The KRIGE2D Procedure

PROC KRIGE2D Statement

  • PROC KRIGE2D options;

The PROC KRIGE2D statement invokes the KRIGE2D procedure. Table 55.2 summarizes the options available in the PROC KRIGE2D statement.

Table 55.2: PROC KRIGE2D Statement Options

Option

Description

DATA=

Specifies input data set

IDGLOBAL

Uses ascending observation numbers as observation labels across BY groups

IDNUM

Uses observation number as observation labels

NOPRINT

Suppresses the normal display of results

OUTEST=

Writes kriging predictions and standard errors

OUTNBHD=

Writes neighborhood information for each grid point

PLOTS

Specifies plot display and options

PREDICTION

Produces the kriging prediction plot

SEMIVARIOGRAM

Produces the semivariogram used for the kriging prediction

SINGULARMSG=

Controls the number of warning messages displayed for a singular matrix


You can specify the following options in the PROC KRIGE2D statement.

DATA=SAS-data-set

specifies a SAS data set that contains the x and y coordinate variables and the VAR= variables in the PREDICT statement.

IDGLOBAL

specifies that ascending observation numbers be used across BY groups for the observation labels in the appropriate output data sets and the OBSERVATIONS plot, instead of resetting the observation number in the beginning of each BY group. The IDGLOBAL option is ignored if no BY variables are specified. Also, if you specify the ID statement, then the IDGLOBAL option is ignored unless you also specify the IDNUM option in the PROC KRIGE2D statement.

IDNUM

specifies that the observation number be used for the observation labels in the appropriate output data sets and the OBSERVATIONS plot. The IDNUM option takes effect when you specify the ID statement; otherwise, it is ignored.

NOPRINT

suppresses the normal display of results. The NOPRINT option is useful when you want only to create one or more output data sets with the procedure. Note: This option temporarily disables the Output Delivery System (ODS); see the section ODS Graphics for more information.

OUTEST=SAS-data-set
OUTE=SAS-data-set

specifies a SAS data set in which to store the kriging predictions, standard errors, and grid location. For details, see the section OUTEST=SAS-data-set.

OUTNBHD=SAS-data-set
OUTN=SAS-data-set

specifies a SAS data set in which to store the neighborhood information for each grid point. Information is written to this data set only if one or more PREDICT statements have options that specify local kriging. For details, see the section OUTNBHD=SAS-data-set.

PLOTS <(global-plot-option)> <= plot-request <(options)>>
PLOTS <(global-plot-option)> <= (plot-request <(options)> <... plot-request <(options)>>)>

controls the plots produced through ODS Graphics. When you specify only one plot request, you can omit the parentheses around the plot request. Here are some examples:

plots=none
plots=observ
plots=(observ(outl) prediction)
plots=(prediction(fill=pred line=se obs=grad) prediction(fill=se))

ODS Graphics must be enabled before plots can be requested. For example:

ods graphics on;

proc krige2d data=thick;
   coordinates xc=East yc=North;
   predict var=thick r=60;
   model scale=7.4599 range=30.1111 form=gauss;
   grid x=0 to 100 by 10 y=0 to 100 by 10;
run;

ods graphics off;

For more information about enabling and disabling ODS Graphics, see the section Enabling and Disabling ODS Graphics in Chapter 21: Statistical Graphics Using ODS.

If ODS Graphics is enabled but you omitted the PLOTS option or have specified PLOTS=ALL, then PROC KRIGE2D produces a default plot for each MODEL statement of every PREDICT statement that you specify. The default PROC KRIGE2D plot displays a contour plot of the kriging prediction and the gradient of the kriging prediction standard error at every location of the prediction grid, in addition to empty circles that indicate the observation locations. See Figure 55.4 for an example of the default KRIGE2D plot.

The following global-plot-option is available:

ONLY

suppresses the default plot. Only plots that are specifically requested are displayed.

The following individual plot-requests and plot options are available:

ALL

produces all appropriate plots. You can specify other options with ALL. For example, to request the default plot and an additional plot of the predictions, specify PLOTS=(ALL PREDICTION).

EQUATE

specifies that all appropriate plots be produced in a way that the axes coordinates have equal size units.

