An approach to fill in missing data from satellite imagery using data-intensive computing and DINEOF.

Area of Science:

• Remote Sensing
• Data-Intensive Computing
• Oceanography

Background:

• Satellite imagery often contains missing data due to acquisition issues like occlusion and sunlight.
• Processing large, high-resolution satellite images can lead to memory overflow errors.

Purpose of the Study:

• To propose and validate a data-intensive computing approach for filling missing satellite image data.
• To enhance the accuracy of chlorophyll concentration estimation in oceanic regions.

Main Methods:

• Automated downloading and correction of satellite data from multiple sensors (MODIS-TERRA, MODIS-AQUA, VIIRS-SNPP, VIIRS-JPSS-1).
• Merging corrected data into an orthomosaic, splitting it into segments for data infilling, and reassembling the segments.
• Utilizing data-intensive computing techniques on a cluster architecture to handle large datasets.

Main Results:

• The proposed approach successfully merges data from diverse satellite sensors to create a more complete orthomosaic.
• The method effectively fills missing data, producing results comparable to state-of-the-art techniques for chlorophyll concentration estimation.
• The approach avoids memory overflow issues common with large, high-resolution satellite images.

Conclusions:

• The data-intensive approach offers a robust solution for handling missing data in satellite imagery.
• This method provides a more accurate estimation of chlorophyll concentration compared to traditional methods like the mean of pixels.
• The technique is scalable and efficient for processing large volumes of satellite data.