Optimal Mapping When Datasets are Massive
Abstract
Maps are an extremely important part of any military operation and producing timely and accurate maps is an essential planning tool. Ocean-floor or terrain data are static and spatial, while meteorological or visibility data are dynamic and spatio-temporal. Data upon which maps are based can be simultaneously massive and sparse, and they are noisy. In the presence of uncertainty due to missing data and measurement-error noise, spatial and spatio-temporal statistical analysis of massive data sets is challenging. The massiveness causes problems in computing optimal spatial predictors, such as kriging, since one has to solve (and store) systems of equations equal to the size of the data. In addition, a large spatial domain is often associated with nonstationary behavior over that domain. The objectives of this study are as follows: (1) construct a flexible family of nonstationary covariance functions using a truncated set of basis functions, fixed in number; (2) develop the necessary methodology and algorithms for covariance-parameter estimation; (3) derive optimal spatial or spatio-temporal maps that account for uncertainties statistically; and (4) incorporate spatial and spatio-temporal dependencies into the analysis of sensor-network data.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 25, 2008
- Accession Number
- ADA478900
Entities
People
- Noel A. Cressie
Organizations
- Ohio State University