A Principal Component Analysis Noise Filter Value-Added Procedure to Remove Uncorrelated Noise from Atmospheric Emitted Radiance Interferometer (AERI) Observations
The atmospheric emitted radiance interferometer (AERI) is a ground-based passive sensor that measures downwelling emitted spectral radiance from 500-3000 cm-1 at 1 cm-1 resolution (Knuteson et al. 2004a, b). An AERI is located at each of the Atmospheric Radiation Measurement (ARM) Climate Research F...
| Main Authors: | , , |
|---|---|
| Language: | unknown |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://www.osti.gov/servlets/purl/1808544 https://www.osti.gov/biblio/1808544 https://doi.org/10.2172/1808544 |
| Summary: | The atmospheric emitted radiance interferometer (AERI) is a ground-based passive sensor that measures downwelling emitted spectral radiance from 500-3000 cm-1 at 1 cm-1 resolution (Knuteson et al. 2004a, b). An AERI is located at each of the Atmospheric Radiation Measurement (ARM) Climate Research Facility (ACRF) locales: Southern Great Plains (SGP), North Slope of Alaska (NSA), and Tropical Western Pacific (TWP). The AERI radiance observations at these three locales have been used to validate radiative transfer models (e.g., Tobin et al. 1999; Turner et al. 2004), retrieve thermodynamic profiles (e.g., Feltz et al. 2003), and investigate the microphysical properties of clouds (e.g., Turner 2005). The original temporal sampling strategy was a three-minute average of sky radiance every eight minutes; this sampling strategy was selected for the validation of clear-sky radiative transfer models and for atmospheric profiling (Knuteson et al. 2004a). However, cloud microphysical properties can change on the order of seconds in the narrow field-of-view of the AERI. Therefore, the ARM Program is in the process of improving the temporal resolution of the AERI to collect a sky spectrum every 15-30 seconds. The increased temporal resolution results in less averaging performed by the instrument, and hence the larger component of random noise in the sky spectra. A new value-added procedure (VAP) has been developed that uses the high correlation in the observed radiance across the spectrum to reduce the uncorrelated random error in the data using principal component analysis (PCA). The VAP automatically determines the appropriate number of principal components to use in the reconstruction to eliminate as much random noise as possible. A significant reduction in the uncorrelated random error in the data has been proven for both regular temporal data and rapid-sample (RS) data. |
|---|