MOBILITY ENGINEERING Produced by SAE Media Group
MOBILITY ENGINEERING
Magazines
Magazine cover

Current Issue

Archives

Magazine cover

Current Issue

Archives

Magazine cover

Current Issue

Archives

Magazine cover

Current Issue

Archives

Automated Atmospheric Correction of Nanosatellites Using Coincident Ocean Color Radiometer Data

Researchers present a machine-learning-based method for utilizing traditional ocean-viewing satellites to perform automated atmospheric correction of nanosatellite data.

Standard atmospheric correction process implemented within NRL’s APS, built upon NASA’s SeaDAS processing code. (Image: Naval Research Laboratory)

The prevailing mission-based paradigm for ocean color remote sensing typically involves high-cost satellite platforms launched and operated by government agencies such as NASA, NOAA, ESA, and JAXA. These platforms host state-of-the-art ocean-viewing radiometers with design and sensitivity specifications appropriate for delineating a comparatively weak water-leaving radiance from the total radiant signal detected at the top of the atmosphere. The current suite of such operational ocean color sensors includes NASA’s Moderate Resolution Imaging Spectroradi-ometer (MODIS; Aqua satellite), NOAA’s VIIRS (SNPP and NOAA-20 satellites), the Ocean and Land Color Instrument (OLCI; Sentinel-3 A/B satellites), and the Second-Generation Global Imager (SGLI) onboard the GCOM-C satellite. All of these sensors provide multi-spectral band sets (visible, near-infrared (NIR), and shortwave infrared (SWIR)) with daily coverage at approximately kilometer-scale spatial resolution. However, even kilometer-scale spatial resolution may be unable to resolve finer-scale features near rivers and estuaries that are critical for scientific and environmental resource management applications.

In contrast to government-sponsored, large satellite missions, commercial entities are now deploying low-cost cubesats with much higher spatial resolution imaging capability. Planet Labs currently has ~200 (as of November 2022) orbiting nanosatellites that provide high-temporal-resolution monitoring of the Earth’s surface at a spatial resolution previously available only by the high-cost tasking of a few specialized satellites such as World-View. The majority of these nanosatellites (known collectively as PlanetScope (PS)) host a multispectral digital camera with blue, green, red, and NIR bands, which image the earth at very high spatial resolution (3.125 m ground resolution at nadir). PS data have been used to monitor volcano activity, assess vegetative index, aid agriculture studies, study lake dynamics, determine high-resolution topography, detect oil spills, monitor rangeland, and monitor disasters. PS applications for aquatic environments include monitoring coral reefs and water quality, Sargassum detection, detection of river ice and water velocities, high-resolution bathymetry, and monitoring seagrass beds.

Other commercial groups are also exploiting nanosatellite technology for a wide variety of remote sensing applications. However, there have been only a small number of studies to assess these commercial nanosatellite data sources as a viable solution for ocean color remote sensing, i.e., detection of the water-leaving radiant signal in the visible bands after removal of the intervening atmospheric contamination. Maciel et al. studied the potential for cubesats to provide remote sensing reflectance over very turbid inland lakes. Vanhellemont applied the dark spectrum fitting (DSF) aerosol correction method to PS data and also found success in the PS red bands over very turbid waters.

More generalized ocean color applications of PS data will require removal of the atmospheric portion of the total sensor signal at the top of the atmosphere. Nearly 30 years ago, Gordon and Wang set the standard for atmospheric correction by using a relationship between two relatively narrow NIR or SWIR bands to estimate the aerosol radiance contribution in a satellite’s total path radiance. The use of two NIR bands is still one of the primary methods used today for characterizing and removing the aerosol radiance during atmospheric correction. However, the design of many small satellites is focused on terrestrial observation, and these sensors do not have the NIR/SWIR wavelengths needed for the standard method of atmospheric correction for ocean color. This inadequacy suggests that alternative methods should be explored for atmospheric correction. In this paper, we present an alternative atmospheric correction method in order to exploit PS data for ocean color applications. We selected machine learning (ML)-based techniques as a means to convolve data from traditional ocean color sensors, which permit a complete atmospheric correction, with PS data that are otherwise inadequate for this purpose.

