ISAS_V4.1b : Description of the method and user manual.
Contents
1. nc WMO INST TYPE ncchar N PROF STRING4 3720 elements nc WMO INST TYPE long name ncchar Coded instrument type nc WMO INST TYPE conventions ncchar WMO code table 1770 instrument type nc JULD ncdouble N PROF 930 elements nc JULD long name ncchar Julian day UTC of the station relative to REFERENCE DATE TIME nc JULD units ncchar days since 1950 01 01 00 00 00 UTC nc JULD conventions ncchar Relative julian days with decimal part as parts of day nc JULD FillValue_ ncdouble 999999 nc LATITUDE ncdouble N PROF 930 elements nc LATITUDE long name ncchar Latitude of the station best estimate nc LATITUDE units ncchar degree north nc LATITUDE FillValue_ ncdouble 99999 nc LATITUDE valid_min ncdouble 90 nc LATITUDE valid max ncdouble 90 nc LONGITUDE ncdouble N PROF 930 elements nc LONGITUDE long name ncchar Longitude of the station best estimate nc LONGITUDE units ncchar degree east nc LONGITUDE FillValue ncdouble 99999 nc LONGITUDE valid min ncdouble 180 nc LONGITUDE valid max ncdouble 180 nc DEPH ncfloat N LEVELS 152 elements nc DEPH long name ncchar Depth nc DEPH units ncchar meter nc DEPH FillValue_ ncfloat 99999 nc DEPH valid_min ncdouble 0 30 Ei H valid max nc2. 1600 1800 2000 i 4 4 0 5 10 15 20 25 30 salinity 200 400 600 800 1000 1200 1400 1600 1800 2000 35 40 60 F 80 100 F 120 F 140 160 f 35 4 36 2 36 4 35 8 36 35 2 35 6 L 180 10 15 20 25 30 Additional plots are saved in the subdirectory ctrl for one profile of the group showing in blue the raw data in red the interpolated data in green the interpolated data with QC higher 22 than the threshold and that will be excluded The plot title include the name of the corresponding raw data file and the DC_reference number Alertes When data points are excluded a plot is created in the directory alert std and the profile reference is added to the list in the directory alert list The plot shows the temperature and salinity data points in blue the climatology in black the corrected standard deviation criteria as dashed line In red the points excluded by the climatology test and in green the points ecluded by the spike test The plot title gives the DC reference of the profile 1 202458 Error dim or spike 3045982 Error clim or spike 3024589 400 400 800 800 Z 800 800 fe 1000 1000 1200 1200 1400 1400 5 1600 SR 1800 2000 1 1 4 L J 0 5 10 15 20 25 30 35 40 2000 4 4 Preprocessing PRE OA select the data that will be used to perform the analysis over each area All data
3. 930 elements 29 nc DIRECTION long name ncchar Direction of the station profiles nc DIRECTION conventions ncchar A ascending profiles D descending profiles nc DATA CENTRE ncchar N PROF STRING2 1860 elements nc DATA CENTRE long name ncchar Data centre in charge of float data processing nc DATA CENTRE conventions ncchar GTSPP table nc DC REFERENCE ncchar N PROF STRING32 29760 elements nc DC_REFERENCE long name ncchar Station unique identifier in data centre nc DC_REFERENCE conventions ncchar Data centre convention nc DATA STATE INDICATOR ncchar N PROF STRING4 3720 elements nc DATA STATE INDICATOR long name ncchar Degree of processing the data have passed through nc DATA STATE INDICATOR conventions ncchar OOPC table nc DATA MODE ncchar N PROF 930 elements nc DATA MODE long name ncchar Delayed mode or real time data nc DATA MODE conventions ncchar R real time D delayed mode nc INST ENCE ncchar N PROF STRING64 59520 elements nc INST ENCE long name ncchar Instrument type nc INST_ ENCE conventions ncchar Brand type serial number
4. c STRING256 256 c STRING64 64 c STRING32 32 c STRING16 16 c STRING8 8 c STRING4 4 c STRING2 2 c N PROF 930 c N LEVELS 152 c RP NB PROF 995 6 Variables and attributes nc DATA TYPE nc DATA TYPE ncchar STRING16 16 elements comment ncchar Data type nc FORMAT VERSION ncchar STRING4 4 elements nc FORMAT VERSION comment ncchar File format version nc REFERENCE DATE TIME ncchar DATE TIME 14 elements nc REFERENCE DATE TIME comment ncchar Date of reference for Julian days nc REFERENCE DATE TIME conventions ncchar YYYYMMDDHHMISS nc PI NAME ncchar N PROF STRING64 59520 elements nc PI NAME comment ncchar Name of the principal investigator nc PLATFORM NUMBER ncchar N PROF STRING8 7440 elements nc PLATFORM NUMBER long name ncchar Float unique identifier nc PLATFORM NUMBER conventions ncchar WMO float identifier QASTIIII nc CYCLE NUMBER nclong N PROF 930 elements nc CYCLE NUMBER long name ncchar Float cycle number nc CYCLE NUMBER conventions ncchar 0 N 0 launch cycle if exists 1 first complete cycle nc CYCLE NUMBER FillValue_ nclong 99999 nc DIRECTION ncchar N PROF
5. within the area mask and the time interval defined by date AMPL OA are selected At this stage data might be excluded on the instrument type criteria INST EXCL LIST 4 4 1 Running preoa After setting the parameters in the preoa block of the configuration file Preoa can be launched in the matlab window config fname myanalysis prepana confisas OA_ config ISAS txt PREOA main config fname 15 01 2006 TEMP An example of perl script to run PREOA_main over several month and years is given in isas v4 perl Type perl preoa pl 23 4 4 2 The output PREOA outputs are e The temporary files fld et dat for each area placed in the directory dir run preoa The fld files contain the empty anomaly filed for the area on the grid The dat files contain the data to be used by the analysis Naming convention are as follows OA_YYYYMMDD _ iarea_typ PARAM nc e OA identifier for optimal analyse e YYYYMMDD analysis date e iarea area number e typ identifier dat ou fld e PARAM TEMP ou PSAL A copy of the dat and fld files in the data subdirectory of DIR OA CALCUL if option copy preoa 1 is set in the configuration file The files TEMP in ou PSAL in that contain the list of the areas to be processed These files are created by the program PREOA creat configin and copied in the subdirectory config of DIR OA CALCUL if option creat in preoa 1 is set in the configurati
