Usage:
      bufrdump.pl <bufr file(s)>
          [--filter <filter file | filter list>]
          [--param <parameter file | parameter list> [--csv [--delimiter <del>]]
          [--sort]
          [--sort_on <parameter>[-]]
          [--station <station list>]
          [--transform <transformation file>]
          [--lon1 <x1>]
          [--lat1 <y1>]
          [--lon2 <x2>]
          [--lat2 <y2>]
          [--obstype <amdar|ocea|surface|sounding|sounding->]
          [--tablepath <path to BUFR tables>]
          [--help]

Options:
      --filter <filter file | filter list>
                      Decode observations meeting criteria in filter file or
                      filter list only
      --param <parameter file | parameter list> [--csv [--delimiter <del>]]
                      Print parameters in parameter file or comma
                      separated list (e.g. wmonr,TA) only, in same order
                      as they occur there. If using --csv possibly
                      followed by --delimiter <del>, the parameters vill
                      be printed using the CSV (comma-separated values)
                      format, with the delimiter del (default is ';')
      --sort          Sort the decoded observations on station identification;
                      first stations with wmonr, then stations with nationalnr,
                      call sign, buoy_id or aircraft (others left out)
      --sort_on <parameter>[-] Sort the decoded observations on increasing
                      values of parameter, or decreasing values if a '-'
                      follows the parameter name. E.g. --sort_on TA- will
                      sort on decreasing temperatures. Observations not
                      containing the parameter at all will be printed lastly,
                      except when --sort_on is combined with --sort (in which
                      case sorting is done firstly on station identification,
                      secondly on parameter with missing values printed first)
      --station <station list>
                      Print observations for stations in station list only,
                      e.g. wmonr=01384,01492
      --transform <transformation file>
                      Do the transformations of parameter values listed in
                      transformation file
      --lon1 <x1>     Decode observations with longitude >= x1 only
      --lat1 <y1>     Decode observations with latitude >= y1 only
      --lon2 <x2>     Decode observations with longitude <= x2 only
      --lat2 <y2>     Decode observations with latitude <= y2 only
                      x1,y1,x2,y2 should be decimal degrees
      --obstype <amdar|ocea|surface|sounding|sounding->]
                      Force observation type. If this option is not set,
                      will make an educated guess of observation type
                      based on metadata in section 1 of each BUFR message
      --tablepath <path to BUFR tables>
                      Set path to BUFR tables (overrides ENV{BUFR_TABLES})
      --help          Print this Usage (but you might instead prefer to use
                      perldoc bufrdump.pl)

    Options may be abbreviated, e.g. --h or -h for --help.

    To avoid having to use the "--tablepath" option, you are adviced to set
    the environment variable BUFR_TABLES to the directory where your BUFR
    tables are located (unless the default path provided by bufrdump.pl
    works for you).

    The lines in <parameter file>, or the comma separated values in
    <parameter list>, should be name of the parameters you want to be
    printed. For example, if you want only station identification and
    temperature to be printed for a BUFR SYNOP file, either supply

      wmonr,nationalnr,call_sign,TA

    as argument to --params, or supply a <parameter file> which should look
    like this:

      wmonr
      nationalnr
      call_sign
      TA

    If you want "parameter=value" to be printed also when value is missing
    in BUFR message, precede the parameter name with an exclamation mark
    (e.g. '!TA'). Missing values will then be displayed as -32767. If the
    argument to --param is a parameter list, you must prevent the shell from
    attaching special meaning to the exclamation mark by enclosing the list
    in single quotes.

    If the parameter list consists of one parameter only, a comma must be
    appended (e.g. 'wmonr,') because bufrdump.pl uses the appearence of
    comma to signal that this is not a filename but parameter name(s).

