|Bourgault, G - STANFORD UNIVERSITY|
|Journel, A - STANFORD UNIVERSITY|
|Lesch, S - UCR, RIVERSIDE, CA|
Submitted to: Advances in Soil Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: August 8, 1996
Publication Date: N/A
Interpretive Summary: Most individuals think of soil as being homogeneous, but actually it is very heterogeneous. When the physical and chemical properties of soil are characterized, it is found that they are spatially variable, which is to say that the properties of soil vary considerably from one point to the next even within as small a distance as a few inches. The spatial complexity of soil makes it a difficult media to model, particularly if one is trying to model non-point source (NPS) pollutants in soil. NPS pollutants are pollutants such as fertilizers, pesticides, salinity, trace elements, etc. which are spread over broad areas of land and pose a global threat due to their ubiquitous nature and due to their chronic health effects. A branch of statistics known as geostatistics when used in combination with a database of spatial information known as a geographics information system (GIS) offers a unique means of analyzing spatial relations between the diverse data used in modeling NPS pollutants. This report presents a step-by-step case study of geostatistics applied to a soil salinity data handled through a GIS of multivariate, spatially-distributed data where sparse hard (reliable) data coexist with abundant but softer (less reliable) information. This approach provides a potential tool useful for modeling NPS pollutants.
Technical Abstract: In the broadest sense, geographical information systems (GIS) and geostatistics pursue a similar objective, that of analysis and integration of diverse available information in order to build maps which summarize and expand the original data. Geostatisitics provides tools and algorithms for modeling spatial relations between diverse data. Then, these models are used for improved mapping yet retaining a measure of uncertainty of the resulting maps. This report presents a step-by-step case study of geostatistics applied to a soil salinity data set typical of data sets handled through GIS, i.e., multivariate, spatially distrubted where sparse hard data coexist with abundant but softer (less reliable) information. Rather than addressing the specific soil salinity problem, this report aims at making a case for adding geostatistics and uncertainty into the GIS toolbox. Correct analysis and efficient utilization of geographical data requires that the space/time dimensions of these data be accounded for explicitly.