Remote sensing and big data: Google Earth Engine data to assist calibration processes in hydro-sediment modeling on large scales

• Published in: Remote sensing applications Vol. 36
• Main Authors: Rossoni, Renata Barão, Laipelt, Leonardo, Paiva, Rodrigo Cauduro Dias de, Fan, Fernando Mainardi
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2024
• Subjects: Big data processing; Google Earth Engine; Remote sensing; Sediment modeling; Sediment transport; South America; South America; Big data processing; Google Earth Engine; Sediment modeling; Remote sensing; Sediment transport
• ISSN: 2352-9385; 2352-9385
• DOI: 10.1016/j.rsase.2024.101352

Mathematical modeling aids in understanding large-scale erosion and sedimentation. However, sediment transport models calibration is constrained by data scarcity. This study explores the use of remote sensing (RS) imagery to supplement observed data, addressing three key questions: (1) How can high-resolution RS data be obtained using cloud-based methods for hydro-sediment applications, considering river changes? (2) What are the benefits of RS data in data-scarce conditions? (3) How can RS data improve hydro-sediment modeling in data-deficient regions? We developed a method to acquire large-scale RS data using Google Earth Engine (GEE) to obtain red and infrared reflectance from satellite imagery. After filtering errors, the data were used to calibrate a hydro-sediment model. Results showed that RS data, when combined with observed data, provided similar outcomes but performed better for lower values. Calibration with RS data alone improved the Kling-Gupta Efficiency (KGE) by 5%–18% and correlation by 5%–15%. Key conclusions are: (I) Cloud-based calibration is superior to using limited virtual stations; (II) RS data effectively complements observed data in hydro-sediment modeling; (III) Calibration using only RS data is beneficial in ungauged basins and preferable to no calibration. [Display omitted] •Remote sensing data addresses gaps in hydro-sediment model calibration.•GEE enables high-resolution data acquisition for large-scale sediment modeling.•1,237 virtual gauge stations created for better spatial data distribution.•Combining remote sensing with observed data enhances model performance.•Valuable for ungauged basins, improving sediment dynamics representation.