6. COARDS ncechar IFREMER har ARIVO char 20071123T102406 ncchar ISAS V4 0 POSTOA ncchar ARRAGLO2 Tarragl02 har Fabienne Gaillard char 20020115 echar 88 77 0105 4 y echar UVT Rel 2245 cchar 180 echar 7925 Dy ibutes 32 nc time ncfloat time 1 element nc time units ncchar days since 1950 01 01 UTC 00 00 00 nc latitude ncfloat latitude 500 elements nc latitude units ncchar degree north nc latitude valid min ncfloat 90 nc latitude valid max ncfloat 90 nc longitude ncfloat longitude 720 elements nc longitude units ncchar degree east nc longitude valid_ min ncfloat 180 nc longitude valid_max ncfloat 180 nc depth ncshort depth 152 elements nc depth units ncchar m nc depth positive ncchar down nc depth valid min ncshort 0 nc depth valid max ncshort 2000 nc TEMP neshort time depth latitude longitude 54720000 elements nc TEMP long name ncchar Temperature nc TEMP units ncchar degree Celsius nc TEMP valid min ncfloat 23000 nc TEMP valid_max ncfloat 20000 nc TEMP FillValue_ ncshort 32767 nc TEMP add offset ncfloat 20 nc TEMP scale factor ncfloat 0 00100000004749745 nc TEMP
7. GTSPP table nc DC REFERENCE ncchar N PROF STRING32 131488 elements nc DC_REFERENCE long name ncchar Station unique identifier in data centre nc DC_REFERENCE conventions ncchar Data centre convention nc DATA STATE INDICATOR ncchar N PROF STRING4 16436 elements nc DATA STATE INDICATOR long name ncchar Degree of processing the data have passed through nc DATA STATE INDICATOR conventions ncchar OOPC table nc DATA MODE ncchar N PROF 4109 elements nc DATA MODE long name ncchar Delayed mode or real time data nc DATA MODE conventions ncchar R real time D delayed mode nc INST REFERENCE ncchar N PROF STRING64 262976 elements 34 nc INST REFERENCE long name ncchar Instrument type nc INST REFERENCE conventions ncchar Brand type serial number nc WMO INST TYPE ncchar N PROF STRING4 16436 elements nc WMO INST TYPE long name ncchar Coded instrument type nc WMO INST TYPE conventions ncchar WMO code table 1770 instrument type nc JULD ncdouble N PROF 4109 elements nc aren fiat ee ncchar Julian day UTC of the station relative t
8. OA PATH S TOOLBOXPATH setenv MATLAB MATHOME MATHOME bin S MATHOME etc set path SMATLAB Spath alias matlab S MATHOME bin matlab 15 4 2 Configuration file for pre and post processing Before starting the analysis the various path file names and parameters must be defined This is done through the configuration file oa config isas txt that we recommend be placed in the directory confisas An example of configuration file is given in isas v4 doc config ole ole Warning Lines starting with are comment lines the others are read to define the configuration on those do not leave spaces if not required by syntax do not write comments o ole o ol ole ole 2 Standart configuration directories and file names DIR CONFSTD home machine dir run isas isas v4 1 confstd nam clim ISASW 4 nam std ISASW 52 STD nam bathy bathy GLOBALOS V4 0 nc Directory for raw data IR RAW ROOT home7 machine raw data CO DMOCGLO1 iw ole Directory dor standardized data note that a same std dataset can be used for different analysis R STD ROOT home machine dir run ana arragl03 ISAS RESU std ol iw HA ole Directory for the analysis results IR ANA RESU home machine dir run ana arragl05 ISAS RESU iw o Directory for the analysis processing log files plots temporary files DIR ANA RUN home machine dir run ana arra
9. PROF N LEVELS 624568 elements nc TEMP ERR UR long name ncchar Error from unresolved scales nc TEMP ERR UR FillValue ncfloat 99999 nc TEMP ERR UR units ncchar degree Celsius nc TEMP RESID ncfloat N PROF N LEVELS 624568 elements nc TEMP RESID long name ncchar Residuals nc TEMP RESID FillValue_ ncfloat 99999 nc TEMP RESID units ncchar degree Celsius nc RP_DC_REFERENCE ncchar RP NB PROF STRING32 131488 elements nc RP DC REFERENCE long name ncchar DC reference of raw profiles nc RP DC REFERENCE R ncchar RP NB PROF STRING32 131488 elements nc RP DC REFERENCE R long name ncchar DC_reference of STD profiles 36
10. bathymetry file of the NGDG Bathy Etopo2 nc The interpolation is done using the median of the 4 surrounding points 2 5 Areas and masks For the practical implementation of the method the global ocean has been divided in areas that define the group of points to be processed at once To each area we associate a mask defining the area where the information can be used This allows for example to exclude the Mediterranean while analysing the Gulf of Cadiz 2 6 The analysis steps The main processing steps are STD Data with the valid quality control flag QC are selected from the raw data files and interpolated on standard levels A new flag representing the quality of the interpolation is associated with the each data PRE OA This pre processing step gathers all data and statistical information that will be used for the analysis of each area defined in the configuration OA The analysis is performed over each area as described in the method section above POST OA This post processing step gathers the results over each area to form the full 3D field Two files are produced one holding the 3D fields and estimation error the other holding the data and residuals 11 3 The directories 3 1 program directory This directotory contains all programs required to perform the analysis It is organized as follows 3 1 1 Configuration directory confstd Contains all files defining the standard configurationuse
