7.2
A new infrared land surface emissivity database for the Community Radiative Transfer Model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 21 January 2010: 8:45 AM
B313 (GWCC)
Ronald L. Vogel, NOAA/NESDIS/STAR, Camp Springs, MD; and Q. Liu, B. Ruston, Y. Han, and F. Weng

Land surface emissivity is an important parameter in satellite remote sensing studies involving surface radiative properties, including surface-atmosphere interactions, surface energy balance, and assimilation of satellite radiances into numerical weather prediction models. In particular, emissivity is necessary for converting satellite measured brightness temperature to kinetic surface temperature. We seek to improve the infrared land surface emissivity component of the Community Radiative Transfer Model (CRTM). This model is used operationally at NOAA/NCEP for assimilating satellite radiances into the Global Forecast System numerical weather prediction model.

Currently, the CRTM contains a database of land surface emissivity values for the visible and infrared spectrum (0.2 - 15.0 μm) using a surface classification scheme consisting of 24 classes, i.e. each spectrum represents the emissivity of each of the surface classes. There is no temporal variation in the emissivity spectra, i.e. the spectra do not account for seasonal vegetation changes. The emissivity database was developed for the NPOESS algorithm program, but the derivation of the spectra was not recorded.

We developed a new set of infrared emissivity spectra, specifically designed for the surface classification scheme used by the Global Forecast System. In addition, a second emissivity database was designed for the IGBP classification scheme, a widely used surface classification scheme in the scientific community. The new database provides spectral surface emissivity (0.4 – 14.0 μm) and accounts for temporal vegetation change by using green vegetation fraction based on 25 years of AVHRR data (Jiang et al., 2008). The new emissivity spectra were generated by combining reflectance spectra from JPL's ASTER Spectral Library (Baldridge et al., 2008) for various surface materials in proportions representative of the surface classes of the classification schemes. The proportions were determined by finding the best fit (lowest bias) between CRTM simulations and GOES-12 Imager observations for the 3.9 μm and 10.7 μm channels.

Independent comparisons of CRTM simulations to satellite observations show that use of the new emissivity database reduces bias between CRTM and GOES Imager observations by 80% for the 3.9 μm channel and 15% for 10.7 μm channel, compared to CRTM's older emissivity database, for nighttime scenes across all seasons. Future work includes expanding the bare-soil classes to provide additional emissivity spectra for different soil types, since soil represents the highest variability in surface emissivity. In addition, a BRDF correction for daytime radiance simulations will be included in CRTM.