8. Batch Plotting Module

The batch plotting system is used to generate a number of plots of data in the METviewer database. The usage statement:

---- MVBatch ----

Usage: mv_batch

       [-list]
       [-printSql]
       plot_spec_file
       [job_name]

       where "-list" indicates that the available plot jobs should be listed and no plots run
       "-printSql" print SQL statements
       "plot_spec_file" specifies the XML plot specification document
       "job_name" specifies the name of the job from the plot specification to run

---- MVBatch Done ----

The plot_spec_file passed to the batch module contains information about the plots that will be generated. It is an XML file and its top-level tag is <plot_spec> which contains the following elements, divided into functional sections.

NOTE: All <val> elements should be listed in alphabetical order.

<connection>: Please reference the Common XML Structures documentation.

<rscript>: (optional) This tag specifies the Rscript instance to use, for example /usr/local/bin/Rscript. It defaults to the instance on the user path.

<folders>: A structure that specifies the input and output file system folders.

<r_tmpl>: The folder that contains the R templates used by METviewer, typically R_tmpl/ from the METviewer CVS source tree.
<r_work>: The folder that will contain the data and generated R scripts, must contain the sub-folder include/
<plots>: The output folder that will contain the generated plots.

<date_list>: Please reference the Common XML Structures documentation.

<date_range>: A structure that specifies a date range, including all dates and times between.

<start>: The start date and time of the date list, specified in the format yyyy-MM-dd [HH:mm:ss], or with a <date_offset> structure. See Common XML Structures.
<end>: The end date and time of the date list, specified in the format yyyy-MM-dd [HH:mm:ss], or with a <date_offset> structure. See Common XML Structures.

<date_range_list>: A structure that specifies a list of date ranges, conceptually, a list of <date_range> structures.

<range_start>: The start date and time of the first date range of the list, specified in the format yyyy-MM-dd [HH:mm:ss], or with a <date_offset> structure. See Common XML Structures
<range_end>: The start date and time of the last date range of the list, specified in the format yyyy-MM-dd [HH:mm:ss], or with a <date_offset> structure. See Common XML Structures
<range_length>: The range length, in minutes.
<inc>: The increment, in minutes, between the end of the previous range member and the start of the next.

<plot>: a structure that specifies plot characteristics and may inherit characteristics from other <plot> structures

Attributes:
name: Name for the plot structure, used when plot’s characteristics are inherited by later <plot> structures.

inherits: Name of one or more <plot> structure’s characteristics to inherit, separated by commas (no spaces). Plots named later on the list override the characteristics of those earlier on the list.
run: (optional) TRUE or FALSE, specifying whether or not to run this plot job.

<template>: The name of the R plot script template to use. The following templates are currently supported:

series_plot.R_tmpl: Use for time series, threshold series, date series, and other series type plots which display a fcst_var and statistic vs. A stat_header quantity.
bar_plot.R_tmpl: Use for bar plots of fcst_var/statistic vs. stat_header quantities.
box_plot.R_tmpl: Use for box plots of fcst_var/statistic vs. stat_header quantities.
rhist.R_tmpl: Use for ensemble rank histograms, only fixed data must be specified.
roc.R_tmpl: Use for ROC plots, only fixed data and PCT/CTC data must be specified.
rely.R_tmpl: Use for reliability diagrams, only fixed data must be specified.
ens_ss.R_tmpl: Use ensemble spread/skill diagrams, fixed data and series must be specified.

<indep>: A structure that defines the independent variable and its values for the plot.
Attributes:
name: Database field name of the independent variable.
depends: The field name of one of the plot_fix field structures; will be used to calculate independent variable values.

<val>: A single value.
Attributes:
label: (optional) x-axis label.
plot_val: (optional) x-axis position.

<date_list>: A structure that specifies a list of date strings in a certain format.
<start>: The start date and time of the date list, specified in the format yyyy-MM-dd HH:mm:ss.
<end>: The end date and time, specified in the format yyyy-MM-dd HH:mm:ss.
<inc>: The increment, in seconds, between successive members of the date list.
<label_format>: The date format, specified by the java class SimpleDateFormat.

<series1>: Describes the database field names and values that constitute the various lines, or series, corresponding to the left y axis (y1-axis) of the plot.
<field>: A database field whose values constitute the members of the series.
Attributes:
name: The database field name. Its values are contained in the child tags.

<val>: A single value.

