#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# $Id: FormatNotes,v 1.15 2009/08/08 22:07:27 swift Exp $
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# RCS Log:
#
# $Log: FormatNotes,v $
# Revision 1.15  2009/08/08 22:07:27  swift
# Implemented park-level PTS sampling.
#
# Revision 1.14  2007/06/08 02:16:05  swift
# Minor change to facilitate automatic management of FormatNotes.
#
# Revision 1.13  2007/05/07 21:05:34  swift
# Made corrections in FormatNotes with respect to temperature encoding.
#
# Revision 1.12  2006/11/24 23:59:10  swift
# Changed format description for CP data.
#
# Revision 1.11  2006/09/30 17:32:45  swift
# Modifications to account for a shorter encoding of the high-resolution CTD
# data.
#
# Revision 1.10  2006/01/31 19:26:48  swift
# Eliminated "dbars" and "C" from ParkPt entries.
#
# Revision 1.9  2006/01/06 23:25:41  swift
# Minor syntax fixes.
#
# Revision 1.8  2005/10/25 17:02:46  swift
# Change message extension from 'prf' to 'msg'.
#
# Revision 1.7  2005/09/05 16:04:16  swift
# Fleshed out formatting documentation for iridium messages.
#
# Revision 1.6  2005/03/30 21:11:09  swift
# Description of the encoding of data.
#
# Revision 1.5  2004/12/29 23:12:17  swift
# Deleted format notes that pertained only to ARGOS firmware.
#
# Revision 1.4  2004/04/10 18:38:27  swift
# Format description for use with 28-bit Argos ids.
#
# Revision 1.3  2004/04/10 18:36:54  swift
# Format description for use with 20-bit Argos ids.
#
# Revision 1.2  2003/11/21 02:46:23  swift
# Added documentation for the SBE41 status buts.
#
# Revision 1.1  2003/11/12 23:05:54  swift
# Initial revision
#
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Format specification for APF9 firmware revision: $FwName$-$FwRev$ 
===================================================================================

Iridium message files end with a ".msg" extension.  Each iridium message
file consists of blocks of similar data presented in the order that they
were collected during the profile cycle.  This firmware revision includes
five blocks of data:

   1) Park-phase PTS samples: These are hourly low-power PTS samples collected
      during the park phase of the profile cycle.

   2) Low resolution PTS samples: The deep parts of the profile can be
      represented using low-resolution spot samples collected at
      predetermined pressures.  Low resolution spot sampling in the deep
      water was implemented as an energy savings measure.

   3) High resolution PTS samples: The shallower parts of the profile can be
      represented with high resolution (ie., 2 decibar) bin-averaged PTS
      samples.  In continuous profiling mode,  the CTD samples at 1Hz and
      stores the data for later binning and averaging.

   4) GPS fixes: After the profile is executed and the float reaches the
      surface, the location of the profile is determined via GPS.

   5) Biographical & engineering data: Various kinds of biographical and
      engineering data are collected at various times during the profile
      cycle.  

Usually, only one telemetry cycle is required to upload the data to the
remote host computer.  However, sometimes the iridium connection is broken
or the quality of the connection is so poor that the float will abort the
telemetry attempt, wait a few minutes, and then try again.  Data blocks 4
and 5 will be repeated for each telemetry cycle of a given profile.

A description of the format for each of these blocks of data follows.


1) Park-phase PTS sample format.
-------------------------------

   Hourly low-power PTS samples are collected during the park phase of the
   profile cycle.  The park phase is also when active ballasting is done.
   Each sample includes the date & time of the sample, the unix epoch (ie.,
   the number of seconds since 00:00:00 on Jan 1, 1970), the mission time
   (ie., the number of seconds since the start of the current profile cycle),
   the pressure (decibars), temperature (C), and salinity (PSU).  For example:

                |------- date -----|   UnixEpoch   MTime       P       T       S
      ParkPts:  Jul 26 2009 15:38:33  1248622713   14432  946.39  3.4410 34.8501
      ParkPts:  Jul 26 2009 16:38:31  1248626311   18030  981.49  3.4242 34.8500
      ParkPts:  Jul 26 2009 17:38:31  1248629911   21630  997.73  3.4166 34.8500
      ParkPts:  Jul 26 2009 18:38:31  1248633511   25230 1005.22  3.4103 34.8494
      ParkPts:  Jul 26 2009 19:38:31  1248637111   28830 1008.79  3.4069 34.8490
      ParkPts:  Jul 26 2009 20:38:31  1248640711   32430 1010.48  3.4053 34.8489
      ParkPts:  Jul 26 2009 21:38:31  1248644311   36030 1011.29  3.4046 34.8488
      ParkPts:  Jul 26 2009 22:38:31  1248647911   39630 1011.68  3.4042 34.8487
      ParkPts:  Jul 26 2009 23:38:31  1248651511   43230 1011.84  3.4040 34.8487
      ParkPts:  Jul 27 2009 00:38:31  1248655111   46830 1011.92  3.4040 34.8487



