Climate Classification Preprocessing Algorithm.

[摘要] ABSTRACT Remote sensing has numerous applications in the fields of defense, medical imaging and environmental research.The Climate Classification Preprocessing algorithm (CPA) combines the areas of remote sensing, data mining and climate research in the development of a neural fuzzy-logic algorithm. CPA identifies atmospheric targets utilizing data from the Atmospheric Radiation and Measurement Program (ARM) Tropical Western Pacific site (TWP) without using computational global model data. CPA is a true multi-sensor data mining algorithm that provides key preprocessing figures and graphs necessary to climate research.This research explores multi-sensor responses to random and/or complex atmospheric targets.Through the utilization of CPA, the classification of atmospheric target properties at the ARM observation sites in the TWP is performed through the process of data mining. Data mining is the process of finding patterns or relationships in large data sets.CPA provides and/or identifies reflectivity, velocity and spectral width diagrams, Contour Frequency by Altitude Diagrams (M-CFAD), condensed water content, and mapping of the data on the same time height grid as the radar by utilization of multiple instruments, which results in the graphical classification of atmospheric targets at the ARM TWP site. Consequently, a fully robust classifier for targets based on several different instruments at the TWP site is achieved with an error rate between 3-10 %. CPA is the first step in achieving one of ARM;;s long term goals for the program.Since CPA is a robust atmospheric phase classifier, the algorithm has possible applications such as utilization of the algorithm for possible tracking of insects particularly mosquitoes to track the spread of West Nile Virus and Malaria. Future Research in this area includes the possibility of creating a true multi-retrieval algorithm with CPA at the core of the multi-retrieval algorithm, as well as altering CPA to be a genetic optimization algorithm.

[发布机构] University of Michigan

[效力级别] Data Mining

[关键词] Atmospheric Phase Classifier;Data Mining;Atmosphere Target Identification;Physics;Science;Applied Physics