<series2>: Describes the database field names and values that constitute the various lines, or series, corresponding to the right y axis (y2-axis) of the plot. The structure is the same as <series1> above.

<series_nobs>: Indicates which database fields should be used to calculate the number of observations (NOBS) for the plot.

<dep1_scale> or <dep2_scale>: Specifies a scale factor for a fcst_var/stat pair, for example to convert the units of a statistic.
<field>: Contains the value of the scalar to apply.
Attributes:
name: The fcst_var/stat name pair in the format [fcst_var]_[stat].

<plot_fix>: Describes the database fields and values that are constant on a given plot, or fixed. Each permutation of plot_fix field values will be used to create a database query and all plots will be run with the resulting data held constant. The fields can be used in the tmpl values, described below, to create paths, filenames and labels for the plots.
<field>: A database field. Its values constitute the members of the series.
Attributes:
name: The database field name. Its values are contained in the child tag.

<val>: A single value.
<date_list>: Refers to a previously declared <date_list>, using the name attribute. See Common XML Structures.
<date_range>: Refers to a previously declared <date_range>, using the name attribute. See Common XML Structures.
<date_range_list>: Refers to a previously declared <date_range_list>, using the name attribute. See Common XML Structures.
<set>: A set of values.
Attributes:
name: (optional) Name of the set, used in the field template.

A set can contain one or more of the following:
<val>: A single value.
<date_list>: Refers to a previously declared <date_list>, using the name attribute. See Common XML Structures.

<plot_cond>: Contains a SQL clause that will be inserted with the clause generated by the <plot_fix> information, allowing a user to specify record conditions for plotted statistics. Examples include:

<plot_cond>ld.baser > 0.02</plot_cond>

<agg>: This tag has been replaced by <plot_fix>. Do not use <agg>.

<tmpl>:
<val_map>: A map of key/value pairs used for when templates are populated.
Attributes:
name: The database field name. Its values should be mapped using the enclosed pairs.

<pair>: A single value pair.
<key>: The template field value as taken from the database.
<val>: The value to use as a replacement.

<data_file>: Path and name of the R data.frame data file, inside of the <r_work> folder specified above.
<plot_file>: Path and name of the output plot image file, inside of the <plots> folder specified above.
<r_file>: Path and name of the R script that generates the plot, inside of the <r_work> folder specified above.
<title>: Title that will be placed on top of the plot.
<x_label>: x-axis label that will be placed on the plot.
<y1_label>: Left y axis (y1-axis) label that will be placed on the plot.
<y2_label>: Right y axis (y2-axis) label that will be placed on the plot.
<caption>: Caption that will be placed on the bottom of the plot.
<listDiffSeries1>: Turns on a difference curve should be plotted for the y1 series; calculated as the pairwise difference between the members of the first series minus the second - example:list(c(“series1”,”series2”), c(“series3”,”series4”)).
<listDiffSeries2>: Turns on a difference curve should be plotted for the y2 series; calculated as the pairwise difference between the members of the first series minus the second - example:list(c(“series1”,”series2”), c(“series3”,”series4”)).

Template Formatting

Template values have the syntax {var_name[?param1=val1[&amp;param2=val2[…]]]}. For example, {fcst_lead?format=HH} or {vx_mask?map=masks} or {fcst_thresh?units=in&amp;symbol=letters}. The var_name must be one of the database fields that is used in the plot fixed variables or agg variables. When the template is applied, the current plot values for each variable replaces the tag in the template. The parameters can be used to format, convert or map the values. The dependent variables forecast variables are represented using the syntax {depY_N}, where Y is the y-axis (either 1 or 2) and N is the index of the fcst_var in the <dep> structure, starting with 1. The dependent variable statistics are represented using the syntax {depY_N_statS}, where Y and N match the values of the parent fcst_var, and S is the index of the statistic, starting at 1 for each fcst_var.

map=[name]: Indicates that the template <val_map> structure with the given name should be applied
format=[format_type]: Depending on the variable being formatted the following options apply:
fcst_lead: Use HH for the two-character wide hour, HHmm for the four-character hour and minute
init_hour or valid_hour: Use HH for the two-character wide hour init_hour.
fcst_thresh, fcst_thr, obs_thresh or obs_thr: Apply the specified decimal format to the threshold value, according to the formatting syntax for the java class DecimalFormat.
For any variable, format=R will replace all non-variable characters (like <, = and .) with appropriate letters and underscores.
units=in: Can only be used if the variable is a threshold, converts the threshold value to inches from mm.
symbol=letters: Can only be used if the variable is a threshold, converts the threshold criteria to FORTRAN-like letters, e.g. >= becomes ge.

