Search

We investigated the single scattering optical properties of snow for different ice particle shapes and degrees of microscopic scale roughness. These optical properties were implemented and tested in a coupled atmosphere-snow radiative transfer model. The modeled surface spectral albedo and radiance distribution were compared with surface measurements. The results show that the reflected radiance and irradiance over snow are sensitive to the snow grain size and its vertical profile. When inhomogeneity of the particle size distribution in the vertical is taken into account, the measured spectral albedo can be matched, regardless of the particle shapes. But this is not true for the modeled radiance distribution, which depends a lot on the particle shape. The usual “equivalent spheres” assumption significantly overestimates forward reflected radiances, and underestimates backscattering radiances, around the principal plane. On average, the aggregate shape assumption has the best agreement with the measured radiances to a mean bias within 2%. The snow optical properties with the aggregate assumption were applied to the retrieval of snow grain size over the Antarctic plateau. The retrieved grain sizes of the top layer showed similar and large seasonal variation in all years, but only small year to year variation. Using the retrieved snow grain sizes, the modeled spectral and broadband radiances showed good agreements with MODIS and CERES measurements over the Antarctic plateau. Except for the MODIS 2.13 μm channel, the mean relative model-observation differences are within few percent. The modeled MODIS radiances using measured surface reflectance at Dome C also showed good agreement in visible channels, where radiation is not sensitive to snow grain size and the measured surface bidirectional reflectance is applicable over the Antarctic plateau. But modeled radiances using local, surface-measured reflectance in the near infrared yielded large errors because of the high sensitivity to the snow grain size, which varies spatially and temporally. The CERES broadband shortwave radiance is moderately sensitive to the snow grain size, comparable to the MODIS 0.86 μm channel. The variation of broadband snow reflectance due to the seasonal variation in snow grain size is about 5% in a year over the Antarctic plateau. CERES broadband radiances simulated with grain sizes retrieved using MODIS are about 2% larger than those observed.

Advanced Very High Resolution Radiometer (AVHRR) images of arctic snow reflectance, observed from different satellite passes with the closest pass times on the same day, show different patterns that are related to the sun-satellite geometry. The variation of snow reflectance with view angle can be large or small, depending on the view geometry. Thus, the snow reflectances of a common target, observed from different viewing angles, can be very different and hard to compare. Unfortunately, none represent the actual physical snow albedo, and the error in the converted broadband top of atmosphere (TOA) albedo could easily be larger than 10%, which implies a great effect on the solar radiation balance at TOA. For an accurate estimate of the solar energy budget from satellite observations, an anisotropic correction is required not only for those reflectance images with obvious angular variation, but also for seemingly “isotropic” scenes because this latter case might incorrectly estimate the albedo under general conditions. Two methods, one based on a radiative transfer model and the other on National Oceanic and Atmospheric Administration's (NOAA) Earth Radiation Budget (ERB) experimental data, have been applied to AVHRR data to implement an anisotropic correction. The results show that both methods can remove the systematic variation of snow reflectance with view angle and can significantly reduce the large observed differences in reflectances from different satellite observations of a common snow target. The radiative transfer model-based method, however, unlike the NOAA experimental method, corrects the reflectances in the two AVHRR channels separately, has higher angular resolution than the NOAA experimental method, and can account for variations in snow condition (e.g., grain size). Moreover, the model-based anisotropic correction shows superior self-consistency compared to that produced by NOAA's experimental-based method. Finally, the model-based method provides a sound theoretical basis for anisotropic correction of satellite images of arctic snow reflectance.

Cloud top height is an important variable for both scientific and operational meteorological applications. Unfortunately, direct observations of cloud top height are generally not routinely made. This article develops a new hybrid method for estimating cloud top height from Advanced Very High Resolution Radiometer (AVHRR) data to help provide more abundant and accurate cloud top height estimates at low cost. It takes into account arbitrary viewing and illumination conditions which can affect the accuracy of both cloud shadow detection and cloud shadow-based cloud top height retrievals. Results were validated with data taken by NASA's ER-2 Cloud LIDAR System (CLS). The mean differences between the AVHRR-based cloud height estimates and the LIDAR observations for the two validation images analyzed are 0.45 km and 0.67 km. The sensitivity of the results to the parametrizations used by the new algorithm was also evaluated. Parameters can be chosen to minimize their effect on retrieved cloud top heights; the procedure executes quickly. Given the accuracy of the cloud top height retrievals, the method could be used routinely to provide cloud top heights for a wide variety of meteorological and hydrometeorological applications (e.g., numerical weather prediction, drought monitoring). With the improved spatial resolution (and reduced cost) of data from soon to be launched new satellite observing systems, proportionally higher accuracy and resolution for cloud top heights can be achieved economically using this method.