Triana measures visible and near-infrared reflectance globally from sunrise to sunset at an almost constant scattering angle between 165° and 178°. The scattering angle for any other satellite at a given location varies with time of day and overpass (e.g., Minnis et al., 1998). Triana‚s spatial and spectral coverage and the scattering angles resulting from its unique view are ideally suited for helping us to monitor clouds, a critical component of the climate system, and to determine the statistics of the global distribution of cloud particle shape.
In recent years, advances have been made in our capabilities for monitoring clouds and their constituents. However, statistically reliable measurements of the shapes of ice crystals comprising cirrus clouds are poorly known. Ice-crystal shape and size determine the basic reflectance properties of clouds. Cloud reflectance is a key factor in calculations of how the Earth responds to incoming solar radiation.
From in situ aircraft measurements, it is known that ice-crystal shapes vary considerably from cloud to cloud. But it is not known how frequently or in what conditions a particular crystal shape occurs. These shapes produce very different scattering phase functions (see Figure 17). Ice crystal habit is difficult to monitor because different crystal shapes can produce similar reflectances in a given direction by adjusting the individual crystal sizes. One means to differentiate one crystal habit from another is to analyze simultaneous measurements from two different angles such that the solutions for different habits yield distinctly difference reflectance ratios. The optimal pairs of angles for such measurements include one between 160° and 178° and another between 60° and 180° to maximize the relative differences in the scattering phase functions (e.g., Figure 17).
Nearly simultaneous measurements from two different satellites have been used to determine the correct optical depth by selecting the phase function that yields the same optical depth from both satellite views (Minnis et al., 1993). Because of differences in the shape of the phase function and the asymmetry factors, the optical depth for an ice crystal will differ from that for a water droplet at the Triana scattering angle (~175°). Ratios of reflectance observed at angles other than 175° will also be considerably different at most angles (Figure 18a) thus providing an estimate of phase. Figure 18b shows a matched set of images from GOES-8 (75°W) and GOES-10 (135°W). The GOES-10 reflectances are generally smaller than those observed from GOES-8 which views the entire scene from a scattering angle of ~167°, in the range seen from Triana. The ratios of the GOES-10 reflectances to those from GOES-8 show that, except in the areas with shadows, the values for the cold clouds (see Fig. 18b) are close to 1.0 while the warmer clouds have ratios closer to 0.85. These ratios are consistent with the results on the right in Figure 18a indicating that the colder clouds are composed of hexagonal ice crystals and water droplets comprise the lower clouds. Similar differences in the ratios exist for clouds composed of crystals having different predominant shapes (e.g. Fig. 17).
One of the greatest stumbling blocks to using multiple satellite measurements is calibration. This obstacle can be eliminated by using the technique of Nguyen et al. (1999) to produce near-real-time intercalibration tables normalizing Triana and other satellites to a common standard. This technique uses simultaneous data from two satellites with nearly identical viewing conditions to obtain a relative calibration from one to another. It is currently applied to GOES-8, GOES-10, GMS, NOAA-12, VIRS, and ATSR-2 using the NOAA-14 calibration as a standard. When Triana is in its prescribed orbit, its 645 and 870 nm channels will be calibrated against similar channels on the Terra MODIS instruments. This calibration can then be easily transferred to VIRS, the NOAA-14/15 AVHRRs, and the GEO satellites, including the new Meteosat which will have comparable visible channels, to facilitate scientific analyses of multiple satellite data sets.
Cloudy Triana pixels will be determined via multispectral thresholding against expected clear-sky reflectances. An initial clear-sky reflectance map will be developed for the 645 and 870 nm channels from existing databases used by the CERES program (Trepte et al., 1999; Sun-Mack et al., 1999). These databases will be updated for the 645 and 870 nm channels and for other channels using the initial Triana observations. Screening for clear pixels will involve both subjective˜initially˜and objective minimum reflectance techniques. Shadows will not be problematic because of the Triana scattering angles. Over ocean, the updated bi-directional reflectance model of Minnis and Harrison (1984) will be used for characterizing the reflectance patterns for clear ocean, except near the coasts. Appropriate sets of thresholds will then be established for each channel and surface type to discriminate cloudy and clear pixels automatically in the Triana data set. The resulting cloudy pixels will be used in the algorithms for determining cloud height, while optical depth will be derived using assumed particle sizes and shapes as in Minnis and Smith (1998) based on the cloud height. The clear pixels will be used in other studies including the hot spot analyses discussed below.
A large database of cloud reflectances based on a variety of different particle shapes and sizes will be constructed for the two relevant Triana channels and, for the other satellites, all of the appropriate channels required for particle size, phase, and optical depth retrievals. Current retrieval algorithms will be applied to pixel-level data from the other satellites to obtain solutions for all of the various shapes. These results will then be matched to the Triana pixels. This matching will be accomplished by compiling groups of high-resolution LEO/GEO pixels into the 8-km Triana pixels. The optical depth for each pixel will be computed for each of the solutions using the Triana-observed radiances. Particle shape will be selected by determining which Triana-derived optical depth most closely matches its counterpart from the other satellite.
Extensive GEO and LEO data sets including GOES, GMS, AVHRR, and VIRS are currently downloaded and archived at NASA Langley Research Center. In the future, MODIS and Meteosat data will be included. These data sets will be used to establish a prototype, semi-operational pixel-matching algorithm that can be expanded in the future to a more operational process.