11. can be averaged reduced into super profiles Each of these procesing are detailled below 4 3 1 1 Detection of erroneous data Two different test are succesively applied Distance to climatology A data point will be accepted if the value X verifies Xobs Xctim lt 1 STD az X 6z The scalar a crit_stdin the configuration file has been determined empirically it defines the distance allowed to the climatology The scalar z introduces an additional tolerance relative to the climatology In the vicinity of very strong stratification perfectly good data may differ strongly from the 19 climatology This is taken into account by introducing an additional tolerance proportional to the vertical gradient of the parameter Spike detection A data value is considered as a spike if the following conditions are filled Change of sign of the first derivative for at least one point before or after the point Second derivative criteria normalized by the median in the vicinity of the point 6 P Z 2 crit _ spike 2 mediane iane 4 3 1 2 Interpolation The high resolution data are bin averaged on the standard levels then the remaining levels are interpolated 4 3 1 3 Reduction superobs Data from the same platform which are close in time and space are averaged The control parameters are RED DXMAX Minimum distance in kilometers RED DTMAX minimum time difference in days RED QCMAX
12. comment ncchar Estimated by optimal interpolation nc pct variance ncshort time depth latitude longitude 54720000 elements nc pct variance long name ncchar Error on temperature percent variance nc pct variance units ncchar percent of a priori variance nc pct variance valid min ncfloat 0 nc pct variance valid max ncfloat 100 nc pct variance FillValue ncshort 32767 nc pct variance add offset ncfloat 0 nc pct variance scale factor ncfloat 1 6 3 Data file ncdump arragl02 20020115 dat TEMP nc 6 Generated 23 Nov 2007 44 44 ole ole m Oo nc netcdf arragl02 20020115 dat TEMP nc noclobber if isempty nc return end 6 Global attributes nc Last update ncchar 23 Nov 2007 10 23 05 nc SoftwareVersion ncchar ISAS V4 0 POSTOA ole ole Dimensions nc DATE TIME 14 nce STRING256 256 nc STRING64 64 nce STRING32 32 33 nc STRING16 16 nc STRING8 8 nc STRING4 4 nc STRING2 Zig nc N PROF 4109 nc N LEVELS 152 nc RP_ NB PROF 4109 6 Variables and attributes nc DATA TYPE ncchar ST
13. contient les fichiers de travail et les informations sur le d roulement de l analyse Ils peuvent tre effac s apr s v rification alert Contient un r pertoire list qui regroupe les listes d alertes mises par les diff rents niveaux de traitement et un r pertoire std qui contient les trac s des profils d tect s en alerte par le programme STD main logisas contient les fichiers logs produits par les diff rentes tapes plotisas contient les trac s produits par les diff rentes tapes preoa Contient fichiers fld et dat par zones pr par s par PREOA main Ces fichiers sont recopi s sur la machine de calcul o ils sont compl t s 14 4 Processing an anlysis 4 1 Matlab path Before starting the analysis the matlkab path need to be defined The environment is described in a file isas matlab env that we recommend be placed in the directory confisas See isas_v4 doc config for an example To launch type cd confisas source ISAS matlab env usr bin sh Set RER PE EE RCE PP See Matlab directory EEE EE eee Be Se ee setenv MATHOME home machine matlab dir matlab r2007b setenv TOOLBOXPATH home machine matlab dir outils matlab m map1 4 setenv ISAS HOME home machine isas dir setenv OA HOME ISAS HOME isas v4 1l isas mat 4 1b setenv OA PATH OA HOME std OA HOME preoa OA HOME postoa OA HOME share setenv MATLABPATH
14. elements FillValue ncchar 0O nventions ncchar Q where Q 0 9 ng name ncchar Quality on interpolated temperature Hae CE EE ncfloat N PROF N LEVELS 141360 elements PSAL FillValue_ ncfloat 99999 PSAL long name ncchar Salinity S78 interpolated on Z ncechar PSU valid min ncfloat 0 valid max ncfloat 60 al al ncfloat N_PROF N LEVELS 141360 elements Hi m FillValue_ ncfloat 99999 long_name ncchar Error on interpolated salinity units ncchar PSU valid min ncfloat 0 00100000004749745 EH EH EH HAA we eww nn YH He valid max ncfloat 10 ncfloat N_PROF N LEVELS 141360 elements FillValue_ ncfloat 99999 long_name ncchar Climatology reference of profile units ncechar PSU valid min ncfloat 0 valid max ncfloat 60 _CLIM STD ncfloat N PROF N LEVELS 141360 elements 31 6 2 Field file 3 ncdump arragl02 2 243 241 netcdf arragl02 isempty nc retur Global attributes CONVENTIONS ncch producer agengy project name creation time software version product version data set ncchar data manager nee estimate_date ne south latitude n north
