ENSEMBLES (EN3) quality controlled T and S profiles =================================================== FILES : ****** EN3_changes - changes from the EN2 processing EN3_files - notes about the NetCDF files and their contents EN3_v1c_1950.tar onwards - data in NetCDF format - Each tar file contains all the gzip'd NetCDF data for one year and the statistics file for each month (.stats). (The EN3_v1c_YYYY.MM.stats files contain the stats for that month. After the level dependent O-B and Ob, ObErr, BkErr etc, then there are the summary statistics (find Ops_QCStats). The summary section is repeated. The first set of data is on the observation levels, and the second on model levels.) - The OQCpaper.pdf document contains quite a bit of detail about the system used for EN2. - For some fortran to read the NetCDF, run the program gennet.f at http://www.unidata.ucar.edu/packages/netcdf/Contrib.html - it automatically generates some code. ABOUT THE DATA AND QC: ********************** There are several minor changes to the NetCDF files relative to EN2: - I have added TEMP (in situ temperature; I haven't updated its PARAM). - PROJECT_NAME gives the source WOD05/GTSPP/ARGO and category # - INST_REFERENCE gives more details of instrument type and correction # - SCIENTIFIC_CALIB_COEFFICENT gives depth multiplication factor # It was discovered (August 2007) that the values of these two # variables are incorrect in the NetCDF files due to an indexing error. See below for more details. NetCDF : The fields to be used are : DEPH_CORRECTED - depth (in metres) POTM_CORRECTED - potential temperature (or TEMP for in situ temperature) PSAL_CORRECTED - practical salinity BKPT and BKPS give the background values used in the QC (The DEPH, POTM and PSAL fields have been omited as they were identical to the _CORRECTED values.) *** IMPORTANT NOTE: please check the _QC flags before using the data. A value of 4 for POTM_CORRECTED_QC or POTM_CORRECTED_QC indicates an error in the temperature or salinity value at that level. Similarly a value of 4 for POSITION_QC indicates an error affecting the whole report. (1 indicates a "good" value, intermediate codes and DEPH_CORRECTED_QC are not currently used.) PROJECT_NAME = "WOD05MBT " ! last 3 characters give WOD05 file type PROJECT_NAME = "GTSPPBT " ! last 2 characters give GTSPP obs type PROJECT_NAME = "ARGOD " ! last character indicates delayed mode (D) or real time (R) or adjusted (A) For 1990 onwards the PROJECT_NAME for EN3 WOD05 data has been extended: PROJECT_NAME = "WOD05CTD49 8102" where 49 is the country code (Japan in this case, see Appendix 1. NODC country codes in WOD05 documentation) and 8102 is the platform from secondary header 3. PLATFORM_NUMBER = "3109602" For WOD05 this is made up of the cruise number (up to six digits) and the last two digits of the platform (from secondary header 3); the string is then left justified. This combination was chosen to distinguish different cruises for the track check. # INST_REFERENCE: # For GTSPP this contains the WMO code table 1770 type # (based on the GTSPP metadata two additional code values are used: # 101 - unknown instrument type, fall-rate coeff A=6.472 assumed # 102 - unknown instrument type, no fall-rate correction required) # For WOD05 this is 10000*Sech29 + 10*InstType + CorrType # Sech29 (secondary header 54) is probe type (0-16) # - table 2.13 (p 88) in WOD05 documentation # InstType (variable specific secondary headers) is probe type (0-16) # - table 3.2 (p 102) in WOD05 documentation # CorrType is derived from Sech54 (secondary header 54) # - "Needs depth fix" table 2.20 (p 93) in WOD05 documentation # Sech54 = 0: CorrType = 0 ! No correction needed # Sech54 = 1: CorrType = 1 ! XBT Hanawa et al (1995) correction (*) # Sech54 = 2: CorrType = 2 ! XBT Kisu et al (2005) correction (*) # Sech54 = 103: CorrType = 3 ! XCTD Johnson (1995) correction (*) # Sech54 = 104: CorrType = 4 ! XCTD Mizuno and Watenabe (1998) correction (*) # Sech54 = -1: CorrType = 9 ! Insufficient information - no correction # (* We correct the depths of profiles with CorrType 1 and 2, # for CorrType 1 this is modfied for "cold" profiles, # see section 3.1 of Ingleby and Huddleston (2006). # We do not currently correct CorrType 3 and 4 - XCTDs.) # # SCIENTIFIC_CALIB_COEFFICENT: # Factor used to multiply XBT depths # # It was discovered (August 2007) that the values of these two # variables are incorrect in the NetCDF files due to an indexing error. WMO_INST_TYPE: we use a simplified subset of WMO code table 1770 401 ! XBT/MBT 741 ! TESAC 831 ! ARGO 820 ! BUOYS - An observation is removed as a duplicate if it is within 0.2 deg latitude and longitude and within an hour of another. They are not included in the NetCDF files. If the same observation is available from GTSPP and WOD01 the WOD01 version will be used. This also performs some thinning in data dense areas on the grounds that current data assimilation systems don't have the resolution to make good use of very dense data. - High resolution CTD and XBT data is vertically thinned and this is described in the OQCpaper document (section 2.2). We are supplying a maximum of 150 levels of observation data. (Note that you should read the number of levels from the NetCDF header information rather than assume it is 150, it may be less particularly for earlier decades.) - Super-obs are formed from data that is from the same place and platform reporting frequently. This is particularly the case for TRITON data (West Pacific, 2000 onwards). The original quasi-hourly data is not in the NetCDF file but the super-obs are (indicated by the 4th character of the data state indicator = "S").