Date of Award
Master of Science
Department of Engineering Physics
Peter A. Saunders PhD
Joint histograms of cloud top height (CTH) and optical depth (OD) are created using the World-Wide Merged Cloud Analysis (WWMCA) dataset over a four year period (2014-2017) to identify average cloud field regimes and assess the application of utilizing the WWMCA dataset with the AFIT Sensor and Scene Emulation Tool (ASSET). Two selected regions encompassing the Florida peninsula and a portion of the Pacific Ocean off the west-central coast of South America are examined over the months of January and July. Cloud field regimes are identified by running generated hourly OD-CTH histograms through k-means clustering, with optimal cluster number ( K ) evaluation performed by calculating and comparing silhouette scores and heuristic elbow method results. Varying cluster groupings are plotted out to distinguish discrepancies between these multi-cluster analysis. Initial results indicate K = 3 as the optimal number of clusters to use, generating three major cloud field regimes unique to each region with high relative frequency of occurrences (RFO). Notable departures from silhouette score calculations to cluster evaluation call into question the validity of silhouette score usage to determine optimal K values, which is discussed alongside future improvements and applications of the cloud field regimes identified.
DTIC Accession Number
Almeida, Stewart G., "Identifying Four Year Average Cloud Field Regimes from World Wide Merged Cloud Analysis Dataset By Way of K-Means Clustering" (2021). Theses and Dissertations. 4914.