Skip to main content
ARS Home » Midwest Area » Morris, Minnesota » Soil Management Research » Research » Publications at this Location » Publication #241160

Title: Adapting geostatistics to analyze spatial and temporal trends in weed populations

item COLBACH, NATHALIE - Institut National De La Recherche Agronomique (INRA)
item Forcella, Frank

Submitted to: Book Chapter
Publication Type: Book / Chapter
Publication Acceptance Date: 7/15/2009
Publication Date: 2/22/2011
Citation: Colbach, N., Forcella, F. 2011. Adapting geostatistics to analyze spatial and temporal trends in weed populations. In: Clay, S., editor. GIS Applications in Agriculture: Invasive Species. Vol. 3. New York: CRC Press, Taylor & Francis Group. p. 319-371.

Interpretive Summary:

Technical Abstract: Geostatistics were originally developed in mining to estimate the location, abundance and quality of ore over large areas from soil samples to optimize future mining efforts. Here, some of these methods were adapted to weeds to account for a limited distribution area (i.e., inside a field), variations with time, skewed data distributions, and correlations with species traits. A geostatistical study starts with collecting data and choosing an adequate sampling plan to "catch" the spatial relationships in weed patches to be studied. Exploratory data analysis then looks at data distributions and checks whether the pre-requisites for a geostatistical analysis are fulfilled. If necessary, data are transformed and detrended to meet these pre-requisites. Only then are empirical semivariograms calculated to describe small-scale spatial trends. Fitting models to these data makes possible the estimation of variogram parameters that then were correlated with species traits to explain small-scale spatial trends (e.g., weed patch shapes and sizes over time as a function of species dispersal and germination traits). Semivariogram models are also used to estimate variance for unsampled distances and thus to plot maps with kriging. Cross-semivariograns and co-kriging describe co-variation of variables in space and these relationships are used to estimate a sparsely sampled primary variable with the help of an extensively sampled secondary variable. Here, these methods were adapted to predict weed maps with past observations and variograms. Lastly, error analysis evaluates how close predicted weed maps are to observations. This provides an indication of the risk of spraying insufficiently or unnecessarily when basing herbicide spraying in precision agriculture on weed maps predicted with past observations using cokriging.