    If --csv is used in conjunction with --param, all values will be printed
    using the CSV format, with first line listing the parameters, and with
    missing fields printed as -32767 if the parameter is marked with '!' in
    parameter file or list. With the parameter file above, the listing may
    for example start like

      wmonr;nationalnr;call_sign;TA
      01001;;;-1.5
      ;;LF5U;9.0

    You can choose another delimiter than semicolon by use of option
    --delimiter <del>, e.g. --csv --delimiter ','

    Using --filter will decode only those observations that meet at least
    one of the BUFR descriptor criteria and all of the parameter criteria in
    <filter file>, where the BUFR descriptor criteria should come first in
    filter file followed by a blank line, then comes the parameter criteria
    which should match <param> or !<param> or <param> <operator> <value>
    where operator is one of =, !=, =~, !~, <, <=, > and >=. What follows =~
    and !~ should be a Perl match regular expression. The parameter criteria
    may be phrased as alternatives by separating them with '|' on a single
    line. An example filter file is

      D: 001001 I2.2
      01
      D: 001001 I2.2 001002 I3.3
      03 895
      06 252
      D: 001011 A9
      LF5U

      type = Manned
      NN != 8
      TA >= 5
      TA < 9.5
      RR_24

    which decodes all observations with block number 01, two other specific
    wmo stations and one specific ship, where stations should be manned and
    have cloud cover with a value different from 8, and have temperature
    between 5 and 9.5 degrees Celsius, and contain precipitation for last 24
    hours. Comment lines starting with # will be ignored.

    Another example: the filter file (starting with a blank line!)

      call_sign =~ /^L[A-N]..$/
      obstime >= '2012-02-10 06:00:00'
      HW | HWA | PW | PWA
      FF > 10 | FG_010 > 10

    will print only those ship observations for which the 4 character
    call_sign starts with 2 letters in the interval LA-LN, and having
    obstime larger or equal to the datetime given, and containing wave data
    (specifically: height or period of waves, manually or automatically
    measured), and with wind or 10 minutes gust more than 10 m/s.

    For convenience, when there are no BUFR descriptor criteria, you might
    provide the filter criteria on the command line. Example:

    --filter 'wmonr,TA > 0,RR_12 | RR_24, !FF'

    will decode only observations with wmonr, having positive temperature
    and containing precipitation for 12 or 24 hours and not reporting wind.
    If (like for --param) the filter list consists of one criterium only, a
    comma must be appended.

    To avoid the need of creating a filter file when observations for some
    few stations are requested, you can provide the stations in a comma
    separated list after option --station. Some examples:

      --station wmonr=01001,01152,01492
      --station nationalnr=614_0050410003,637_108
      --station call_sign=LF5U
      --station buoyid=64607,64609
      --station aircraft=EU3421,JHCWUURA

    You cannot mix different kinds of stations this way (before '=' you must
    choose either wmonr, nationalnr, call_sign, buoy_id or aircraft). Note
    also that providing the stations in the BUFR descriptor part (first
    part) of the filter file will speed up execution time considerably,
    compared to using option --station. It is possible to combine --filter
    with --station if done with some care, e.g. specifying WMO block 01 and
    the required parameters in filter file, then the requested stations in
    station list.

    The --transform option is provided mainly to be able to use other units
    than what is default in bufrdump.pl. The transformation file should list
    the transformations wanted, one per line as

      <parameter> = <perl expression involving $x>

    where $x is original value of the parameter.

    For example, the following transformation file will display wind speed
    FF and wind gust FG in knots instead of m/s, rounded to one decimal, and
    cloud cover NN in % (instead of the default which is using WMO code
    table 2700, roughly counting octas):

      FF = sprintf("%.1f", $x*1.9438)
      FG = sprintf("%.1f", $x*1.9438)
      NN = int($x*12.5 + .5)

    If --transform is combined with --filter, the filter criteria should
    refer to the transformed values. E.g. if the above NN transform to % is
    to be applied for sky not all covered by clouds, you should use NN !=
    100 instead of NN != 8 in filter file.

    The --obstype option might be handy in some special cases, like when you
    are interested only in the surface part of oceanographic data (then use
    '--obstype surface'), or when you want to see only levels with vss>0 in
    high resolution radiosonde data (then use '--obstype sounding-'), or
    when data category and/or data sub-category in the BUFR messages have
    unusual values.