# Deep Feature Gaussian Processes for Single-Scene Aerosol Optical Depth Reconstruction

arXiv: https://arxiv.org/abs/2405.17262 · HTML: https://mlsanity.com/paper/2405.17262/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 28 references that are in the map.
- First posted on arXiv 27 May 2024.

## Builds on (most cited first, 7 of 7)

- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015)
- [Deep Neural Networks as Gaussian Processes](https://mlsanity.com/paper/1711.00165/) (2017)
- [GPyTorch: Blackbox Matrix-Matrix Gaussian Process Inference with GPU Acceleration](https://mlsanity.com/paper/1809.11165/) (2018)
- [Scalable Variational Gaussian Process Classification](https://mlsanity.com/paper/1411.2005/) (2014)
- [Few-Shot Hyperspectral Image Classification With Unknown Classes Using Multitask Deep Learning](https://mlsanity.com/paper/2009.03508/) (2020)
- [Practical Bayesian Modeling and Inference for Massive Spatial Datasets On Modest Computing Environments](https://mlsanity.com/paper/1802.00495/) (2018)
- [Active Ensemble Deep Learning for Polarimetric Synthetic Aperture Radar Image Classification](https://mlsanity.com/paper/2006.15771/) (2020)

## Built on by (most cited first, 2 of 2)

- [Daily Land Surface Temperature Reconstruction in Landsat Cross-Track Areas Using Deep Ensemble Learning With Uncertainty Quantification](https://mlsanity.com/paper/2502.14433/) (2025)
- [Resolution Revolution: A Physics-Guided Deep Learning Framework for Spatiotemporal Temperature Reconstruction](https://mlsanity.com/paper/2507.09872/) (2025)

## Abstract (arXiv)

Remote sensing data provide a low-cost solution for large-scale monitoring of air pollution via the retrieval of aerosol optical depth (AOD), but is often limited by cloud contamination. Existing methods for AOD reconstruction rely on temporal information. However, for remote sensing data at high spatial resolution, multi-temporal observations are often unavailable. In this letter, we take advantage of deep representation learning from convolutional neural networks and propose Deep Feature Gaussian Processes (DFGP) for single-scene AOD reconstruction. By using deep learning, we transform the variables to a feature space with better explainable power. By using Gaussian processes, we explicitly consider the correlation between observed AOD and missing AOD in spatial and feature domains. Experiments on two AOD datasets with real-world cloud patterns showed that the proposed method outperformed deep CNN and random forest, achieving R2 of 0.7431 on MODIS AOD and R2 of 0.9211 on EMIT AOD, compared to deep CNN's R2 of 0.6507 and R2 of 0.8619. The proposed methods increased R2 by over 0.35 compared to the popular random forest in AOD reconstruction. The data and code used in this study are available at urlhttps://skrisliu.com/dfgp.