Optical Data Fusion for Daily Gap-Free Surface Reflectance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing satellite data fusion methods struggle to achieve high spatial resolution and high revisiting frequency simultaneously due to cloud contamination and sensor mechanical issues, leading to data gaps and inaccurate predictions.
Innovation Solution
The STAIR method partitions satellite images into homogeneous segments, adaptively corrects missing pixels using an adaptive-average correction process, and integrates multiple sources of optical data to generate daily, high-resolution, cloud-/gap-free surface reflectance products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-/medium-resolution satellite data (e.g., Landsat, Sentinel-2) are used, then spatial resolution is improved, but temporal sampling frequency deteriorates (multiple days to weeks)
Solution Approach 1:
The patent combines multiple satellite data sources (Landsat, Sentinel-2, MODIS) with different spatial and temporal characteristics into a unified fusion product. By merging these data sources, the system achieves both high spatial resolution (from Landsat/Sentinel-2) and high temporal frequency (from MODIS daily observations), resolving the contradiction between spatial and temporal sampling rates.
Solution Approach 2:
The patent employs dynamic temporal fusion that adapts the contribution of different data sources based on temporal proximity to the target date. The system dynamically selects and weights satellite images based on their temporal distance from the target date, allowing flexible adjustment of temporal sampling frequency while maintaining spatial resolution.
2Measurement precision
If clear Landsat-MODIS image pairs are required for accurate fusion, then prediction accuracy is improved, but data availability deteriorates due to cloud contamination
Solution Approach 1:
The patent applies adaptive weighting that allows partial use of cloud-contaminated images by assigning lower weights to affected pixels or regions. Instead of requiring complete clarity, the system uses available clear pixels and partially processes contaminated areas with reduced confidence, maintaining data availability while preserving accuracy where possible.
Solution Approach 2:
The patent converts cloud-contaminated data into useful information by using the temporal proximity of these images. Even though cloud-contaminated images have lower quality, their temporal closeness to the target date provides valuable spatiotemporal information that can be weighted appropriately to improve overall fusion accuracy and data availability.
3Reliability
If Landsat-MODIS image pairs months apart from target date are used, then data availability is improved, but prediction accuracy deteriorates
Solution Approach 1:
The patent implements dynamic temporal weighting where the contribution of each satellite image to the fusion result is determined by its temporal distance from the target date. Images closer in time receive higher weights, while older images receive lower weights, creating a dynamic balance between data availability and prediction accuracy based on temporal proximity.
Solution Approach 2:
The patent changes the weighting parameter based on temporal distance, where the weight assigned to each image pair is a function of its temporal proximity to the target date. This parameter change allows the system to automatically adjust the influence of older versus newer images, optimizing the trade-off between data availability and prediction accuracy.
4Reliability
If multiple satellite data sources are integrated, then data completeness is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational process into distinct stages: data acquisition from multiple sources, temporal matching and pairing, quality assessment and weighting, and final fusion computation. This segmentation allows each stage to be optimized independently and facilitates parallel processing, reducing overall computational complexity while maintaining data completeness.
Solution Approach 2:
The patent applies local quality assessment where different regions of the image are evaluated independently for cloud contamination and data quality. This allows the system to process only the necessary portions of each satellite image at full resolution, reducing computational load while maintaining completeness in areas with available data.
Data Source
AI summary
Aspects of the subject disclosure may include, for example, performing, by a processing system, image fusion using two or more groups of images to generate predicted images, wherein each group of the two or more groups has one of a different resolution, a different frequency temporal pattern or a combination thereof than another of the two or more groups. Gap filling can be performed by the processing system to correct images of the two or more groups. Additional embodiments are disclosed.