<dep>: Information about the dependent variables of the plot. Each curve of the plot is assumed to have a constant fcst_var and constant statistic. The stats plotted on the two dependent axes (y1 and y2) are specified using the identical dep1 and dep2 structures. Fixed values, which correspond to each fcst_var, are specified using the fix structure. The MODE statistics documentation can be viewed below on this page.
<dep1>: The fcst_var/stat pairs for the first dependent axis (y1).
<fcst_var>: The fcst_var for which the enclosed list of stats will be plotted.
Attributes:
name: The fcst_var whose statistics will be plotted.

<stat>: The name of the stat to plot. Its name must be identical to the name in the MET User’s Guide.

<dep2>: The fcst_var/stat pairs for the second dependent axis (y2). Its structure is identical to that of <dep1>
<fix>: Do not use <fix>, use <plot_fix> instead.

<agg_stat>: Contains information about how to calculate and display aggregated statistics. If present, agg_stat is turned on. Depending on the selected <dep> statistics, only certain settings are appropriate.
<agg_ctc>: TRUE or FALSE, indicating whether or not to aggregate CTCs.
<agg_sl1l2>: TRUE or FALSE, indicating whether or not to aggregate SL1L2s.
<agg_sal1l2>: TRUE or FALSE, indicating whether or not to aggregate SAL1L2s.
<agg_pct>: TRUE or FALSE, indicating whether or not to aggregate PCTs.
<agg_nbrcnt>: TRUE or FALSE, indicating whether or not to aggregate NBR_CNTs.
<agg_ssvar>: TRUE or FALSE, indicating whether or not to aggregate SSVARs.
<agg_vl1l2>: TRUE or FALSE, indicating whether or not to aggregate VL1L2s.
<agg_val1l2>: TRUE or FALSE, indicating whether or not to aggregate Vector Anomaly Partial Sums.
<agg_grad>: TRUE or FALSE, indicating whether or not to aggregate GRADs.
<boot_repl>: Number of bootstrapping replications, use 1 for no bootstrapping.
<boot_random_seed>: Bootstrapping seed.
<boot_ci>: Type of confidence interval to calculate, passed to the boot.ci() R function (e.g. bca).
<cache_agg_stat>: true or false, turns on/off the prevention the reuse of existing bootstrapping output data. Default: FALSE
<circular_block_bootstrap>: true or false, turns on/off the circular block bootstrap method. Default: TRUE. Python only.

When using <agg_stat>, the following constraints and conditions apply:

  • To turn off bootstrapping, set <boot_repl> to 1.

  • agg_stat types must only be used with appropriate statistics for each, and only one can be used on a single plot.

  • agg_stat is performed once for each plot that is created.

  • A plot cannot have both agg_stat and non-agg_stat data on the same plot.

  • If the agg_stat difference curve is turned on, the plot1 difference curve must be turned off.

  • To display the calculated bootstrap CIs, set the plot_cis to “boot”.

  • Bootstrapping CIs cannot be calculated for PSTD stats, instead use the “brier” CI type for Brier score.

  • The plot must contain statistics for only one fcst_var.

  • All series on the plot must be derived from the same data

  • If agg_stat is turned on, event equalization must be turned off (it is done automatically).

  • agg_stat cannot be used on the same plot with calc_stat.

<calc_stat>: Contains flags telling METviewer to calculate the plot statistics from CTCs or partial sums. If present, calc_stat is turned on. Depending on the selected <dep> statistics, only certain settings are appropriate.
<calc_ctc>: TRUE or FALSE, indicating whether or not to calculate CTC stats.
<calc_sl1l2>: TRUE or FALSE, indicating whether or not to calculate SL1L2 stats.
<calc_sal1l2>: TRUE or FALSE, indicating whether or not to calculate SAL1L2 stats.
<calc_vl1l2>: TRUE or FALSE, indicating whether or not to calculate VL1L2 stats.

When using <calc_stat>, the following constraints apply:

  • calc_ctc, calc_sl1l2, calc_sal1l2, calc_vl1l2 must only be used with appropriate statistics for each, and only one can be used on a single plot.

  • A plot cannot have both calc_stat and non-calc_stat data on the same plot.