15. l int rieur d un sous r pertoire les donn es sont organis es par ann e La d nomination des fichiers suit la convention de la documentaion Coriolis Antonio 2007 Un niveau mois bassin a t conserv pour assurer la compatibilit avec l organisation pr c dente mais son usage n es pas recommand 3 4 R pertoire confisas Contient le fichier de configuration de l analyse pour d finir les chemins et l ensemble des param tres ajustables de l analyse Un exemple de fichiers figure dans le r pertoire doc config model isas_matlab env chemins matlab oa_config isas txt fichier de configuration d finissant les param tres des programmes matlab tel qu il sera lu par isas mat TEMP 6 2006 in TEMP cnf PSAL cnf mod les de fichiers de configuration pour les programmes f90 recopier dans le r pertoire f90 correspondant 3 5 R pertoire dir_resu Ce r pertoire contient les r sultats de l analyse Ces r sultats sont distribu s dans trois sous r pertoires chacun contenant un r pertoire par ann e std fichiers contenant les donn es sur niveau standard ces fichiers correspondent aux fichiers raw mais peuvent tre regroup s par mois field Fichiers contenant les champs analys s sur la grille r guli re data fichiers contenant les donn es utilis es pour le calcul du champ field ainsi que les r sidus d analyse Dir_resu 13 3 6 R pertoire dir_run Ce r pertoire
16. latitude n west longitude n east longitude n ncc nc o ole Dimensions 1 LS 500 720 time depth latitude longitude nc nc nc nc 20 06 Variables and attr _20020115 fld TEMP nc nc PSAL CLIM STD FillValue ncfloat 99999 nc PSAL CLIM STD long name ncchar Standard deviation of climatology reference of profile nc PSAL CLIM STD units ncchar PSU nc PSAL CLIM STD valid min ncfloat 0 nc PSAL CLIM STD valid_max ncfloat 60 nc PSAL QC ncchar N PROF N LEVELS 141360 elements nc PSAL QC FillValue_ ncchar 0 nc PSAL QC conventions ncchar Q where Q 0 9 nc PSAL QC long name ncchar Quality on interpolated salinity nc RP DC REFERENCE ncchar RP NB PROF STRING32 31840 elements nc RP DC REFERENCE long name ncchar DC_ reference of raw profiles nc RP DC REFERENCE R ncchar RP NB PROF STRING32 31840 elements nc RP DC REFERENCE R long name ncchar DC_ reference of STD profiles nc RP TEMP QC STD ncchar RP NB PROF 995 elements nc RP TEMP QC STD long name ncchar Temperature QC flag from STD nc RP PSAL QC STD ncchar RP NB PROF 995 elements nc RP PSAL QC STD long name ncchar Salinity QC flag from STD 0020115 fld TEMP nc 6 Generated 23 Nov 2007 noclobber n end ar
17. xl4 4C C R d P P C C RY CZ eq 1 It should be noticed that this formalism provides at the same time an estimate of the misfit between observations and analysis also called analysis residuals y y R C R d C is the covariance matrix between analyzed points and data points C is the covariance matrix between data points and R is the error covariance matrix It combines the measurement error and the representativity error The error on the estimation is given by the diagonal of the P matrix usually normalized by the a priori variance o Co C R Cc a ae xi ii 2 2 Statistical information Statistical information on the field and data noise are introduced through the covariance matrices that appear in equation 1 We assume that the covariances of the analyzed field can be specified by a structure function modeled as the sum of two gaussians the first term i 1 corresponding to the large scale field LS the second i 2 to the mesoscale MS 2 d d dt C dx dy dt X _ 0 exp die a oe where dx dy dt are the space and time separation L L L the corresponding e folding ix iy scales The weight given to each ocean scale is controlled by the variances 0 The total variance is computed as the variance of the anomaly relative to the monthly reference field It is considered as the sum of four terms 2 2 2 2 2 Doi Ors tOus Our t Our where o an
18. F Gaillard R Charraud ARIVO Re N 08 03 ISAS_V4 1b Description of the method and user manual O W c IRD March 2008 Historique Auteur Mise a jour Date F Gaillard Cr ation du document V4 beta 03 02 2007 R Charraudeau V4 00 Version fran aise 23 11 2007 F Gaillard V4 01 Version frangaise 11 02 2008 F Gaillard V4 1b English version 19 03 2008 Content 1 Introduction 2 Method and configuration 2 1 2 2 2 3 2 4 2 5 2 6 Estimation method Statistical information The datasets Configuration Areas and masks The analysis steps 3 The directories 3 1 3 2 3 3 3 4 3 5 3 6 program directory The data directories R pertoire raw les donn es brutes R pertoire confisas R pertoire dir_resu R pertoire dir_run 4 Processing an anlysis 4 1 42 43 4 4 4 5 4 6 Matlab path Configuration file for pre and post processing Standardisation Preprocessing Analysis Post Processing 5 References 6 Annex 6 1 STD file 6 2 Field file 6 3 Data file CN 10 11 12 12 13 13 13 13 14 15 15 16 19 23 24 27 28 29 29 32 33 1 Introduction ISAS In Situ Analysis System is an analysis tool for the temperature and salinity fields originally designed for the synthesis of ARGO dataset it has been tested for the first time on the POMME area in the North East Atlantic 2000 It is developed and maintained at LPO La
19. FSTD bathy GLOBALOS V4 0 nc 300 300 21 covar ls x y t in km km days 21 covar ms t in days 1 4 var weigh LS S UR 4 Nee bare X Y Z t covariance dependency 1 yes 0 no 1 32 fact Variance 2 412 5 QC Max Mx std LL Cov max if gt 1 no oversampling test 33 11 oversample alpha fct test Si fct test lt 10 autorise l augment de l erreur partout 4 5 1 5 data Contains the fld and dat files created and optionnally copied by PREOA Those files will be completed by OA 25 4 5 2 Running ISAS_f90 The program can be run in interactive mode cd my analysis f90 calculateur isas f90 0A main lt config TEMP 2006 in It can also be launched in batch mode this allows to loop over dates and parameters The way batches are run is machine dependent Examples are given here for SGI ICE 8200 Launch with qsub my batch where my batch contains bin csh get the path for library MKL source usr share modules init csh module load cmkl recent setenv MKL SERIAL YES cd my analysis 90 foreach year 2003 2004 2005 2006 foreach month 1 23 45 67 8 9 10 11 12 foreach param TEMP PSAL date calculateur isas f90 0OA main lt config Sparam month year in date end end end 4 5 3 Outputs 4 5 3 1 err Contains a short log file with the list of processes files and any error message issued by the program This file must be screened carafull