NONE

suppresses all plots.

OBSERVATIONS <(observations-plot-options)>
OBSERV <(observations-plot-options)>
OBS <(observations-plot-options)>

produces the observed data plot. Only one observations plot is created if you specify the OBSERVATIONS option more than once within a PLOTS option.

The OBSERVATIONS option has the following suboptions:

GRADIENT

specifies that observations be displayed as circles colored by the observed measurement.

LABEL < ( label-option ) >

labels the observations. The label is the ID variable if the ID statement is specified; otherwise, it is the observation number. The label-option can be one of the following:

EQ=number

specifies that labels show for any observation whose value is equal to the specified number.

MAX=number

specifies that labels show for observations with values smaller than or equal to the specified number.

MIN=number

specifies that labels show for observations with values equal to or greater than the specified number.

If you specify multiple instances of the OBSERVATIONS option and you specify the LABEL suboption in any of those, then the resulting observations plot displays the observations labels. If more than one label-option is specified in multiple LABEL suboptions, then the prevailing label-option in the resulting OBSERVATIONS plot emerges by adhering to the choosing order: MIN, MAX, EQ.

OUTLINE

specifies that observations be displayed as circles with a border but with a completely transparent fill.

OUTLINEGRADIENT

is the same as OBSERVATIONS(GRADIENT) except that a border is shown around each observation.

SHOWMISSING

specifies that observations with missing values be displayed in addition to the observations with nonmissing values. By default, missing values locations are not shown on the plot. If you specify multiple instances of the OBSERVATIONS option and you specify the SHOWMISSING suboption in any of those, then the resulting observations plot displays the observations with missing values.

If you omit any of the GRADIENT, OUTLINE, and OUTLINEGRADIENT suboptions, the OUTLINEGRADIENT is the default suboption. If you specify multiple instances of the OBSERVATIONS option or multiple suboptions for OBSERVATIONS, then the resulting observations plot honors the last specified GRADIENT, OUTLINE, or OUTLINEGRADIENT suboption.

PREDICTION <(prediction-plot-options)>
PRED <(prediction-plot-options)>

specifies that the kriging prediction plot be produced. You can specify the PREDICTION option multiple times in the same PLOTS option to request instances of plots with the following prediction-plot-options:

ALPHA=number

specifies a parameter to obtain the confidence level for constructing confidence limits based on the prediction standard error. The value of number must be between 0 and 1, and the confidence level is $1-$number. The default is ALPHA=0.05; this corresponds to the confidence level of 95%, or about 1.96 times the prediction standard error. The ALPHA= suboption is used only for prediction plots in one dimension, and it is incompatible with the FILL and LINE suboptions.

CLONLY

specifies that only the confidence limits be shown in a prediction plot without the predicted values. This suboption can be useful for identifying confidence limits when the prediction standard error is small at the prediction locations. CLONLY is used only for prediction plots in one dimension, and it is incompatible with the FILL and LINE suboptions.

CONNP

specifies that grid points that you provide as individual prediction locations be connected with a line on the area map. This suboption is ignored when you have a single grid point, a prediction grid in two dimensions, or when you also specify the NOMAP suboption. The CONNP suboption is incompatible with the FILL and LINE suboptions.

FILL=NONE | PRED | SE

produces a surface plot for either the predicted values or the standard errors. FILL=SE is the default. However, if you omit the FILL suboption, the behavior depends on the LINE suboption as follows: If you specify LINE=NONE or entirely omit the LINE suboption, then the FILL suboption is set to its default value. If LINE=PRED or LINE=SE, then the FILL suboption is set to the same value as the LINE suboption.

LINE=NONE | PRED | SE

produces a contour line plot for either the predicted values or the standard errors. LINE=PRED is the default. However, if you omit the LINE suboption the behavior depends on the FILL suboption as follows: If you specify FILL=NONE or entirely omit the FILL suboption, then the LINE suboption is set to its default value. If FILL=PRED or FILL=SE, then the LINE suboption is set to the same value as the FILL suboption.

NOMAP

specifies that the prediction plot be produced without a map of the domain where you have observations. The NOMAP suboption is used in the case of prediction in one dimension or at individual points. It is incompatible with the FILL and LINE suboptions.