  • calc_stat cannot be used on the same plot with agg_stat.

<roc_calc>: Contains flags telling METviewer how to calculate the points on the ROC curve: from PCTs or CTCs. This setting is required if the roc.R_tmpl is used, otherwise, it is ignored.
<roc_pct>: TRUE or FALSE, indicating whether or not to calculate ROC points using Probabilistic Contingency Table Counts (PCTs).
<roc_ctc>: TRUE or FALSE, indicating whether or not to calculate ROC points using Contingency Table Counts (CTCs).
When using <roc_calc>, the following constraints apply:

  • Only one of roc_pct or roc_ctc can be used on a single plot.

  • If using roc_pct, select a list of probabilistic threshold values as a fcst_thresh <plot_fix> value.

  • If using roc_pct, select a single obs_thresh <plot_fix> value.

  • If using roc_ctc, select two or more fcst_thresh <plot_fix> values and a single obs_thresh <plot_fix> value.

<normalized_histogram>: Contains flags telling METviewer which type of histogram to build: normalized or raw counts plot. This setting is required if the rhist.R_tmpl is used, otherwise, it is ignored. The default value is TRUE (normalized histogram).
<normalized_histogram>: TRUE or FALSE, indicating whether to build a normalized or raw counts histogram.

<ensss_pts>: Number of binned points.

  • If the number of binned points is not specified or < 1, use a default.

  • If the number of binned points is > 10 , use 1.

  • In other cases use the result of division of the number of points on 10.

<execution_type>: Rscript or Python, indicating whether to use Rscript or Python for the statistics calculation and plotting (if available). This is an optional element. The default value is Rscript.
<event_equal>: true or false, turns on the event equalizer.
<event_equal_m>: true or false, turns on the event equalizer for multiple events.
<vert_plot>: true or false, turns on vertical levels plotting.
<x_reverse>: true or false, reverse the direction of the x-axis.
<num_stats>: true or false, turns on the number of stats for each independent variable value, shown along the x2 axis.
<indy1_stag>: true or false, indicates whether y1 series points should be staggered at each independent variable value, for visibility.
<indy2_stag>: true or false, indicates whether y2 series points should be staggered at each independent variable value, for visibility.
<grid_on>: true or false, turns on gridding on the plot.
<sync_axes>: true or false, indicates that the y1 and y2 axes should be the same.
<dump_points1>: true or false, turns on a report for the y1 series points.
<dump_points2>: true or false, turns on a report for the y2 series points.
<log_y1>: true or false, turns on log-scale on the y1 axis.
<log_y2>: true or false, turns on log-scale on the y2 axis.
<varianceInflationFactor>: true or false, include/exclude the variance inflation factor when computing standard errors for means and medians - TRUE: include, FALSE: exclude - default value: TRUE.
<order_series>: true or false, include/exclude the variance inflation factor when computing standard errors for means and medians - TRUE: include, FALSE: exclude - default value: TRUE.

The following group of settings are each optional and act as placeholders for R plotting function arguments. Specified values should follow the format of the default values listed. The values are passed directly to plotting functions which can be viewed in the R template files for further details. The R function and parameter name that the value is used for is listed after the tag name below. The default value is also shown for each setting.