20. RING16 16 elements nc DATA TYPE comment ncchar Data type nc FORMAT VERSION ncchar STRING4 4 elements nc FORMAT VERSION comment ncchar File format version nc REFERENCE DATE TIME ncchar DATE TIME 14 elements nc REFERENCE DATE TIME comment ncchar Date of reference for Julian days nc REFERENCE DATE TIME conventions ncchar YYYYMMDDHHMISS nc PI NAME ncchar N PROF STRING64 262976 elements nc PI NAME comment ncchar Name of the principal investigator nc PLATFORM NUMBER ncchar N PROF STRING8 32872 elements nc PLATFORM NUMBER long name ncchar Float unique identifier nc PLATFORM NUMBER conventions ncchar WMO float identifier QA9IIIII nc CYCLE NUMBER nclong N PROF 4109 elements nc CYCLE NUMBER long name ncchar Float cycle number nc CYCLE NUMBER conventions ncchar 0 N 0 launch cycle if exists 1 first complete cycle nc CYCLE NUMBER FillValue_ nclong 99999 nc DIRECTION ncchar N PROF 4109 elements nc DIRECTION long name ncchar Direction of the station profiles nc DIRECTION conventions ncchar A ascending profiles D descending profiles nc DATA CENTRE ncchar N PROF STRING2 8218 elements nc DATA CENTRE long name ncchar Data centre in charge of float data processing nc DATA CENTRE conventions ncchar
21. allow an additional 0 6 daT dZ or dS dZ distance to climatology ole ole o alpha_clim 2 o Criteria for spike detection based on second derivative Small value detects smaller spikes scrit_ spike temp 200 first pass scrit spike psal 200 scrit_ spike temp 800 second pass scrit_ spike psal 800 crit spike temp 800 crit spike psal 800 INT NB MIN Min number of points to perform interpolation 17 INT NB MIN 2 ole Parameters for reduction of nearby profiles creates super profiles ole ED DXMAX max distance in km ED DTMAX max time interval in days ED QCMAX QC max defined by STD preocess used to build the superprofiles AP AP ol An ole ole ED_DXMAX 15 ED_DTMAX 7 ED_QCMAX 4 fe Definition of default errors associated to each data type will be used only if no error is specified within the raw data file RR 0 03 05 2 OL 203 SQT 01 002 H D D lt w Oo eee R R R R R R ole ole ole PREOA Preprocessing ol May be used to overwrite STD name SPRF_STD ST_RAOAGLO1 _ Instrument type excluded SINST EXCL LIST 1 800 900 exclude XBTs of all types INST EXCL LIST INST _EXCL_LIST 1 800 900 List of areas to be analyzed REA LIST 101 141 201 241 301 388 401 403 D oe Z D D time interval in days look wi
22. at N PROF N LEVELS 624568 elements nc TEMP ERR ME FillValue ncfloat 99999 nc TEMP ERR ME long name ncchar Error on interpolated temperature nc TEMP ERR ME units ncchar degree Celsius nc TEMP ERR ME valid_min ncfloat 0 00100000004749745 nc TEMP ERR ME valid max ncfloat 10 nc TEMP CLIM ncfloat N PROF N LEVELS 624568 elements nc TEMP CLIM FillValue_ ncfloat 99999 nc TEMP CLIM long name ncchar Climatology reference of profile nc TEMP CLIM units ncchar degree Celsius nc TEMP CLIM valid min ncfloat 3 nc TEMP CLIM valid max ncfloat 40 35 nc TEMP CLIM STD ncfloat N PROF N LEVELS 624568 elements nc TEMP CLIM STD FillValue ncfloat 99999 nc TEMP CLIM STD long name ncchar Standard deviation of climatology reference of profile nc TEMP CLIM STD units ncchar degree Celsius nc TEMP CLIM STD valid min ncfloat 0 nc TEMP CLIM STD valid max ncfloat 40 nc TEMP QC ncchar N PROF N_LEVELS 624568 elements nc TEMP QC FillValue_ ncchar 0 nc TEMP QC conventions ncchar Q where Q 0 9 nc TEMP QC long name ncchar Quality on interpolated temperature nc TEMP ERR UR ncfloat N
23. boratoire de Physique des Oc ans within the ARIVO project and has been made available to the Coriolis datacenter The analysis is performed on the daily datasets prepared by Coriolis for the various operational users The background and statistical information required to complement the observations are provided with the software as part of the configuration Charraudeau et Gaillard 2007 For each analysis date the results are provided as two NetCDF files one holding the data and analysis residuals the other holding the gridded fields and estimation error expressed as percentage of a priori variance This document describes the method and how it is implemented within ISAS The main steps of the process are detailed and examples of configuration files are provided 2 Method and configuration 2 1 Estimation method ISAS uses estimation theory for mapping of a scalar field on a regular grid from sparse and irregular data Bretherton et al 1976 The interpolated field represented by the state vector x is constructed as the departure from a reference field values at the grid points x This reference is derived from previous knowledge climatology or forecast Only the unpredicted part of the observation vector or departure from the reference field at the data points called innovation is used d x The analyzed field x is obtained as a linear combination of the innovation and is associated with a covariance matrix P x
