Canopy emissivity characterization from hyper spectral infrared obervations

Surface emissivity is one of the most critical parameters when dealing with the use and interpretation of high spectral resolution observations in the infrared. The spectral signature of the surface becomes relevant in the range with higher transmittance and, therefore, an incorrect model for the surface emissivity could sensitively bias, e.g., the retrieval of geophysical parameters in the lower part of the atmosphere.

In this work we illustrate a techniques which has been used to characterize the surface emissivity of the area around our arm site at the IMAA institute in southern Italy. The area surrounding the arm site is mostly covered by a forest. For this kind of vegetation (plant canopy), the surface emissivity is typically modeled with that of a black-body (emissivity 1 and no dependence on the field of view angle). Our results show that this is, indeed, not the case and the above straightforward model can produce highly biased retrieval for temperature and water vapor.

The analysis has been performed by using NAST-I (NPOESS Aircraft Sounder Testbed Interferometer) data recorded during the Italian phase of the EAQUATE (European AQUA Thermodynamic Experiment) campaign. The international experiment EAQUATE was held in September 2004 in Italy and in the United Kingdom. One of the goals of the Italian campaign was to demonstrate the benefit of ground-based and airborne systems to validate satellite observations (and related retrieval products) of hyper spectral infrared sensors, such as AIRS spectrometer aboard the AQUA satellite.

The set of NAST-I spectra, which have been used in the analysis, were recorded over a plant canopy (deciduous and conifer forest), while ancillary radio-sonde observations for temperature and water vapor were performed from the arm site. These ancillary information, along with other observations from the arm site, allows us to have an independent knowledge of the thermodynamic atmospheric state during the NAST-I observations. Synthetic spectral radiance can be, therefore, computed and compared to the observations. The resulting spectral residual, at wave number of high transmittance, can be then minimized by properly adjusting the spectral emissivity. In order to filter out the possible effect of the uncertainties on the trace gases amount, synthetic ad observed spectral radiance were re-sampled at a rate of 2 cm-1. The forward model LBLRTM was used in our analysis.

The resulting emissivity has been compared to laboratory measurements (Salisbury database) and a good agreement was found in the long wave region, whereas a relatively large discrepancy was found in the shortwave region. The emissivity was then used to invert the NAST-I observations for temperature and water vapor retrieval and a very nice agreement was found with radiosonde observations.