Evaluation of Generalized Thresholds in an Objective Multispectral Satellite Cloud Analysis
Abstract
An objective algorithm is developed and evaluated for classifying 11 cloud types. The multispectral technique uses daytime images of AVHRR channel 1 (0.68 microns), channel 4 (10.8 microns), and channel 5 (12.0 microns). Visual reflectance is scaled by the solar zenith angle in order to prepare standardized albedos. Infrared brightness temperatures provide cloud height information through comparison with a representative temperature sounding. Channel 4-5 temperature differences (split-window) distinguish between ice clouds with variable emissivities and thick precipitating clouds. A statistical texture analysis on the standardized albedos separates stratiform from cumuliform clouds. Generalized thresholds were set for each of these parameters to discriminate between cloud types. The evaluation judges model performance based on four case studies located at different latitudes with varying solar zenith angles. Overall results compiled between the cases show a 67% agreement for manual and model classifications. General results compiled for clear, low clouds, middle clouds, high clouds, and precipitation cloud types were even more successful with 94%, 90%, 50%, 87%, and 95% agreement, respectively. The color enhanced automated product provides a legible, quick and accurate tool for cloud type analysis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1990
- Accession Number
- ADA232716
Entities
People
- Thomas J. Neu
Organizations
- Naval Postgraduate School