24. d 6 are the two terms appearing in the equation for the covariance structure se 2 2 4 p 2 The remaining sum 0 1 the total error variance 0 corresponds to the 2 measurement errors and represents small scales unresolved by the analysis and lt hy 2 considered as noise sometimes called representativity errors A unique O profile has been computed from the measurement errors of the standard database and substracted from the total variance to obtain the ocean variance first three terms of the sum The ocean variance is adjusted to remain larger than o and can be multiplied by a factor to account for the under sampling of the ocean variability We express the variances associated to each scales as a function of the ocean variance by introducing normalized weights 2 2 Ors Wis 2 2 Ous WysO 2 2 Our WuirO Wis Wus Wir 1 The free parameters of the system are the weights that define the distribution of variance over the different scales The error matrix combines the measurement error and the representativity error due to unresolved scales it is assumed diagonal although this is only a crude approximation since both errors are likely to be correlated for measurements obtained with the same instrument or within the same area and time period The large scale lengths are taken isotropic and equal to 300km the target Argo resolution the corresponding time scale is set to 3 weeks The meso scale leng
25. d by the analysis Charraudeau et Gaillard 2007 e Bathymetry Climatology annual and monthly Variances Covariance scales Definition for analysis areas and masks 3 1 2 Documentaion directory doc Contient les documentations sur la description et la mise en oeuvre des programmes ISAS V4 config pdf ISAS V4 prog pdf 3 1 3 Matlab scripts isas_mat Contains the matlab scripts used for the standardisation the pre and post processing 3 1 4 Fortran programs isas_f90 Contains source codes makefile and executable for the analysis 3 1 5 Directory perl Contains perl script that allow to loop over different analysis dates and parameters 12 3 2 The data directories La mise en oeuvre d ISAS se base sur quatre r pertoires principaux Le r pertoire dir _raw contient l ensemble des donn es brutes fournies par Coriolis Le r pertoire dir _confisas contient les fichiers de configuration Le r pertoire dir resu contient les r sultats des analyses Le r pertoire dir run contient les alertes trac s fichiers log et fichiers interm diaires des analyses Les deux premiers r pertoires sont fournis par l utilisateur les deux autres sont cr es par l analyse 3 3 R pertoire raw les donn es brutes Ce dossier est pr par par l utilisateur et doit respecter l organisation qui suit Contient un sous r pertoire par lot de donn es homog nes issues d un m me fournisseur A
26. double 10000 P units lt 2 ig fe ie AA woe EH a Hi El A m ncfloat N PROF N LEVELS 141360 elements P FillValue_ ncfloat 99999 P long name ncchar Ocean temperature T90 interpolated on Z ncechar degree Celsius P valid min ncfloat 3 P valid max ncfloat 40 tA Ra R 0 ZZ Zo TETE oii 5 HAA Ae EH ncfloat N PROF N LEVELS 141360 elements FillValue ncfloat 99999 iong name ncchar Error on interpolated sunits ncchar degree Celsius valid min ncfloat 0 00100000004749745 valid max ncfloat 10 ncfloat N PROF N LEVELS 141360 elements M FillValue_ ncfloat 99999 iong name ncchar Climatology reference of profile M units ncchar degree Celsius M valid min ncfloat 3 valid max ncfloat 40 HD HE Et LE 9 AMR ERR TD 2s fe Ste he Y QAaAAAAA m me PRP RE Qn QQ A U Q 0 KAA HE oO 2 2 25 22 TRI SEE AN ncfloat N_PROF N_LEVELS 141360 elements FillValue_ ncfloat 99999 J Onon name ncchar Standard deviation of climatology ile mie le units RS Nee ea armen wit valid min ncfloat 0 valid max ncfloat 40 ncchar N PROF N LEVELS 141360
27. e log files and plots The results are saved in two files In DIR ANA RESU data the NetcCDF dat file that contains the data and residuals used by all the areas In DIR ANA RESU field the NetCDF file fld that contains the global 3D gridded fields and error File naming convention are as follows nameana YYYYMMDD _ typ PARAM nc e nameana analysis identifier ANA NAME in the config file e YYYYMMDD analysis day e PR identifier dat ou fld e PARAM TEMP or PSAL 27 5 References Antonio J 2007 Outils d analyse de donn es in situ ISAS formats et nomenclatures en version 3 7 Document provisoire Coriolis Gaillard F et E Autret 2006 Climatologie et statistique de l Atlantique Nord Projet GMMC 2003 Bretherton F R Davis and C Fandry 1976 A technique for objective analysis and design of oceanic experiments applied to Mode 73 Deep Sea Research 23 1B 559 582 Charraudeau R et F Gaillard 2007 ISAS_V4 Mise en place de la configuration Rapport LPO 07 09 88 p 28 6 Annex 6 1 STD file o ole ncdump ST _RAOAGLO1 20020100 PR PF nc 47 32 m O nc netcdf ST RAOAGLO1 20020100 PR PF nc if isempty nc return end 6 Global attributes Generated 23 Nov 2007 noclobber nc Last update ncchar 05 Nov 2007 17 01 58 nc SoftwareVersion ncchar ISAS V4 03 STD Dimensions c DATE TIME 14
28. gl105 ISAS RUN ole directory for 90 DIR_OA_CALCUL home2 computer user OA run arragl05 ole AREA LIMITS 81 80 180 180 ole ole STD Standardisation ole ole ocean list list of directories to explor DIR RAW LIST none ole Specifique Coriolis 16 2 TYP LIST List of file types to process TYP LIST PR TE PR XB PR CT PR MO PR PF PR BA PRF RAW identifier for raw files PRF RAW CO DMOCGLO1 PRE STD identifier for STD files PRF_STD ST_DMQCGLO1_ month _grp 1 all data within a month are grouped no group 0 month _grp 1 all RAW data from different basins are grouped in a single STD needed for compatibility with previous versions of Coriolis file naming convention ocean _grp 1 use adjust 1 use adjusted value if exist else 0 use_adjust 0 5 QC TS flags ok Temp and Psal QC_ZP flags ok Pres and depth sfOC_ XY flags ok Position and date QC_TS 125 QC_ZP 0 125 OC XY 0 125 ole Parameter for the profile control o ole Criteria for comparing to climatology crit std clim number of standard deviation allowed relative to the climatology scrit_std_clim 8 first pass 2 Scrit std clim 15 second pass ole ole crit _std_clim 15 ole alpha clim Stratification correction A correction proportionnal to the vertical gradient is added to the standard deviation If alpha_clim 0 6 we
