Publication Date

9-1-2004

Degree Program

Department of Geography and Geology

Degree Type

Master of Geoscience

Abstract

Spatial variability of precipitation is examined over the state of Kentucky and surrounding areas. The study focuses on the analysis of monthly precipitation totals from the period of 1961-2000. The purpose of the study is to develop a set of indices to represent the spatial variability of the study area for a given month. Various exploratory data analysis methods such as variography, kriging, and cluster analysis were used. The study attempts to quantify the second order (local) effects of the spatial variation of precipitation as a means to provide insight into the prediction of precipitation randomness. This task can be a difficult one due to the distinction between first and second order effects being somewhat arbitrary. The study proposes that a qualitative map of mean monthly precipitation can be classified through the use of a quantitative measure. This approach allowed for the unique classification of numerous months of precipitation through the use of a standard methodology. The researcher found that trying to capture the spatial variation of precipitation with two indices is an arduous task. Months were classified based on percentiles of the variogram cloud. The data were condensed into distance bins for analysis and used for calculation of the indices. A Short Range Index (SRI) and Long Range Index (LRI) were calculated for each month. The indices were then analyzed through the use of cluster analysis. The Partioning Around Medoids (PAM) method was used for the analysis providing an average silhouette value of .32. The study found that the methods applied did not efficiently capture the spatial variability of precipitation across the study area. However, this study has provided insight into the methodologies that can be applied to investigate spatial patterns of precipitation.

Disciplines

Atmospheric Sciences | Climate | Geography | Meteorology