Mary Ellen Craddock, Northrop Grumman Corporation; Danny Felton, Northrop Grumman Corporation; Heather Kiley, Northrop Grumman Corporation; Randall J. Alliss, Northrop Grumman
Keywords: atmospherics, machine learning, validation
Abstract:
In December of 2021, NASA launched its Laser Communications Relay Demonstration (LCRD). LCRD will demonstrate bi-directional space to ground optical communications to two optical ground stations (OGS). The OGS at the Table Mountain Facility, known as OGS-1, and the OGS-2 at Haleakala summit will serve as the two ground sites. In mid-January 2022, LCRD demonstrated its first light through NASA optical telescopes. Over the next two years, LCRD will demonstrate the benefits of optical communications. Laser communications provide secure, high data rate transmission in the absence of strong atmospheric fading that includes cloud liquid water, ice, and atmospheric aberrations produced by pancake layer density gradients. Two of the major goals of LCRD are to quantify the impacts of the atmosphere on optical transmissions and to predict link handovers between OGS-1 and OGS-2. The Atmospheric Monitoring System (AMS) deployed to OGS-2 in 2017 has been collecting atmospherics in real-time and is now supporting the goal of quantifying the impacts on LCRD optical transmissions. For example, data from the AMS is providing minute by minute estimates of atmospheric fades due to clouds from a laser ceilometer as well as horizon to horizon imagery of the clouds. Additionally, AI-powered atmospheric decision aids based on the AMS are being run to support link handovers. Northrop Grumman is using a U-Net neural network and multi-layer perceptron model trained on high performance computing GPUs. The resulting decision aids are developed using many terabytes of AMS data collected over the last several years. Results prior to launch of LCRD showed a remarkable ability to predict the short term atmospheric and space environment in and around the line of sight to the spacecraft. For example, the ten-minute cloud prediction skill score beats a basic persistence forecast. This implies that the decision aids are able to predict a change in state of the atmospheric transmission, something a persistence forecast is unable to do. In addition, these predictions are showing very little bias.
This talk will provide an overview of the AMS, its real-time data collects, and its predictive capabilities. In addition, the talk will report on the development of atmospheric predictive models at OGS-2 for time scales between zero and 48 hours and subsequent validation efforts using in situ instrument and LCRD data. Comparisons to LCRD link quality performed during an on-site visit to OGS-2 highlight the benefits of using in situ data to support the validation of the atmospheric estimated link fades and predictions. The ultimate goal of this work is to show that atmospheric characterization and prediction is essential for any ground based optical system whether it be for space situational awareness (SSA) or optical communication applications. The talk will show how the independent data from LCRD may be used to validate ground-based characterization and prediction systems.
Date of Conference: September 27-20, 2022
Track: Atmospherics/Space Weather