2) Low resolution PTS sample format.
------------------------------------
   
   The SBE41CP that is used on iridium floats has features that enable
   subsampling of the water column (similar to the SBE41) as well as the
   ability to bin-average a continuous sampling of the water column.  For
   subsampled data, the values of pressure, temperature, and salinity are not
   encoded but are given in conventional units (decibars, C, PSU).  For
   example:
      
      $ Discrete samples: 6
      $       p        t        s
        1021.60   5.1719  34.6362 (Park Sample)
        1994.88   2.4792  34.7165
        1947.50   2.5454  34.7142
        1898.08   2.6260  34.7115
        1847.83   2.7133  34.7076
        1797.26   2.8271  34.7017
        1746.87   2.8982  34.6964



3) High resolution PTS sample format.
-------------------------------------   

   For continuously sampled data, 2-decibar bins are used for bin-averaging.
   These data are encoded as three 16-bit integers (PTS) and then an 8-bit
   integer that represents the number of samples in the bin:
   
      # Nov 05 2006 23:38:59 Sbe41cpSerNo[1520] NSample[11134] NBin[495]
      00000000000000[2]
      002B5CE0885115
      003C5CE188511F
      00505CE088511E
      00645CE088511D
      00785CDF88521C
      008C5CDF88521B
      00A05CDF88511A
      00B45CDC88511A
      00C85CAD885A1A
      00DC5C8D886418
      00F05C88886919
      01045C88886A19
      01185C74887618
      012C5C4E88BB15
      [snippage...]
      2684144D874814
      2698145C874D14
   
   The first 4-bytes of the encoded sample represents the pressure in
   centibars.  The second 4-bytes represents the temperature in
   millidegrees.  The third 4-bytes represent the salinity in parts per
   million.  The final 2-bytes represent the number of samples collected in
   the 2dbar pressure bin.
   
   For example, the encoding: 2698145C874D14 represents a bin with (0x14=) 20
   samples where the mean pressure was (0x2698=) 988.0dbars, the mean
   temperature was (0x145C=) 5.212C, and the mean salinity was (0x874D=)
   34.637PSU.  The PTS values were encoded as 16-bit hex integers according to
   the functions below.
   
   Integers in square brackets '[]' indicate replicates of the same encoded
   line.  For example, a line that looks like: 00000000000000[2] indicates
   that there were 2 adjacent lines with the same encoding....all zeros in
   this case.
   
   /*------------------------------------------------------------------------*/
   /* function to encode the number of samples as an 8-bit unsigned integer  */
   /*------------------------------------------------------------------------*/
   /**
      This function encodes the number samples in the bin average as an 8-bit
      unsigned integer with protection against overflow.  The encoding accounts
      for the full range of 16-bit unsigned integers but only values in the
      open range: 0<NSample<255 are representable.  This encoding makes full
      use of all 8-bits.
   
         \begin{verbatim}
         input:
            NSample ... The number of samples in the bin-average.
   
         output:
            1) Values greater than or equal to 255 are mapped to 0xff.
   
            2) All other values are expressed as an 8-bit unsigned integer.
         \end{verbatim}
   */
   unsigned char EncodeN(unsigned int NSample)
   {
      /* prevent overflow of the sample counter */
      unsigned int N = (NSample>=255) ? 0xff : NSample;
      
      return N;
   }
   
   /*------------------------------------------------------------------------*/
   /* function to encode pressure as a 2-byte unsigned integer               */
   /*------------------------------------------------------------------------*/
   /**
      This function implements the hex-encoding of IEEE-formattted floating
      point pressure data into 2-byte signed integers with 2's complement
      representation.  The encoding formula accounts for the full range of
      32-bit IEEE floating point values but only values in the open range:
      -3276.7<p<3276.7 are representable.  This encoding makes full use of all
      16-bits.
   
         \begin{verbatim}
         input:
            p ... The pressure (decibars) expressed as a floating point value.
   
         output:
            1) Nonfinite values (Inf, -Inf, NaN) are mapped to the sentinel hex
               value: 0x8000.
   
            2) Pressure values less than -3276.7 are mapped to the sentinel
               value: 0x8001.
   
            3) Pressure values greater than 3276.7 are mapped to the sentinel
               value: 0x7fff.
   
