Rand E. Feind

Cloud Liquid Water Measurements on the Armored T-28: Intercomparison between Johnson–Williams Cloud Water Meter and CSIRO (King) Liquid Water Probe

Abstract Comparisons are made between liquid water concentration (LWC) readings obtained from a Johnson–Williams (J–W) cloud water meter and a King (Commonwealth Scientific and Industrial Research Organisation) liquid water probe, both mounted on the armored T-28 research aircraft during penetrations of springtime convective storms in Oklahoma and Colorado. The King probe readings are almost always higher, being up to twice those of the J–W instrument in clouds with narrower cloud droplet spectra. In clouds with broader droplet spectra, the ratio often climbs to three or greater. The King probe responds partially to drops larger than cloud droplet size, and also to some ice particles, so its reading can be higher than the cloud LWC present. However, this and earlier comparisons by others indicate that the primary reason for this discrepancy is that the J–W probe often underestimates the cloud LWC due to incomplete response to larger cloud droplets. Thus, published studies involving cloud LWC in convective...

The ASTER polar cloud mask

This research is concerned with the problem of producing polar cloud masks for satellite imagery. The results presented are for Thematic Mapper (TM) data from the northern and southern polar regions, however, the techniques discussed will be applied to ASTER data when it becomes available. A series of classification techniques have been implemented and tested, the most promising of which is a neural network classifier. To use a neural network classifier, the pixels in the data must be transformed into feature vectors, some of which are used for training the network and the remainder of which are reserved for testing the final system. The first challenge is the identification of pure pixel samples from the imagery. The Interactive Visual Image Classification System (IVICS) was developed specifically for this project to make this task simpler for the human expert. After labeling the pixels, the feature vectors are generated. One hundred and forty potential vector elements, consisting of linear and nonlinear combinations of the satellite channel data, have been identified. Because smaller input vectors reduce the difficulty of training and can improve classification accuracy, the set of potential vector elements must be reduced. Two techniques have been tested: a histogram-based selection method and a fuzzy logic method. Both have proven effective for this task. Although the polar region is the only area considered in this work, a system that can produce cloud masks for all areas of the globe will be required. Thus, speed, extensibility, and flexibility requirements must be added to the accuracy constraint. To achieve these goals, a two-stage classification approach is used. The first stage uses a series of static and adaptive thresholds derived from statistical analysis of the polar scenes to reduce the set of possible classes to which a pixel may be assigned, once a cluster of classes has been selected, a neural network trained to distinguish between the classes in the cluster is used to make the ultimate classification.

Cloud fraction and cloud shadow property retrievals from coregistered TIMS and AVIRIS imagery: the use of cloud morphology for registration

The Thermal Infrared Multispectral Scanner (TIMS) and the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) were operated simultaneously from the ER2 aircraft during a March 1990 test over the Rio Bravo region, Belize. Coregistration of the imagery obtained by these two instruments is necessary to utilize the data effectively. A technique for registering the TIMS imagery to AVIRIS imagery is presented. It takes advantage of the morphology of the fair weather cumulus (FWC) clouds present in the imagery for estimating inter-sensor distortions. It relies on an iterative process in which skew, nearest neighbor sampling, and cross-correlation (1D and 2D) are applied. Comparison between the AVIRIS three-band ratio (3BR) imagery and the coregistered TIMS imagery shows that TIMS is superior in detecting thin cloud and cloud edge pixels, especially over shadowed background. Although the seven scenes analyzed in the study were obtained within the same one-hour time period and over the same geographical region, the optimum temperature threshold for cloud detection, with respect to the background temperature, varies significantly from 2.1 to 3.3/spl deg/C. These values agree with the AVIRIS 3BR cloud fraction equivalent temperature thresholds to within 0.5/spl deg/C. When applying a cloud shadow mask from the AVIRIS near infrared imagery to the coregistered TIMS background imagery, a 1/spl deg/C temperature differential is found between the shadowed and nonshadowed background. This significant radiative cooling by Fair Weather Cumulus cloud shadows could introduce errors in surface emissivity retrievals by other Earth Observing System (EOS) investigators. >

Analysis of cloud optical thickness and effective radius retrievals for broken cloud fields

