"Everything should be made as simple as possible, but not simpler." - Albert Einstein

Seminar Announcement

Spatial Analysis of Soil Infiltration Data  at the Central Plains Experimental Range (CPER) in Northern Colorado

Ahmed Mohamed, M.S. Candidate, Department of Statistics, Colorado State University.

Thursday, November 20, 2008

2:00 p.m. 223 Weber

ABSTRACT

Abstract

Soil water infiltration data collected at the Central Plains Experimental Range (CPER) were analyzed using spatial correlation models.  Data were collected on four experimental plots that were grided into 1 m2 cells. Plots A1 and A2 were 30 m2 (3 m×10 m) each, and plot B1 and B2 were 50 m2 (105 m ) each. Soil physical properties and vegetation cover are important factors in determining soil infiltration. The infiltration data as well as soil bulk density (BD), soil moisture, and vegetation cover data were collected from each cell in the experiment. A double-ring infiltrometer was used to measure cell infiltration rate. A core sampler of known volume was used to collect cell soil sample from which soil BD and soil moisture were measured at two depths each. The vegetation cover was estimated from the digital image of the cell. The soil BD, soil moisture, and vegetation cover measurements were incorporated in the infiltration spatial analysis as covariates.

Spatial models with covariates were compared to no-covariate spatial models.  For each spatial model, three spatial covariance functions were compared: the spherical, the exponential, and the Gaussian covariance function. The semivariogram function was also used to evaluate the infiltration spatial model. Due to the apparent edge effects on the infiltration surfaces of the plots, the analysis was performed on three data sets; the A1 and A2 data set, the B1 and B2 data set, and the combined plots data set. The infiltration spatial analysis of the three data sets supported the use of the covariate spherical model without nugget effect. The estimated variograms of the three data sets showed a downward turn of the function at distances beyond the apparent sill (the variance of the spatial process), which contradicts the theoretical behavior of a variogram function. The drop of the variogram function may be explained by plot infiltration edge effects. The infiltration spatial analysis of the three data sets confirms the existence of a strong spatial correlation.

Advisory Committee:

Phil Chapman, Adviser

Myung-Hee Lee, Committee Member

Kenneth Berry (Sociology)