Exploration of a Statistical Approach for the Calibration of the NOAA CrIS Sensors Using Machine Learning

The Cross-Track Infrared Sounder (CrIS) is a Michelson interferometer currently onboard three polar-orbiting satellites as a part of the Joint Polar Satellite System (JPSS), the Suomi National Polar-orbiting Partnership (SNPP), NOAA-20, and NOAA-21 satellites. The CrIS sensor measures the infrared radiance at top of the atmosphere at three spectral infrared bands: longwave (LWIR) from 650 to 1095 cm-1, midwave (MWIR) from 1210 to 1750 cm-1, and shortwave (SWIR) from 2155 to 2550 cm-1, in the form of interferograms. Observations from the CrIS sensor are critical for weather forecasting and environmental monitoring. This work is dedicated to explore and assess the capabilities of a statistical approach with Artificial Intelligence (AI) to perform the calibration of the CrIS observations, and identify the benefits and the challenges of the process. This includes quantifying the volume of data and the sampling of the data in time and space required to capture changes associated to the sounder as part of the calibration process. The computational efficiency and the quality that AI has shown in some technological applications makes it a promising tool to help aid the calibration of CrIS observations. A specific type of AI, neural networks, define a series of linear and nonlinear operations to transform the input (uncalibrated radiances) into a desired output (calibrated radiances), using a variable system of weights and biases to fine-tune the model based on its performance. Once a neural network is trained on a large amount of sample data, it can be used to speed up the large-scale calibration of CrIS observations, given that the AI calibrated radiances meet specific requirements. Because this is a largely unexplored area of research, this work is trying to lay the groundwork for future developments.

Disclaimer: The scientific results and conclusions, as well as any views or opinions expressed herein, are those of the author(s) and do not necessarily reflect those of NOAA or the Department of Commerce.