<plot_type>: bitmap(type) - plot image type - default value: png256
<plot_height>: bitmap(height) - height of the display region, in specified units - default value: 8.5
<plot_width>: bitmap(width) - width of the display region, in specified units - default value: 11
<plot_res>: bitmap(res) - resolution, in dots per inch - default value: 72
<plot_units>: bitmap(units) - units for height and width - default value: in
<mar>: par(mar) - lines of margin in form c(bottom, left, top, right) - default value: c(8, 4, 5, 4)
<mgp>: par(mgp) - margin line for axis title, axis labels and axis line in form c(title, labels, line) - default value: c(1, 1, 0)
<cex>: par(cex) - magnification for plot text - default value: 1
<title_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 2
<title_size>: mtext(cex) - relative magnification - default value: 1.4
<title_offset>: mtext(padj) - axis label perpendicular location adjustment - default value: -.4
<title_align>: mtext(adj) - axis label parallel location adjustment - default value: .5
<xtlab_orient>: axis(las) - axis label orientation - 0: parallel to axis, 1: horiz, 2: perp to axis, 3: vert - default value: 1
<xtlab_perp>: axis(padj) - axis label perpendicular location adjustment - default value: -.75
<xtlab_horiz>: axis(hadj) - axis label parallel location adjustment - default value: .5
<xtlab_freq>: axis(labels) - frequency of axis labels, 0 to disable - default value: 0
<xtlab_size>: axis(cex) - relative axis label magnification - default value: 1
<xlab_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 1
<xlab_size>: mtext(cex) - relative axis label magnification - default value: 1
<xlab_offset>: mtext(padj) - axis label perpendicular location adjustment - default value: 2
<xlab_align>: mtext(adj) - axis label parallel location adjustment - default value: .5
<ytlab_orient>: axis(las) - axis label orientation - 0: parallel to axis, 1: horiz, 2: perp to axis, 3: vert - default value: 1
<ytlab_perp>: axis(padj) - axis label perpendicular location adjustment - default value: .5
<ytlab_horiz>: axis(hadj) - axis label parallel location adjustment - default value: .5
<ytlab_size>: axis(cex) - relative axis label magnification - default value: 1
<ylab_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 1
<ylab_size>: mtext(cex) - relative axis label magnification - default value: 1
<ylab_offset>: mtext(padj) - axis label perpendicular location adjustment - default value: -2
<ylab_align>: mtext(adj) - axis label parallel location adjustment - default value: .5
<grid_lty>: abline(lty) - line type - 0: blank, 1: solid, 2: dashed, 3: dotted, 4: dotdash, 5: longdash, 6: twodash - default value: 3
<grid_col>: abline(col) - line color - default value: #CCCCCC
<grid_lwd>: abline(lwd) - line width - default value: 1
<grid_x>: abline(v) - list of positions for the gridlines - default value: listX
<x2tlab_orient>: axis(las) - axis label orientation - 0: parallel to axis, 1: horiz, 2: perp to axis, 3: vert - default value: 1
<x2tlab_perp>: axis(padj) - axis label perpendicular location adjustment - default value: 1
<x2tlab_horiz>: axis(hadj) - axis label parallel location adjustment - default value: .5
<x2tlab_size>: axis(cex) - relative axis label magnification - default value: .8
<x2lab_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 1
<x2lab_size>: mtext(cex) - relative axis label magnification - default value: .8
<x2lab_offset>: mtext(padj) - axis label perpendicular location adjustment - default value: -.5
<x2lab_align>: mtext(adj) - axis label parallel location adjustment - default value: .5
<y2tlab_orient>: axis(las) - axis label orientation - 0: parallel to axis, 1: horiz, 2: perp to axis, 3: vert - default value: 1
<y2tlab_perp>: axis(padj) - axis label perpendicular location adjustment - default value: .5
<y2tlab_horiz>: axis(hadj) - axis label parallel location adjustment - default value: .5
<y2tlab_size>: axis(cex) - relative axis label magnification - default value: 1
<y2lab_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 1
<y2lab_size>: mtext(cex) - relative axis label magnification - default value: 1
<y2lab_offset>: mtext(padj) - axis label perpendicular location adjustment - default value: 1
<y2lab_align>: mtext(adj) - axis label parallel location adjustment - default value: .5
<legend_size>: legend(size) - relative magnification - default value: .8
<legend_box>: legend(bty) - legend box type - o: box, n: none - default value: o
<legend_inset>: legend(inset) - inset distance from margin - default value: c(0, -.25)
<legend_ncol>: legend(ncol) - number of columns in legend - default value: 3
<caption_weight>: mtext(font) - 1: plain text, 2: bold, 3: italic, 4: bold italic and 5: symbol - default value: 1
<caption_col>: mtext(col) - font color - default value: #999999
<caption_size>: mtext(cex) - relative magnification - default value: .8
<caption_offset>: mtext(padj) - axis perpendicular location adjustment - default value: 5
<caption_align>: mtext(adj) - axis parallel location adjustment - 0:left, 1: right - default value: 0
<box_pts>: points() - draw points instead of boxes - TRUE: draw, FALSE: no draw - default value: FALSE.
<box_outline>: boxplot(outline) - draw outliers - TRUE: draw, FALSE: no draw - default value: TRUE.
<box_boxwex>: boxplot(boxwex) - scale factor to apply to all box widths - default value: 1
<box_notch>: boxplot(notch) - draw box notches - TRUE: notches on, FALSE: notches off - default value: FALSE
<box_avg>: points() - draw box averages as asterisks - TRUE: on, FALSE: off - default value: FALSE
<rely_event_hist>: - draw histogram of events by threshold on reliability diagram - TRUE: draw, FALSE: no draw - default value: TRUE
<ci_alpha>: alpha value used to calculate the standard median error confidence intervals - default value: .05
<eqbound_low>: lower equivalence bounds expressed in standardized mean difference (Cohen’s dz) - default value: -0.001
<eqbound_high>: upper equivalence bounds expressed in standardized mean difference (Cohen’s dz) - default value: 0.001

