Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List

Full text release has been delayed at the author's request until December 21, 2025

ETD Abstract Container

Abstract Header

Improving Satellite Data Quality and Availability: A Deep Learning Approach

Mukherjee, Rohit
http://rave.ohiolink.edu/etdc/view?acc_num=osu1598286115582789

Abstract Details

2020, Doctor of Philosophy, Ohio State University, Geography.
Remote Sensing offers a unique perspective of our Earth and is crucial for managing its resources. Currently, there is no single satellite data product that is suitable for all applications. Satellite data are limited by their spatial, spectral, and temporal resolution. Additionally, satellite images can be affected by sensor noise and cloud cover. One of the solutions to overcome these limitations is by combining existing satellite products to minimize the drawbacks of a dataset. In this dissertation, we improve the spatial and temporal resolution of satellite data products, minimize sensor noise, and remove cloud cover from satellite images by combining data from multiple satellite sensors using deep learning methods. Deep learning has been successful in natural image superresolution, denoising, and translation and these methods perform efficiently given sufficiently large datasets and computational resources. Therefore, publicly available satellite datasets and recent computational advancements provide an ideal opportunity for applying deep learning for our tasks. In our first study, we downscale low resolution optical and thermal spectral bands of MODIS to match higher resolution NIR and Red bands. Information extraction from satellite data often requires the combined use of multiple spectral bands. Usually, the low-resolution bands are downscaled using naive interpolation methods or high-resolution bands are upscaled to create spectral indices. We train a deep learning model for downscaling MODIS spectral to create a spatially consistent MODIS dataset. Our model is compared to a state-of-the-art satellite image downscaling method and a deep learning image superresolution method. Additionally, we investigate the importance of prior natural images towards downscaling satellite images. Next, we increase the effective spatial resolution and denoise MODIS spectral bands with the help of Landsat 8 images. MODIS and Landsat 8 have similar measurement principles and acquire spectral bands over similar wavelength ranges. Therefore, we use higher spatial resolution Landsat 8 images as ground truth during training time to generate an enhanced MODIS data product. We compare our model to state-of-the-art deep learning architectures SRCNN and a deep residual network. Additionally, we investigate the performance of multiple loss functions on our and the benefit of fine-tuning our model on a specific study area. Finally, we increase the temporal resolution of Sentinel-2 data product by combining it with Landsat 8. Data availability is a major issue in the field of remote sensing and moderate spatial resolution satellite imagery like Sentinel-2 and Landsat 8 is limited by their relatively lower temporal resolution. Landsat 8 and Sentinel-2 have revisit times of 16 and 5 days, respectively. We train a generative adversarial network, to transform Landsat 8 spectral bands to Sentinel-2 to generate a combined Sentinel-2 like dataset with an improved temporal resolution of approximately 3 days. Our study proposes a methodology to transform Landsat 8 Green and NIR spectral bands to predict Sentinel-2 like Red Edge spectral bands which are unavailable in Landsat 8. Due to the image acquisition time delay between the satellite sensors, we remove cloud cover from Sentinel-2 images by transforming Landsat 8 bands. Additionally, we minimize the image registration mismatch between Landsat 8 and Sentinel-2.
Desheng Liu, Dr (Advisor)
Alvaro Montenegro, Dr (Committee Member)
Srinivasan Parthasarathy, Dr (Committee Member)
Rongjun Qin, Dr (Committee Member)
124 p.
Geographic Information Science; Geography; Remote Sensing
MODIS, Landsat 8, Sentinel-2, Downscaling, Denoising, Temporal Resolution, Spatial Resolution, Spectral Resolution, Satellie Image Fusion, Deep Learning, Generative Adversarial Networks

Recommended Citations

Citations

  • Mukherjee, R. (2020). Improving Satellite Data Quality and Availability: A Deep Learning Approach [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1598286115582789

    APA Style (7th edition)

  • Mukherjee, Rohit. Improving Satellite Data Quality and Availability: A Deep Learning Approach. 2020. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1598286115582789.

    MLA Style (8th edition)

  • Mukherjee, Rohit. "Improving Satellite Data Quality and Availability: A Deep Learning Approach." Doctoral dissertation, Ohio State University, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=osu1598286115582789

    Chicago Manual of Style (17th edition)

osu1598286115582789
© 2020, some rights reserved.Creative Commons License Improving Satellite Data Quality and Availability: A Deep Learning Approach by Rohit Mukherjee is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.
Release 3.2.12