A novel framework to assess all-round performances of spatiotemporal fusion models

• DOI: 10.1016/j.rse.2022.113002
• Publication Date: 2022-03-23T07:28:36Z
• Authors: Xiaolin Zhu; Wenfeng Zhan; Junxiong Zhou; Xuehong Chen; Zicong Liang; Shuai Xu; Jin Chen
• Keywords: Sensor Fusion

Spatiotemporal data fusion, as a feasible and low-cost solution for producing time-series satellite images with both high spatial temporal resolution, has undergone rapid development over the past two decades more than one hundred spatiotemporal fusion methods developed. Accuracy assessment of fused is crucial users to select appropriate real-world applications. However, commonly used metrics do not comprehensively cover multiple aspects image quality, contain redundant information, are comparable across different study areas. To address these problems, this proposed novel framework assess all-round performances methods. Four accuracy metrics, including RMSE, AD, Edge, local binary patterns (LBP), were selected optimal set according criteria. These only quantify spectral information in but also greatly alleviate redundancy feature computational simplicity. Furthermore, inspired by Taylor diagrams, we designed an performance (APA) diagram provide visual tool comprehensive methods, supporting cross-comparison considering effects input land surface characteristics. The case three typical sites demonstrated that can better differentiate six This new promote guide suitable applications, well facilitate establishment standard procedure