            4) All other values are expressed in millibars rounded to the
               nearest integer and expressed as a 16-bit signed integer in
               2's-complement form.
         \end{verbatim}
   */
   unsigned int EncodeP(float p)
   {
      /* initialize with the mapping for a nonfinite pressure */
      long int P = 0x8000L;
   
      /* make sure long ints are at least 3 bytes long */
      assert(sizeof(long int)>=3);
      
      if (finite(p))
      {
         /* assign out-of-range values to sentinel values */
         if (p>=3276.7) P=0x7fffL; else if (p<=-3276.7) P=0x8001L;
   
         /* encode the pressure as the number of centibars (rounded) */
         else P = (long int)(10*(p + ((p<0) ? -0.05 : 0.05)));
   
         /* express in 16-bit 2's-complement form */
         if (P<0) P+=0x10000L;
      }
   
      return P;
   } 
   
   /*------------------------------------------------------------------------*/
   /* function to encode salinity as a 2-byte unsigned long integer          */
   /*------------------------------------------------------------------------*/
   /**
      This function implements the hex-encoding of IEEE-formattted floating
      point salinity data into 16-bit unsigned integers with 2's complement
      representation.  The encoding formula accounts for the full range of
      32-bit IEEE floating point values but only values in the open range:
      -4.095<s<61.439 are representable.  This encoding makes full use of all
      16-bits.
   
         \begin{verbatim}
         input:
            s ... The salinity (PSU) expressed as a floating point value.
   
         output:
            1) Nonfinite values (Inf, -Inf, NaN) are mapped to the sentinel hex
               value: 0xf000.
   
            2) Salinity values less than -4.095 are mapped to the sentinel
               value: 0xf001.
   
            3) Salinity values greater than 61.439 are mapped to the sentinel
               value: 0xefff.
   
            4) All other values are expressed in parts-per-ten-million
               rounded to the nearest integer and expressed as a 16-bit
               signed integer in 2's-complement form.
         \end{verbatim}
   */
   unsigned int EncodeS(float s)
   {
      /* initialize with the mapping for a nonfinite salinity */
      long int S = 0xf000L;
   
      /* make sure that long integers have at least three bytes */
      assert(sizeof(long int)>=3);
   
      if (finite(s))
      {
         /* assign out-of-range values to sentinel values */
         if (s>=61.439) S=0xefffL; else if (s<=-4.095) S=0xf001L;
   
         /* encode the salinity as the number of parts-per-ten-million (rounded) */
         else S = (long int)(1000*(s + ((s<0) ? -0.0005 : 0.0005)));
   
         /* express in 16-bit 2's-complement form */
         if (S<0) S+=0x10000L;
      }
      
      return S;
   }
   
   /*------------------------------------------------------------------------*/
   /* function to encode temperature as a 2-byte unsigned integer            */
   /*------------------------------------------------------------------------*/
   /**
      This function implements the hex-encoding of IEEE-formattted floating
      point temperature data into 16-bit unsigned integers with 2's complement
      representation.  The encoding formula accounts for the full range of
      32-bit IEEE floating point values but only values in the open range:
      -4.095<t<61.439 are representable.  This encoding makes full use of all
      16-bits.
   
         \begin{verbatim}
         input:
            t ... The temperature (C) expressed as a floating point value.
   
         output:
            1) Nonfinite values (Inf, -Inf, NaN) are mapped to the sentinel hex
               value: 0xf000.
   
            2) Temperature values less than -4.095 are mapped to the sentinel
               value: 0xf001.
   
            3) Temperature values greater than 61.439 are mapped to the sentinel
               value: 0xefff.
   
            4) All other values are expressed in tenths of millidegrees Celsius
               rounded to the nearest integer and expressed as a 16-bit signed 
               integer in 2's-complement form.
         \end{verbatim}
   */
   unsigned int EncodeT(float t)
   {
      /* initialize with the mapping for a nonfinite temperature */
      long int T = 0xf000L;
   
      /* make sure that long integers have at least three bytes */
      assert(sizeof(long int)>=3);
   
      if (finite(t))
      {
         /* assign out-of-range values to sentinel values */
         if (t>=61.439) T=0xefffL; else if (t<=-4.095) T=0xf001L;
   
         /* encode the temperature as the number of tenths of millidegrees (rounded) */
         else T = (long int)(1000*(t + ((t<0) ? -0.0005 : 0.0005)));
   
         /* express in 16-bit 2's-complement form */
         if (T<0) T+=0x10000L;
      }
      
      return T;
   }
   


4) GPS fix format.
------------------
   
   Each telemetry cycle begins with the float attempting to acquire a GPS
   fix.  The fix includes the amount of time required to acquire the fix,
   the longitude and latitude (degrees), the date & time of the fix, and the
   number of satellites used to determine the fix.  For example:

      # GPS fix obtained in 98 seconds.
      #         lon     lat mm/dd/yyyy hhmmss nsat
      Fix: -152.945  22.544 09/01/2005 104710    8

   Positive values of longitude, latitude represent east, north hemispheres,
   respectively. Negative values of longitude, latitude represent west,
   south hemispheres, respectively.  The date is given in month-day-year
   format and the time is given in hours-minutes-seconds format.

   If no fix was acquired then the following note is entered into the
   iridium message:

      # Attempt to get GPS fix failed after 600 seconds.



5) Biographical & engineering data.
-----------------------------------

   These data have the format, "key"="value", as shown in the following
   examples:

      ActiveBallastAdjustments=5
      AirBladderPressure=119
      AirPumpAmps=91
      AirPumpVolts=192
      BuoyancyPumpOnTime=1539

   Interpretation of these data requires detailed knowledge of firmware
   implementations.  