This work was performed by Sean McCarthy, Summer Crawford, Christopher Wood, and Mark D. Lewis for the Naval Research Laboratory. For more information, download the Technical Support Package (free white paper) under the Communications category.

Topics:
Aerospace Cameras Coatings, colorants and finishes Communications Data acquisition and handling Imaging Imaging and visualization Machine Vision Optics Remote sensing Satellite communications Satellite/Spacecraft Communications Satellites Sensors and actuators Telecommunications Vehicles Water
This Brief includes a Technical Support Package (TSP).
Document cover
Automated Atmospheric Correction of Nanosatellites Using Coincident Ocean Color Radiometer Data

(reference ARL-96553) is currently available for download from the TSP library.

Don't have an account?

More From SAE Media Group

    Magazine cover
    Aerospace & Defense Technology Magazine

    This article first appeared in the June, 2023 issue of Aerospace & Defense Technology Magazine (Vol. 8 No. 4).

    Read more articles from this issue here.

    Read more articles from the archives here.

    Overview

    The document presents a study focused on enhancing the accuracy of ocean color measurements from nanosatellites through automated atmospheric correction using machine learning techniques. The research is particularly relevant given the increasing number of nanosatellites, such as those operated by Planet Labs, which provide high-temporal-resolution monitoring of the Earth's surface. As of November 2022, Planet Labs had approximately 200 orbiting nanosatellites, offering data that was previously only available from high-cost specialized satellites.

    The study employs a machine-learning approach that utilizes multispectral data from the Suomi National Polar-orbiting Partnership (NPP) Visible Infrared Imaging Radiometer Suite (VIIRS). The primary objective is to correct atmospheric interference in the data collected by nanosatellites, thereby improving the accuracy of ocean color measurements. The methodology involves creating predictive models for remote sensing reflectance (nLw) at various visible wavelengths, specifically focusing on a linear regression model validated against in situ measurements from the Marine Optical Buoy (MOBY) and AERONET-OC.

    The authors detail the process of model development, which includes training the model on a portion of the dataset and validating it against a withheld subset to ensure accuracy. The study highlights the use of decision trees as a supervised learning method, which is advantageous due to its applicability for both regression and classification tasks, lack of requirement for feature scaling, and ease of interpretation through visualization. The decision tree algorithm is described as a non-parametric supervised method that utilizes a binary tree structure to assign target values based on data features.

    The results indicate that the machine-learning models successfully predict nLw values, demonstrating the potential for improved oceanographic applications through the integration of nanosatellite data with traditional ocean color satellite measurements. This research not only contributes to the field of remote sensing but also emphasizes the importance of leveraging advanced algorithms to enhance the quality of environmental monitoring from space.

    In summary, the document outlines a significant advancement in the use of machine learning for atmospheric correction of nanosatellite data, showcasing its implications for ocean color measurement accuracy and broader environmental monitoring efforts.

    Top Stories

    INSIDERManned Systems

    Feature Image Venus Aerospace’s Rotating Detonation Rocket Engine Completes First Flight...

    INSIDERSoftware

    Feature Image Bombardier is Digitally Upgrading its Aircraft Design, Engineering and...

    INSIDERMaterials

    Feature Image New Copper Alloy Could Provide Breakthrough in Durability for Military Systems

    NewsEnergy

    Feature Image ACT Expo 2025: Heavy-Duty EVs, H2 Trucks and Tariff Talk Dominate Day One

    INSIDERResearch Lab

    Feature Image Quarterhorse Hypersonic Test Aircraft Completes First Flight

    INSIDERCommunications

    Feature Image How the US Army is Advancing Research in Robotics, AI and Autonomy

    Webcasts