Interpolacin
Páginas: 46 (11263 palabras)
Publicado: 1 de abril de 2011
nadaUsing analytical tools when generating surfaces
There are many steps that you go through when creating a surface. In each step, you specify a number of parameters. Geostatistical Analyst provides a series of dialog boxes containing analytical tools to assist you in determining the values for the parameters. Some of these dialog boxes and tools are applicable to almost all interpolationmethods, such as specifying the search neighborhood, cross-validation, and validation. Others are specific to the geostatistical methods (kriging and cokriging), such as modeling semivariograms, transformations, detrending, declustering, and checking for the bivariate normal distribution.
Within each dialog box, there are a series of tasks that can be accomplished using the tools. Depending onyour data, none, some, or all of the tasks and their parameters might be explored. As with all parameters, Geostatistical Analyst provides reliable defaults, some of which have been calculated specifically for your data. However, you may have additional insight into your data from prior knowledge of the phenomena under study or that you gained through the exploratory tools provided withGeostatistical Analyst, which you can use to refine the parameters to create a more accurate surface.
About determining neighborhood search size
Things that are closer to one another are more alike—as the data locations become farther away from a location where the value is unknown, they may not be as useful when predicting the value at an unmeasured location. At some distance, the points will have nocorrelation with the prediction location, and they may even be located in an area much different than the unknown location. Therefore, it is common practice to specify a search neighborhood that limits the number and configuration of the points that will be used in the predictions.
There are two controlling mechanisms to limit the points used, namely specifying the shape of the neighborhood andestablishing constraints on the points within the shape.
Determining neighborhood search size gives you task-based information.
Specifying the shape of the neighborhood
The shape of the search neighborhood will be dictated by the input data. For instance, if your data is evenly sampled and has no directional autocorrelation (isotropy), you will want to include points evenly in alldirections from the unknown location. To do so, you will probably want to specify your neighborhood shape to be a circle. However, if you know that there is directional autocorrelation (anisotropy), such as is caused by wind pollutants, you may want to use an elliptical search neighborhood scheme with the major axis parallel to the wind when interpolating a surface. You will do so because you know thatpoints upwind and downwind from the prediction location are more like the prediction location's value at longer distances than those perpendicular to the wind's direction.
The search neighborhood shape should be based on an understanding of the spatial locations and the spatial autocorrelation of the dataset. Understanding the spatial locations and autocorrelations is done through the ESDA toolsand with the tools within the Geostatistical Analyst Wizard.
The search neighborhood in this example can be defined as an ellipse with semimajor and semiminor axes of 200 km and 80 km, respectively, and with a rotation angle of 340°.
[pic]
The Searching Neighborhood dialog box allows you to define the length of the semimajor and semiminor axes and the direction of the semimajor axis. Acircle will have equal lengths for semimajor and semiminor axes.
Establishing constraints
Once the shape is defined, the second mechanism for controlling the neighborhood involves establishing constraints within the shape. First, the number of points to be used within the neighborhood for the predictions must be defined. Geostatistical Analyst allows you to select the desirable and minimum...
There are many steps that you go through when creating a surface. In each step, you specify a number of parameters. Geostatistical Analyst provides a series of dialog boxes containing analytical tools to assist you in determining the values for the parameters. Some of these dialog boxes and tools are applicable to almost all interpolationmethods, such as specifying the search neighborhood, cross-validation, and validation. Others are specific to the geostatistical methods (kriging and cokriging), such as modeling semivariograms, transformations, detrending, declustering, and checking for the bivariate normal distribution.
Within each dialog box, there are a series of tasks that can be accomplished using the tools. Depending onyour data, none, some, or all of the tasks and their parameters might be explored. As with all parameters, Geostatistical Analyst provides reliable defaults, some of which have been calculated specifically for your data. However, you may have additional insight into your data from prior knowledge of the phenomena under study or that you gained through the exploratory tools provided withGeostatistical Analyst, which you can use to refine the parameters to create a more accurate surface.
About determining neighborhood search size
Things that are closer to one another are more alike—as the data locations become farther away from a location where the value is unknown, they may not be as useful when predicting the value at an unmeasured location. At some distance, the points will have nocorrelation with the prediction location, and they may even be located in an area much different than the unknown location. Therefore, it is common practice to specify a search neighborhood that limits the number and configuration of the points that will be used in the predictions.
There are two controlling mechanisms to limit the points used, namely specifying the shape of the neighborhood andestablishing constraints on the points within the shape.
Determining neighborhood search size gives you task-based information.
Specifying the shape of the neighborhood
The shape of the search neighborhood will be dictated by the input data. For instance, if your data is evenly sampled and has no directional autocorrelation (isotropy), you will want to include points evenly in alldirections from the unknown location. To do so, you will probably want to specify your neighborhood shape to be a circle. However, if you know that there is directional autocorrelation (anisotropy), such as is caused by wind pollutants, you may want to use an elliptical search neighborhood scheme with the major axis parallel to the wind when interpolating a surface. You will do so because you know thatpoints upwind and downwind from the prediction location are more like the prediction location's value at longer distances than those perpendicular to the wind's direction.
The search neighborhood shape should be based on an understanding of the spatial locations and the spatial autocorrelation of the dataset. Understanding the spatial locations and autocorrelations is done through the ESDA toolsand with the tools within the Geostatistical Analyst Wizard.
The search neighborhood in this example can be defined as an ellipse with semimajor and semiminor axes of 200 km and 80 km, respectively, and with a rotation angle of 340°.
[pic]
The Searching Neighborhood dialog box allows you to define the length of the semimajor and semiminor axes and the direction of the semimajor axis. Acircle will have equal lengths for semimajor and semiminor axes.
Establishing constraints
Once the shape is defined, the second mechanism for controlling the neighborhood involves establishing constraints within the shape. First, the number of points to be used within the neighborhood for the predictions must be defined. Geostatistical Analyst allows you to select the desirable and minimum...
Leer documento completo
Regístrate para leer el documento completo.