29. is launched from a steaming ship It measures temperature and salinity in the case of XCTD the measurement depth is deduced from the XBT fall rate The accuracy is 0 1 C and most XBT reach 800 m e CTD This high quality measurement is obtained from a research vessel in the context of a scientific cruise Pressure and temperature sensors are carefully calibrated and water samples are taken to adjust the salinity measurement Standard procedure were defined for the WOCE experiment they lead to accuracies of 0 001 C and 0 001 PSU e Time series Time series of pressure temperature and salinity are recorded at high time resolution hours from sensors installed on fixed points mooring or drifting buoys The measurement depth is usually constant The sensors are similar to those used on the profiling floats 2 4 Configuration The configuration proposed with ISAS V4 is describred in detail in Charraudeau and Gaillard 2007 The horizontal grid is 2 degree Mercator limited to 77S 77N It is thus isotropic and the resolution increases with latitude from 0 5 at equator to 0 1125 77N The vertical levels are given below 12 grid Mercator 05 F E F N i STD_LEVEL A 0 3 5 5 100 110 10 800 820 20 2000 STD LEVEL B 2020 20 2500 2550 50 58001 The bathymetry is an interpolation over our grid of the file etopo2bedmap nc produced by MERCATOR from the 2 minutes
30. maximum QC flag after standardisation 4 3 2 Running STD After setting the parameters of the STD block in the configuration file STD main can be launched in the matlab execution window config fname my analysis confisas OA config ISAS txt to process 10 days starting on july 14 2006 dd 14 mm O7 yyyy 2006 nb days 10 To process a full month ex july 2006 dd 0 mm 07 yyyy 2006 nb days 0 or anything this value is ignored STD main config fname dd mm yyyy nb days An example of perl script to run STD main over several month and years is given in isas v4 perl std Type perl std pl 20 4 3 3 Outputs 4 3 3 1 Data files on standard levels Results are writen as NetCDF files in the directory dir resu std The naming convention is as follows ST CCCCCCCC YYYYMMDD PR YY nc e ST identifies STD data e CCCCCCCC dataset name e YYYYMMDD date of observation if day 00 file contains the whole month e PR identifies profile data e YY data types according to Coriolis convention 4 3 3 2 Listing log file The log file can be found in dir run logisas Example of log file gt gt gt gt gt gt gt Running ISAS V4 0 STD Last update 18 Apr 2007 18 27 25 KKKKKKKK File 1 KKKKKKKK File processed CO_TST 20040701 PR TE Number of profiles read 3 Number of valid profiles QC_posdate 3 Number of profiles kept 3 Nu
31. mber of profiles per type TMs 0 Se 20 ES 3y total 3 Number of profiles without depth 1 Number of profiles per type Te 0z Sas 0 TSE 335 COEAL 3 3 profiles CPU time total seconds 16 20 CPU read 1 76 depth 1 48 STD check 10 53 Stdlev 1 62 KKKKKKKK File 31 KKKKKKKK File processed CO_TST 20040731 PR TE Number of profiles read 18 Number of valid profiles QC posdate 18 Number of profiles kept 18 Number of profiles per type Ts Gy Sue 0 RSs 125 total T8 Number of profiles without depth 12 Number of profiles per type Te 6 SE Oy TS 12 total 18 18 profiles CPU time total seconds 18 40 CPU read 0 96 depth 1 77 STD check 14 41 Stdlev 1 94 112 Multiple profiles platform 13009 nb av 2 platform 13009 nb av 2 platform 15001 nb av 2 platform CGDV nb av 3 platform CGDV nb av 2 Final number of profiles per type 261 Ts 35 Ss 90 TS 226 processing time red 1 53 prep 0 05 write 6 00 21 4 3 3 3 Control plots Different types of plots can be found in dir run plotisas std Standard plot level PLOT_CONV 1 A plot showing all profiles is produced ST S T 0040700 R gT nc temperature 400 F 600 F 800 F 1000 F 1200 F 1400 F 1600 F 1800 F 2000 0 25 High level plot PLOT_CONV gt 1 CO ST 0040701 R A 1798588 Or r r r ak MO ht ape 200 fa x 400 4 600 800 F i 1000 F 4 1200 1400
32. o REFERENCE DATE TIME is nc JULD units ncchar days since 1950 01 01 00 00 00 UTC nc Sal bey are ncchar Relative julian days with decimal part as parts of day nc JULD FillValue_ ncdouble 999999 nc LATITUDE ncdouble N PROF 4109 elements nc LATITUDE long name ncchar Latitude of the station best estimate nc LATITUDE units ncchar degree north nc LATITUDE FillValue_ ncdouble 99999 nc LATITUDE valid_min ncdouble 90 nc LATITUDE valid max ncdouble 90 nc LONGITUDE ncdouble N PROF 4109 elements nc LONGITUDE long name ncchar Longitude of the station best estimate nc LONGITUDE units ncchar degree east nc LONGITUDE FillValue ncdouble 99999 nc LONGITUDE valid min ncdouble 180 nc LONGITUDE valid max ncdouble 180 nc DEPH ncfloat N LEVELS 152 elements nc DEPH long name ncchar Depth nc DEPH units ncchar meter nc DEPH FillValue_ ncfloat 99999 nc DEPH valid_min ncdouble 0 nc DEPH valid_max ncdouble 10000 nc TEMP ncfloat N PROF N LEVELS 624568 elements nc TEMP FillValue_ ncfloat 99999 nc TEMP long name ncchar Ocean temperature T90 interpolated on Z levels nc TEMP units ncchar degree Celsius nc TEMP valid min ncfloat 3 nc TEMP valid max ncfloat 40 nc TEMP ERR ME ncflo