The following group of settings should be set to lists of values where the length of each list is equal to the number of series that are specified for the plot. Calculating the number of series and furthermore, the order of the series, can be done by considering the series1, series2 and dep values. For each dep1 statistic, enumerate each series1 permutation. Then, do the same for each dep2 statistic.

<plot_ci>: confidence interval type for each series - options: none, norm, boot, brier, std - example: c(“norm”, “none”)
<show_signif>: if the statistically significant CI’s should be highlighted c(FALSE, TRUE)
<plot_disp>: indicates which series are visible or not using TRUE/FALSE - example: c(TRUE, FALSE)
<colors>: lines(col) - line color for each series - example: c(“#33FF00FF”, “#00FF19FF”)
<pch>: lines(pch) - point symbol - see example:(points) - example: c(19, 19)
<type>: lines(type) - series line type - p: points, l: lines, o: overplotted points and lines, b or c: points joined by lines, s or S: stair steps, h: histogram, n: nothing - example: c(“b”, “h”)
<lty>: lines(lty) - line type - 0: blank, 1: solid, 2: dashed, 3: dotted, 4: dotdash, 5: longdash, 6: twodash - example: c(1, 2)
<lwd>: lines(lwd) - line width - example: c(2, 2)
<con_series>: if a series contains NAs, specifies whether to connect the points or not - 0: do not connect, 1: connect - example: c(0, 1)

<legend>: legend labels for each series - example: c(“series1”, “series2”)
<order_series>: the order of the series - example: c(1, 2, 3)
<x1_lim>: lower y axis (x1-axis) limits - example: c(10,15) or c(“2017-05-10 00:00:00”,”2017-05-15 06:00:00”)
<y1_lim>: left y axis (y1-axis) limits - example: c(0,5)
<y1_bufr>: left y axis (y1-axis) top and bottom buffer - example: .04
<y2_lim>: right y axis (y2-axis) limits - example: c(0,1)
<y2_bufr>: right y axis (y2-axis) top and bottom buffer - example: .04
<plot_cmd>: extra R commands to execute at the end of the plot script - example: par(usr=listY1Usr); abline(h=1, lty=1, col=”#999999FF”);

Elements specific to ROC curve:
<add_point_thresholds>: TRUE or FALSE, indicating whether or not to display threshold points on the plot. The default value is TRUE (display threshold points).
<reverse_connection_order>: TRUE or FALSE, indicating whether or not to reverse the connection order of the points on the graph created from CTC data. The default value is FALSE (standard order).

Elements specific to Rely plot:
<add_skill_line>: TRUE or FALSE, indicating whether or not to display skill line on the graph. The default value is TRUE (display skill line).
<add_reference_line>: TRUE or FALSE, indicating whether or not to display the reference line on the graph. The default value is TRUE (display reference line).

8.1. MODE Statistics

8.1.1. Single Object Statistics

MODE statistics are broken up into two categories: single and pair statistics. Single statistics are suffixed with a three letter code which indicates the group of objects over which the statistic should be calculated or aggregated. The syntax for the three letter code is as follows: STATNAME_CCC where STATNAME is one of the Single Object Statistics in the table below, and CCC follows the following system:

[A|F|O|D] - indicating All, Forecast, Observed or a Difference between the forecast statistic and the observed statistic

[A|S|C] - indicating All, Simple or Cluster objects

[A|M|U] - indicating All, Matched or Unmatched objects

Single Object Stat

METviewer Stat Name

SUM(area) / total

ACOV

COUNT(object_id)

CNT

COUNT(object_id)

CNTSUM

centroid_x

CENTX

centroid_y

CENTY

centroid_lat

CENTLAT

centroid_lon

CENTLON

axis_avg

AXAVG

length

LEN

width

WID

area

AREA

area_threshold

AREATHR

curvature

CURV

curvature_x

CURVX

curvature_y

CURVY

complexity

CPLX

intensity_10

INT10

intensity_25

INT25

intensity_50

INT50

intensity_75

INT75

intensity_90

INT90

intensity_nn

INTN

intensity_sum

INTSUM

In addition to the single object statistics, METviewer supports ratios of single object counts and areas, and also some contingency table statistics. These statistics are specified in the table below with their corresponding interpretation. For the area-weighted version of these statistics, replace RATIO_ with AREARAT_ and OBJ with OBJA.