OBS=obs-options

produces an overlaid scatter plot of the observations in addition to the specified contour plots. The following obs-options are available:

GRAD

specifies that observations be displayed as circles colored by the observed measurement. The same color gradient displays the prediction surface and the observations. Observations where the prediction is close to the observed values have similar colors—the greater the contrast between the color of an observation and the surface, the larger the prediction standard error is at that point.

LINEGRAD

is the same as OBS=GRAD except that a border is shown around each observation. This option is useful for identifying the location of observations where the standard errors are small, because at these points the color of the observations and the color of the surface are indistinguishable.

NONE

specifies that no observations be displayed.

OUTL

specifies that observations be displayed as circles with a border but with a completely transparent fill.

OBS=NONE is the default when you specify a grid in two dimensions, and OBS=LINEGRAD is the default used in the area map when you have a grid in one dimension. However, the default PROC KRIGE2D plot for a surface grid displays the observations locations as outlines.

SHOWD

specifies that the horizontal axis in scatter plots of linear prediction grids show the distance between grid points instead of the grid points’ coordinates. When the area map is displayed, the prediction locations are also connected with a line. In all other grid configurations the SHOWD suboption is ignored, and it is incompatible with the FILL and LINE suboptions.

SHOWP

specifies that the grid points in band plots of linear prediction grids be shown as marks on the band plot. In all other grid configurations the SHOWP suboption is ignored, and it is incompatible with the FILL and LINE suboptions.

TYPE=BAND | SCAT

requests a particular type of plot when you have a linear grid, regardless of the default PREDICTION plot behavior in this case. The TYPE suboption is incompatible with the FILL and LINE suboptions.

If you specify multiple instances of the ALPHA, FILL, LINE, OBS, or TYPE suboptions in the same PREDICTION option, then the resulting predictions plot honors the last value specified for any of the suboptions. Any combination where you specify FILL=NONE and LINE=NONE is not available. When the prediction grid is in two dimensions, only the FILL, LINE, and OBS suboptions apply. If you specify incompatible suboptions in the same PREDICTION plot, then the plot instance is skipped.

The PREDICTION option produces a surface or contour line plot for grids in two dimensions and a band plot or scatter plot with error bars for grids in one dimension or individual points. In two dimensions the plot illustrates the predicted values and prediction error at each grid point. By default, when you specify a linear grid with fewer than 10 points, PROC KRIGE2D produces a PREDICTION scatter plot for each one of the prediction grid points. For 10 or more points in a linear grid, the PREDICTION plot is a band plot of the predicted means and the confidence limits at the 95% confidence level. You can override the default behavior in linear grids with the TYPE suboption. Prediction at individual locations always produces a PREDICTION scatter plot.

In cases of prediction in one dimension or at individual points, an area map is produced that shows the observations and the grid points. Band plots of linear grids display the grid points as a line on the map. When you specify individual prediction locations, the grid points are indicated with marks on the area map. The area map appears on the side of the prediction band plot or scatter plot, unless you specify the NOMAP suboption. You can also label the individual grid points or the ends of linear grid segments with the LABEL option of the GRID statement.

SEMIVARIOGRAM <(semivar-plot-option)>
SEMIVAR <(semivar-plot-option)>

specifies that the semivariogram used for the kriging prediction be produced. You can use the following semivar-plot-option:

MAXD=number

specifies a positive value for the upper limit of the semivariogram horizontal axis of distance. The SEMIVARIOGRAM plot extends by default to a distance that depends on the correlation model range. You can use the MAXD= option to adjust the default maximum distance value for the plot.

The SEMIVARIOGRAM option produces a plot for each correlation model that you specify for your prediction tasks. In an anisotropic case, the plot is not produced if you assign different anisotropy angles for different model components. The only exception is when you specify zonal components at right angles with the nonzonal model components. Also, the SEMIVARIOGRAM option is ignored for models that consist of purely zonal components.

SINGULARMSG=number
SMSG=number

controls the number of warning messages displayed for a singular matrix. When local kriging is performed, a separate kriging system is solved for each grid point. Anytime a singular matrix is encountered, a warning message is displayed up to a total of number times. The default is SINGULARMSG=10.