The Havemann-Taylor Fast Radiative Transfer Code (HT-FRTC): A multipurpose code based on principal components. (November 2018)

• Record Type: Journal Article
• Authors: Havemann, Stephan; Thelen, Jean-Claude; Taylor, Jonathan P.; Harlow, R. Chawn
• Is Part Of: Journal of quantitative spectroscopy & radiative transfer. Volume 220(2018)
• Page Start: 180
• Page End: 192
• Publication Date: 2018-11
• Subjects: Hyperspectral -- Fast radiative transfer -- Sensor-agnostic principal component -- Kernel regression; 42.68.-W -- 42.68.Ay -- 95.75.-Z -- 95.75.Pq; Spectrum analysis -- Periodicals; Radiation -- Periodicals; Analyse spectrale -- Périodiques; Rayonnement -- Périodiques; Radiation; Spectrum analysis; Periodicals; 543.0858
• DOI: 10.1016/j.jqsrt.2018.09.008
• Languages: English
• ISSNs: 0022-4073

Abstract

Highlights: A new fast radiative transfer code that works across the electromagnetic spectrum. Application of Principal Components of very high spectral resolution. A sensor independent formulation for satellite, airborne and ground-based sensors. Line-by-line like, hyperspectral and broadband simulations. Kernel regression for increased accuracy and robustness. Abstract: The HT-FRTC can be used across the electromagnetic spectrum from the microwave through to the ultraviolet to calculate transmittance, radiance and flux spectra, which are represented by their principal components. The code uses monochromatic calculations at a small number of frequencies, which are selected by k-means clustering, to predict the principal component scores. It has been found that kernel regression yields more robust predictions than a linear regression. The principal components cover the spectrum at a very high spectral resolution, similar to that of conventional line-by-line models such that the individual spectral lines are resolved. This approach allows very fast line-by-line-like, hyperspectral and broadband simulations for satellite-based, airborne and ground-based sensors. The principal components are derived during a code training phase from monochromatic simulations for a diverse set of atmospheres and surfaces. They are sensor independent and no extra training is required for additional sensors. The HT-FRTC has been trained with all of the trace gases in the HIgh-resolution TRANsmission molecular absorption database (HITRAN) and on a large variety of surface emissivity/reflectance spectra. It can be run for any Lambertian or specular surface. The HT-FRTC is a plane-parallel one-dimensional model but some effects of the Earth's sphericity are accounted for. Solar, lunar and cosmic microwave background sources have been included. Scattering by frozen and liquid cloud, precipitation particles and twenty different aerosol species has been included, as well as Rayleigh scattering which is significant in the short-wave. The scattering phase function can be fully accounted for by an integrated monochromatic version of the Edwards-Slingo spherical harmonics radiation code or approximately by a modification to the extinction (Chou scaling).