33. on file A log file Attention Erase all files from previous runs in preoa directory berore running PREOA The process that copies the files on the fortran computer takes all files found in the directory files from previous runs which have not been overwritten will be taken into acount and may produce inconsistencies 4 5 Analysis 4 5 1 Subdirectories on OA computer 24 4 5 1 1 Directory isas_f90 The objective analysis has been coded in fortran 90 to improve the memory use The source codes makefile and executable from isas_f90 must be copied in this directory and recompiled for the computer if necessary 4 5 1 2 Directory CONFSTD This directory contains files for e the bathymetry e the covariance scales e thea priori variance for temperature e thea priori variance for salinity 4 5 1 3 Analysis directory The parent directory is named DIR_OA_CALCUL 4 5 1 4 config Contains the list of area copied by PREOA for each parameter For example TEMP_1_2006 in contains the list of NetCDF files to be processed data 2006 TEMP S log 2006 TEMP config TEMP cnf 164 number of files area to process OA 20060101 101 dat TEMP nc OA 20060101 102 dat TEMP nc It should also contain the configuration file for the analysis TEMP cnf or PSAL cnf TEMP home2 mycomputer user OA run CONFSTD ISASW 52 STD TEMP nc home2 mycomputer user OA run CONFSTD ISASW 4 ann COVS nc home2 mycomputer user OA run CON
34. th is proportionnal to the Rossby Radius computed from the annual climatology In the equatorial band this value is bounded by the large scale length in the zonal direction and by the length scale of the adjacent zones in the meridional direction At high latitudes it is bounded by the resolution of the estimation grid 2 3 The datasets We briefly describe here the characteristics of the data types taken into account at the moment These dataset have different accuracy resolution and sampling that depend mostly on the sensor and on the storage and transmission system used Temperature and salinity measurements are obtained from autonomous instruments drifting or anchored or from instruments deployed with a ship The data are transmitted in real time by satellite or in delayed mode The main characteristics of the most common instruments are given below e Profiling floats The autonomous floats are part of the ARGO program they collect vertical profiles of temperature and salinity as a function of pressure between their maximum pressure usually 2000 dbars and the surface At the end of the profile that takes nearly 5 hours the profiler transmits the data to a satellite and dives toward its parking depth 1000 dbars waiting for the next cycle 10 days later Nominal accuracy of the data is assumed to be 0 01 C and 0 01 PSU At present time a vertical profile is described by approximately 100 pts e XBT An eXpendable BathyThermograph
35. thin day AMPL OA and day AMPL OA MPL OA 30 D o Copies NetCDF files on the fortran computer DIR_OA_CALCUL copy _preoa 1 Creates the input file for the fortran computer creat_in_preoa 1 o ole ole POSTOA Post Processing ole ole AR Arivo RA Re analysis AT Atlantic X1 analysis identifier 18 ANA NAME arragl05 Reference climatology month or year clim ref oa M month or clim ref oa Y year clim ref oa M Spatial filtering on areas boundaries applied on points with err gt err max filter err max 80 data set DATA _SET CO_DMQCGLO1 Sproduct version PRODUCT VERSION arragl05 SProject name PROJECT NAME ARIVO Data manager DATA MANAGER Fabienne Gaillard 2 splot Plot options 0 no plots 3 maxplots pause PLOT_CONV 1 LANG En 4 3 Standardisation 4 3 1 Description The first step in the analysis is an interpolation of the raw data on the standard levels of the analysis grid It is partly independent of the analysis in the sense that the dataset produced can be used for different analysis A new QC is introduced it represents the quality of the interpolation the closest to a measured value the lowest the QC flag value To avoid spoiling the analysis with eroneous data a control is performed before the interpolation Finally oversampled points such as repeated fixed points CTD drfting buoys mooring
36. y to chck that the processing has ended normally 4 5 3 2 log The log file contain statistical information on the processing for each area and each level of analysis KERERE Area 218 Nb profiles 85 Nb level 152 Nb analysis points nlon nlat 24 28 cpu distance calculations 0 038 Level 1 Nb ana points 535 Nb ovsamp 0 Nb data 58 ano max 11 892 inov min 2 729 inov max 1 526 fld min 1 566 fld max 0 076 cond 0 4999E 00 cpu Analysis 0 006 Level 2 Nb ana points 535 Nb ovsamp 0 Nb data 58 ano max 11 892 inov min 2 677 inov max 1 478 fld min 1 527 fld max 0 096 cond 0 5000E 00 cpu Analysis 0 006 26 Level Bin ek XXXXX cpu total area 1 582 4 5 3 3 data files The analysis results are store in the fld and dat data files that now contain the gridded anomaly fields and corresponding error and the data residuals respectively 4 6 Post Processing During this last part of the processing POSTOA concatenates all processed areas and datasets It also convert anomalies to absolute values The files are read in DIR OA CALCUL and results are written in DIR ANA RESU 4 6 1 Running POSTOA POSTOA is launched with the same arguments as PREOA In the matlab window config fname myanalysis prepana confisas OA_ config ISAS txt POSTOA main config fname 15 01 2006 TEMP Perl script are also provided 4 6 2 Outputs The processing can be checked looking at th
Download Pdf Manuals
Related Search