be accessed through https://github.com/2023ATAI/Landbench1.0 .