Radiative transfer model results at two wavelengths (i.e., visible and near IR) can be applied to satellite imagery to infer optical properties of clouds. In this study, the authors examine several types of cloud fields to gain an understanding of the effect of broken cloudiness on cloud property retrievals. Reliable estimates of cloud properties can be obtained for large scale, stratiform cloud fields over water. The same appears to be true even for optically thin (/spl tau/<10), broken cloud fields over water. However, such is not the case for continental fair weather cumulus cloud fields. Three-dimensional cloud effects cause significant disagreement between measured and modelled reflectance. The authors also examine the effects of spatial resolution on cloud property retrievals. When simulating imagery obtained by a low spatial resolution instrument with their high spatial resolution imagery, the 3D cloud effects are filtered out and subsequent retrievals appear reasonable; however, because of the nonlinear relationship between reflectance and optical properties, some bias is likely.<<ETX>>

The Effect Of Spatial Resolution Upon Cloud Optical Property Retrievals Part I: Optical Thickness

High spectral and spatial resolution Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) imagery is used to study the effects of spatial resolution upon fair weather cumulus cloud optical thickness retrievals. As a preprocessing step, a variation of the Gao and Goetz three-band ratio technique is used to discriminate clouds from the background. The combination of the elimination of cloud shadow pixels and using the first derivative of the histogram allows for accurate cloud edge discrimination. The data are progressively degraded from 20 m to 960 m spatial resolution. The results show that retrieved cloud area increases with decreasing spatial resolution. The results also show that there is a monotonic decrease in retrieved cloud optical thickness with decreasing spatial resolution. It is also demonstrated that the use of a single, monospectral reflectance threshold is inadequate for identifying cloud pixels in fair weather cumulus scenes and presumably in any inhomogeneous cloud field. Cloud edges have a distribution of reflectance thresholds. The incorrect identification of cloud edges significantly impacts the accurate retrieval of cloud optical thickness values.

ASTER polar cloud mask

An algorithm is currently under development that will provide a classification mask for ASTER imagery obtained poleward of 60 N and 60 S. The classification mask will be a product available through EOSDIS and is called the ASTER polar cloud mask. Ten classes are currently in the mask and include six clear classes (water, slush/wet ice, ice/snow, land, shadow on land, and shadow on ice/snow) and four cloud classes (thin cloud over ice/snow, water, or land, and thick cloud). The algorithms is designed as a four stage process. In the first stage the data are median filtered, sampled to 30 m spatial resolution, normalized, and navigated to coastlines and ancillary Earth surface databases. In the second stage, through adaptive thresholding, simple decision surfaces, and ancillary data, the class ambiguity of each pixel is reduced from ten to two to four classes. In the third stage, additional features are utilized in a paired- histogram classification methodology to make the final pixel classification. And finally, in the fourth stage, a simple spatial consistency check is performed over the entire classification mask to detect isolated pixel classifications. Over 3700 samples have been extracted and labeled to date representing over one million pixels from 82 Landsat TM circumpolar scenes. Tests of the algorithm on the labeled samples indicate that the clear/cloud classification accuracy is greater than 90 percent and subjective evaluation of the classification masks supports that result.

Effect of spatial resolution on cloud area retrievals: a comparison of results derived from the airborne visible/infrared imaging spectrometer (AVIRIS) and the thermal infrared multispectral scanner

Thermal Infrared Multispectral Scanner (TIMS) imagery and Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) imagery are registered to each other for seven scenes. Cloud area, as determined by applying the 3-band ratio method to AVIRIS imagery, is used to determine the optimum temperature threshold in the TIMS imagery. For this set of scenes, a threshold of 2 degree(s) to 3 degree(s) below the background temperature provides for the most accurate cloud pixel identification. Five to 8% differences in cloud area are found when comparing cloud pixel identification between AVIRIS and TIMS. Some of the differences are due to misregistration; however, at least half are due to differences in the mensuration process. It is demonstrated that cloud edges have a distribution of temperature thresholds, indicating the difficulty of locating cloud edges with a single temperature threshold. It is found that cloud edges occupy nearly half of the entire range of scene temperatures, significantly overlapping the distribution of background temperatures.