Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → États-Unis → 1997 → Geostatistical analysis of precipitation in southwest Saudi Arabia

Colorado State University (1997)

Geostatistical analysis of precipitation in southwest Saudi Arabia

Subyani, Ali M

Titre : Geostatistical analysis of precipitation in southwest Saudi Arabia

Auteur : Subyani, Ali M

Université de soutenance : Colorado State University

Grade : Doctor of Philosophy (PhD) 1997

Résumé
Precipitation is a hydrological phenomena that varies in magnitude in space as well as in time and requires suitable tools in order to predict values in space and time. This study investigated the feasibility of applying the Theory of Regionalized Variables, Univariate and Multivariate Geostatistics, to precipitation in the southwest region of Saudi Arabia. The major goals of this work were : (1) to analyze and model the spatial variability of precipitation, (2) to interpolate kriging maps at each grid point for different seasons, (3) to develop new graphical methods to describe the structural continuity of precipitation, (4) to analyze and model the structural cross-correlation of precipitation and elevation for different seasons, (5) to investigate whether cokriging would improve the accuracy of precipitation estimates by including elevation as a secondary variable, and (6) to compare prediction errors and prediction variances with those of kriging and cokriging methods for different seasons. The results from the spatial structural models of seasonal precipitation showed that : (1) the exponential structures with a nugget effect fit to winter and spring data, and (2) the spherical structures with no nugget effect fit to summer and fall data. Two new methods were proposed to describe the spatiotemporal structural continuity of precipitation. Method 1, called Combined-Variogram, attempted to combine different seasonal variograms by using the seasonal correlations. The results from testing the historical and synthetic generated data for lag-zero cross-correlation show that there is spurious cross-correlation between two uncorrelated processes, and we can not go farther to combine different seasonal variograms. Method 2, called Distance-Time, was developed to describe graphically the seasonal structural continuity variation of precipitation. This method reflects the climate factor affecting precipitation in the area, and from the variogram contour lines, the degree of the structural continuity of precipitation can be visualized. The cross-spatial structural models and cokriging method were applied to improve the accuracy of precipitation estimates by including elevation as a secondary variable. The study shows that the cross-variogram models for winter and spring seasons show more consistence with the spherical model, higher range, and absence of a nugget effect, compared with variogram models of the same seasons, due to highly significant correlation of precipitation with elevation. The summer cross-variogram is less consistent, with exponential model and lower range, due to no correlation of precipitation with elevation of the same season. The results of the cokriging method give a slightly more uniform distribution of precipitation than kriging in all seasons except fall, where no obvious cross-variogram models were evident. Cokriging estimates were slightly reduced, relative to kriging estimates. However, cokriging estimation variances were reduced by 25%, relative to kriging in all seasons. The most significant reduction was found for the spring season.

Mots clés : Cokriging, Statistics, kriging, Hydrology, Atmosphere, Pure sciences (more...) Earth sciences

Search Oxford Libraries On Line (SOLO)

ProQuest : Dissertations & Theses

Page publiée le 24 février 2015, mise à jour le 18 novembre 2018