contour lines incorrect when data range is a few orders of magnitude

[karlmsmith]

Reported by @AnsleyManke on 1 Oct 2010 23:41 UTC
The question reported in
http://www.pmel.noaa.gov/maillists/tmap/ferret_users/fu_2010/msg00552.html

Is a case where the data field has a range of several orders of magnitude but a FILL/LINE plot uses a /LEVELS setting to plot only a range of the smaller values. The contour lines don't match the color fill.

Here is an example with etopo05.

yes? use etopo05
yes? let var = if abs(rose) gt 3000 then abs(rose/100) else abs(rose)

yes? ! See the values. They range over a couple orders of magnitude.
yes? list/wid=80/x=75w:71w/y=16s:14s var
             VARIABLE : IF ABS(ROSE) GT 3000 THEN ABS(ROSE/100) ELSE ABS(ROSE)
             FILENAME : etopo05.cdf
             FILEPATH : /home/shiva/data/
             SUBSET   : 49 by 25 points (LONGITUDE-LATITUDE)
      ... listing every   6th point
                 74.998W74.498W73.998W73.498W72.998W72.498W71.998W71.498W70.998W
                3421   3427   3433   3439   3445   3451   3457   3463   3469
 14S     / 913:    39.    43.    39.    40.    43.    41.    42.    43.    50.
 14.083S / 912:    38.    42.    40.    40.    43.    42.    42.    41.    49.
 14.166S / 911:    37.    41.    40.    40.    44.    43.    41.    40.    48.
 14.25S  / 910:    32.    41.    41.    42.    44.    44.    41.    40.    47.
 14.333S / 909:  2743.    41.    42.    43.    44.    44.    40.    41.    45.
 14.416S / 908:  2552.    41.    43.    43.    46.    45.    42.    41.    44.
 14.5S   / 907:  2362.    40.    44.    43.    47.    46.    43.    41.    43.
 14.583S / 906:  1981.    39.    44.    44.    48.    47.    43.    41.    41.
 14.666S / 905:  1601.    37.    44.    44.    48.    49.    44.    40.    40.
 14.75S  / 904:  1372.    35.    42.    43.    48.    49.    45.    40.    40.
 14.833S / 903:  1143.    32.    41.    43.    47.    49.    46.    40.    41.
 14.916S / 902:  1204.  2743.    37.    43.    45.    48.    47.    43.    43.
 15S     / 901:  1265.  2286.    33.    44.    43.    47.    48.    46.    46.
 15.083S / 900:  1143.  2209.    34.    38.    44.    47.    47.    46.    47.
 15.166S / 899:  1021.  2133.    36.    33.    44.    46.    46.    46.    48.
 15.25S  / 898:   678.  1981.    37.    32.    45.    44.    48.    46.    48.
 15.333S / 897:   336.  1829.    37.    32.    46.    42.    49.    45.    47.
 15.416S / 896:   206.  1806.    34.    34.    43.    42.    46.    45.    46.
 15.5S   / 895:   105.  1783.    30.    37.    40.    43.    43.    44.    46.
 15.583S / 894:   223.  1272.  2743.    31.    38.    38.    46.    44.    45.
 15.666S / 893:   250.   762.  2438.  2499.    36.    33.    48.    44.    44.
 15.75S  / 892:   894.    13.  2133.  2202.    30.    30.    45.    44.    45.
 15.833S / 891:  1503.   218.  1829.  1905.  2439.  2819.    43.    44.    46.
 15.916S / 890:  2120.   515.  1562.  1745.  2309.  2362.    37.    43.    46.
 16S     / 889:  2515.   770.  1296.  1585.  2179.  1905.    32.    42.    46.

yes? ! FILL/LINE plot, Lines don't match. See the upper panel, attached plot. 
yes? fill/line/lev=(-inf)(40,50,2)(inf)/x=75w:71w/y=16s:14s var

yes? ! if we redefine the variable so it only has the smaller values, 
yes? ! then the plot is fine. See the lower panel.
yes?  fill/line/lev=(-inf)(40,50,2)(inf)/x=75w:71w/y=16s:14s if var le 60 then var

Migrated-From: http://dunkel.pmel.noaa.gov/trac/ferret/ticket/1757

[karlmsmith]

Attachment from @AnsleyManke on 1 Oct 2010 23:46 UTC