Interpretation

METviewer Stat Name

What percentage of simple objects are forecast?

RATIO_FSA_ASA

What percentage of simple objects are observation?

RATIO_OSA_ASA

What percentage of simple objects are matched?

RATIO_ASM_ASA

What percentage of simple objects are unmatched?

RATIO_ASU_ASA

What percentage of simple forecast objects are matched?

RATIO_FSM_FSA

What percentage of simple forecast objects are unmatched?

RATIO_FSU_FSA

What percentage of simple observation objects are matched?

RATIO_OSM_OSA

What percentage of simple observation objects are unmatched?

RATIO_OSU_OSA

What percentage of simple matched objects that are forecast?

RATIO_FSM_ASM

What percentage of simple matched objects that are observation?

RATIO_OSM_ASM

What percentage of simple unmatched objects that are forecast?

RATIO_FSU_ASU

What percentage of simple unmatched objects that are observation?

RATIO_OSU_ASU

What percentage of all objects are simple?

RATIO_ASA_AAA

What percentage of all objects are cluster?

RATIO_ACA_AAA

What percentage of all forecast objects are simple?

RATIO_FSA_FAA

What percentage of all forecast objects are cluster?

RATIO_FCA_FAA

What percentage of all observation objects are simple?

RATIO_OSA_OAA

What percentage of all observation objects are cluster?

RATIO_OCA_OAA

What percentage of cluster objects are forecast?

RATIO_FCA_ACA

What percentage of cluster objects are observation?

RATIO_OCA_ACA

What is the ratio of simple forecasts to simple observations (frequency bias)?

RATIO_FSA_OSA

What is the ratio of simple observations to simple forecasts (1 / frequency bias)?

RATIO_OSA_FSA

What is the ratio of cluster objects to simple objects?

RATIO_ACA_ASA

What is the ratio of simple objects to cluster objects?

RATIO_ASA_ACA

What is the ratio of cluster forecast objects to simple forecast objects?

RATIO_FCA_FSA

What is the ratio of simple forecast objects to cluster forecast objects?

RATIO_FSA_FCA

What is the ratio of cluster observation objects to simple observation objects?

RATIO_OCA_OSA

What is the ratio of simple observation objects to cluster observation objects?

RATIO_OSA_OCA

Hits = (FSM + OSM)/2

OBJHITS

Misses = OSU

OBJMISSES

False Alarms = FSU

OBJFAS

Correct Negatives (areas only) = [(TOT - FSA) + (TOT - OSA)] / 2

(not implemented)

CSI = hits / (hits + misses + false alarms) = [(FSM + OSM)/2] / [(FSM + OSM)/2 + OSU + FSU]

OBJCSI

PODY = hits / (hits + misses) = [(FSM + OSM)/2] / [(FSM + OSM)/2 + OSU]

OBJPODY

FAR = false alarms / (hits + false alarms) = FSU / [(FSM + OSM)/2 + FSU]

OBJFAR

FBIAS = RATIO_FSA_OSA = FSA / OSA (a.k.a. RATIO_FSA_OSA above)

OBJFBIAS

8.1.2. Pair Object Statistics

MODE object pair statistics are specified using the following format: STATNAME_CC where STATNAME is one of the Object Pair Statistics in the table below, and CC follows the following system:

[A|S|C] - indicating All, Simple or Cluster pairs

[A|M|U] - indicating All, Matched or Unmatched objects

Object Pair Stat

METviewer Stat Name

centroid_dist

CENTDIST

boundary_dist

BOUNDDIST

convex_hull_dist

HULLDIST

angle_diff

ANGLEDIFF

area_ratio

AREARAT

intersection_area

INTAREA

union_area

UNIONAREA

symmetric_diff

SYMDIFF

intersection_over_areai

INTOVERAREA

complexity_ratio

CMPLXRATIO

percentile_intensity_ratio

PERCINTRATIO

interest

INT

MAX(interest)

MAXINT

MAX(interest) for fcst objects

MAXINTF

MAX(interest) for obs objects

MAXINTO