Search

Since the 2018 launch of NASA's ICESat‐2 satellite, numerous studies have documented the bathymetric measurement capabilities of the space‐based laser altimeter. However, a commonly identified limitation of ICESat‐2 bathymetric point clouds is that they lack accompanying spectral reflectance attributes, or even intensity values, which have been found useful for benthic habitat mapping with airborne bathymetric lidar. We present a novel method for extracting bathymetry from ICESat‐2 data and automatically adding spectral reflectance values from Sentinel‐2 imagery to each detected bathymetric point. This method, which leverages the cloud computing systems Google Earth Engine and NASA's SlideRule Earth, is ideally suited for "big data" projects with ICESat‐2 data products. To demonstrate the scalability of our workflow, we collected 3,500 ICESat‐2 segments containing approximately 1.4 million spectrally‐attributed bathymetric points. We then used this data set to facilitate training of a deep recurrent neural network for classifying benthic habitats at the ICESat‐2 photon level. We trained two identical models, one with and one without the spectral attributes, to investigate the benefits of fusing ICESat‐2 photons with Sentinel‐2. The results show an improvement in model performance of 18 percentage points, based on F1 score. The procedures and source code are publicly available and will enhance the value of the new ICESat‐2 bathymetry data product, ATL24, which is scheduled for release in Fall 2024. These procedures may also be applicable to data from NASA's upcoming CASALS mission. Plain Language Summary: NASA's Ice, Cloud, and land Elevation Satellite 2 (ICESat‐2) uses green laser beams to map the elevation of the Earth's surface. These beams can also penetrate water, allowing the satellite to map the seafloor elevation up to a certain depth. These measurements are important for research on many topics, including shallow marine habitat mapping. In many cases, ICESat‐2 seafloor depths are combined with color information from satellite imagery. However, it is difficult to combine large quantities of these two data types because they are distributed by different systems and come in different formats. In this paper, we present a workflow for ICESat‐2 seafloor mapping that uses free, cloud computing resources from NASA and Google. By leveraging these computing resources, we can combine large amounts of ICESat‐2 data with satellite imagery. These large data sets are highly sought after for projects in artificial intelligence. We demonstrate this by building our own artificial intelligence model to map shallow marine habitats from ICESat‐2 data. A statistical analysis shows that combining ICESat‐2 with satellite imagery increases accuracy by 18 percentage points. Future NASA satellite missions will use similar sensors to ICESat‐2 and this work may be useful for future research with those satellites. Key Points: We present a cloud computing workflow for ICESat‐2 bathymetry extraction and data fusion that facilitates deep learningWe introduce a novel approach for extracting bathymetric photons from ICESat‐2 using techniques from graph theory and network analysisWe demonstrate the efficacy of our techniques by training a recurrent neural network to classify benthic habitats at the photon scale [ABSTRACT FROM AUTHOR]

The ICESat‐2 (Ice, Cloud and Land Elevation Satellite‐2) photon‐counting laser altimeter technology required the design and development of very sophisticated onboard algorithms to collect, store and downlink the observations. These algorithms utilize both software and hardware solutions for meeting data volume requirements and optimizing the science achievable via ICESat‐2 measurements. Careful planning and dedicated development were accomplished during the pre‐launch phase of the mission in preparation for the 2018 launch. Once on‐orbit all of the systems and subsystems were evaluated for performance, including the receiver algorithms, to ensure compliance with mission standards and satisfy the mission science objectives. As the mission has progressed and the instrument performance and data volumes were better understood, there have been several opportunities to enhance ICESat‐2's contributions to Earth observation science initiated by NASA and the ICESat‐2 science community. We highlight multiple updates to the flight receiver algorithms, the onboard software for signal processing, that have extended ICESat‐2's data capabilities and allowed for advanced science applications beyond the original mission objectives. Plain Language Summary: NASA launched its second Earth observing laser altimeter in 2018 with mission objectives of collecting observations in support of Earth science as a window into climate change impacts on our planet. Pre‐launch studies focused on specific instrument settings and on‐board data processing to support the mission objectives without violating data volume constraints. Once the instrument was on‐orbit and operational, evaluation of the algorithms for success in signal detection, signal finding, and signal telemetry was undertaken. In response to the evaluation, updates have been made to optimize the data provided by the mission. Key Points: Since the 2018 ICESat‐2 (Ice, Cloud and Land Elevation Satellite‐2) launch multiple updates have been made to the onboard flight receiver algorithm parameters to improve access and utility of the data for a multi‐disciplinary science communityThe adjustments have been primarily in the in the vertical telemetry band settings and have facilitated enhanced observations of blowing snow and increased detection opportunities of bathymetry in nearshore environmentsThe parameter changes have been made to mitigate data losses in certain situations and to advance science applications outside of the primary science objectives of the ICESat‐2 mission [ABSTRACT FROM AUTHOR]

Early spaceborne laser altimetry mission development starts in pre-phase A design, where diverse ideas are evaluated against mission science requirements. A key challenge is predicting realistic instrument performance through forward modeling at an arbitrary spatial scale. Analytical evaluations compromise accuracy for speed, while radiative transfer modeling is not applicable at the global scale due to computational expense. Instead of predicting the arbitrary properties of a lidar measurement, we develop a baseline theory to predict only the distribution of uncertainty, specifically for the terrain elevation retrieval based on terrain slope and fractional canopy cover features through a deep neural network Gaussian mixture model, also known as a mixture density network (MDN). Training data were created from differencing geocorrected Global Ecosystem Dynamics Investigation (GEDI) L2B elevation measurements with 32 independent reference lidar datasets in the contiguous U.S. from the National Ecological Observatory Network. We trained the MDN and selected hyperparameters based on the regional distribution predictive capability. On average, the relative error of the equivalent standard deviation of the predicted regional distributions was 15.9%, with some anomalies in accuracy due to generalization and insufficient feature diversity and correlation. As an application, we predict the percent of elevation residuals of a GEDI-like lidar within a given mission threshold from 60°S to 78.25°N, which correlates to a qualitative understanding of prediction accuracy and instrument performance. [ABSTRACT FROM AUTHOR]