Adversarial Attacks against a Satellite-borne Multispectral Cloud Detector

TL;DR

This paper demonstrates that deep learning-based satellite cloud detectors are vulnerable to adversarial attacks, which can bias cloud detection results by superimposing optimized patterns in multispectral satellite imagery.

Contribution

It introduces the first adversarial attack methodology targeting multispectral satellite cloud detection and explores mitigation strategies for these vulnerabilities.

Findings

Adversarial patterns can bias cloud detection in satellite imagery.

Attacks are effective across multispectral bands, including non-visible spectra.

Mitigation strategies can reduce attack success.

Abstract

Data collected by Earth-observing (EO) satellites are often afflicted by cloud cover. Detecting the presence of clouds -- which is increasingly done using deep learning -- is crucial preprocessing in EO applications. In fact, advanced EO satellites perform deep learning-based cloud detection on board the satellites and downlink only clear-sky data to save precious bandwidth. In this paper, we highlight the vulnerability of deep learning-based cloud detection towards adversarial attacks. By optimising an adversarial pattern and superimposing it into a cloudless scene, we bias the neural network into detecting clouds in the scene. Since the input spectra of cloud detectors include the non-visible bands, we generated our attacks in the multispectral domain. This opens up the potential of multi-objective attacks, specifically, adversarial biasing in the cloud-sensitive bands and visual camouflage in the visible bands. We also investigated mitigation strategies against the adversarial attacks. We hope our work further builds awareness of the potential of adversarial attacks in the EO community.