| 1 | % ================================================================ |
|---|
| 2 | % Chapter observation operator (OBS) |
|---|
| 3 | % ================================================================ |
|---|
| 4 | \chapter{Observation and model comparison (OBS)} |
|---|
| 5 | \label{OBS} |
|---|
| 6 | |
|---|
| 7 | Authors: D. Lea, M. Martin, K. Mogensen, A. Vidard, A. Weaver... % do we keep that ? |
|---|
| 8 | |
|---|
| 9 | \minitoc |
|---|
| 10 | |
|---|
| 11 | |
|---|
| 12 | \newpage |
|---|
| 13 | $\ $\newline % force a new line |
|---|
| 14 | |
|---|
| 15 | The observation and model comparison code (OBS) reads in observation files (profile |
|---|
| 16 | temperature and salinity, sea surface temperature, sea level anomaly, sea ice concentration, |
|---|
| 17 | and velocity) and calculates an interpolated model equivalent value at the observation |
|---|
| 18 | location and nearest model timestep. The resulting data are saved in a ``feedback'' file (or |
|---|
| 19 | files). The code was originally developed for use with the NEMOVAR data assimilation code, but |
|---|
| 20 | can be used for validation or verification of model or any other data assimilation system. |
|---|
| 21 | |
|---|
| 22 | The OBS code is called from \np{opa.F90} for model initialisation and to calculate the model |
|---|
| 23 | equivalent values for observations on the 0th timestep. The code is then called again after |
|---|
| 24 | each timestep from \np{step.F90}. To build with the OBS code active \key{diaobs} must be |
|---|
| 25 | set. |
|---|
| 26 | |
|---|
| 27 | For all data types a 2D horizontal interpolator is needed to interpolate the model fields to |
|---|
| 28 | the observation location. For {\em in situ} profiles, a 1D vertical interpolator is needed in |
|---|
| 29 | addition to provide model fields at the observation depths. Currently this only works in |
|---|
| 30 | z-level model configurations, but is being developed to work with a generalised vertical |
|---|
| 31 | coordinate system. Temperature data from moored buoys (TAO, TRITON, PIRATA) in the |
|---|
| 32 | ENACT/ENSEMBLES data-base are available as daily averaged quantities. For this type of |
|---|
| 33 | observation the observation operator will compare such observations to the model temperature |
|---|
| 34 | fields averaged over one day. The relevant observation type may be specified in the namelist |
|---|
| 35 | using \np{endailyavtypes}. Otherwise the model value from the nearest timestep to the |
|---|
| 36 | observation time is used. |
|---|
| 37 | |
|---|
| 38 | The code is controlled by the namelist \textit{nam\_obs}. See the following sections for more |
|---|
| 39 | details on setting up the namelist. |
|---|
| 40 | |
|---|
| 41 | Section~\ref{OBS_example} introduces a test example of the observation operator code including |
|---|
| 42 | where to obtain data and how to setup the namelist. Section~\ref{OBS_details} introduces some |
|---|
| 43 | more technical details of the different observation types used and also shows a more complete |
|---|
| 44 | namelist. Section~\ref{OBS_theory} introduces some of the theoretical aspects of the |
|---|
| 45 | observation operator including interpolation methods and running on multiple processors. |
|---|
| 46 | Section~\ref{OBS_obsutils} introduces some utilities to help working with the files produced |
|---|
| 47 | by the OBS code. |
|---|
| 48 | |
|---|
| 49 | % ================================================================ |
|---|
| 50 | % Example |
|---|
| 51 | % ================================================================ |
|---|
| 52 | \section{Running the observation operator code example} |
|---|
| 53 | \label{OBS_example} |
|---|
| 54 | |
|---|
| 55 | This section describes an example of running the observation operator code using |
|---|
| 56 | profile data which can be freely downloaded. It shows how to adapt an |
|---|
| 57 | existing run and build of NEMO to run the observation operator. |
|---|
| 58 | |
|---|
| 59 | \begin{enumerate} |
|---|
| 60 | \item Compile NEMO with \key{diaobs} set. |
|---|
| 61 | |
|---|
| 62 | \item Download some ENSEMBLES EN3 data from |
|---|
| 63 | \href{http://www.hadobs.org}{http://www.hadobs.org}. Choose observations which are |
|---|
| 64 | valid for the period of your test run because the observation operator compares |
|---|
| 65 | the model and observations for a matching date and time. |
|---|
| 66 | |
|---|
| 67 | \item Add the following to the NEMO namelist to run the observation |
|---|
| 68 | operator on this data. Set the \np{enactfiles} namelist parameter to the |
|---|
| 69 | observation file name: |
|---|
| 70 | \end{enumerate} |
|---|
| 71 | |
|---|
| 72 | %------------------------------------------namobs_example----------------------------------------------------- |
|---|
| 73 | \namdisplay{namobs_example} |
|---|
| 74 | %------------------------------------------------------------------------------------------------------------- |
|---|
| 75 | |
|---|
| 76 | The options \np{ln\_t3d} and \np{ln\_s3d} switch on the temperature and salinity |
|---|
| 77 | profile observation operator code. The \np{ln\_ena} switch turns on the reading |
|---|
| 78 | of ENACT/ENSEMBLES type profile data. The filename or array of filenames are |
|---|
| 79 | specified using the \np{enactfiles} variable. The model grid points for a |
|---|
| 80 | particular observation latitude and longitude are found using the grid |
|---|
| 81 | searching part of the code. This can be expensive, particularly for large |
|---|
| 82 | numbers of observations, setting \np{ln\_grid\_search\_lookup} allows the use of |
|---|
| 83 | a lookup table which is saved into an ``xypos`` file (or files). This will need |
|---|
| 84 | to be generated the first time if it does not exist in the run directory. |
|---|
| 85 | However, once produced it will significantly speed up future grid searches. |
|---|
| 86 | Setting \np{ln\_grid\_global} means that the code distributes the observations |
|---|
| 87 | evenly between processors. Alternatively each processor will work with |
|---|
| 88 | observations located within the model subdomain (see section~\ref{OBS_parallel}). |
|---|
| 89 | |
|---|
| 90 | A number of utilities are now provided to plot the feedback files, convert and |
|---|
| 91 | recombine the files. These are explained in more detail in section~\ref{OBS_obsutils}. |
|---|
| 92 | |
|---|
| 93 | \section{Technical details} |
|---|
| 94 | \label{OBS_details} |
|---|
| 95 | |
|---|
| 96 | Here we show a more complete example namelist and also show the NetCDF headers |
|---|
| 97 | of the observation |
|---|
| 98 | files that may be used with the observation operator |
|---|
| 99 | |
|---|
| 100 | %------------------------------------------namobs-------------------------------------------------------- |
|---|
| 101 | \namdisplay{namobs} |
|---|
| 102 | %------------------------------------------------------------------------------------------------------------- |
|---|
| 103 | |
|---|
| 104 | This name list uses the "feedback" type observation file input format for |
|---|
| 105 | profile, sea level anomaly and sea surface temperature data. All the |
|---|
| 106 | observation files must be in NetCDF format. Some example headers (produced using |
|---|
| 107 | \mbox{\textit{ncdump~-h}}) for profile |
|---|
| 108 | data, sea level anomaly and sea surface temperature are in the following |
|---|
| 109 | subsections. |
|---|
| 110 | |
|---|
| 111 | \subsection{Profile feedback type observation file header} |
|---|
| 112 | |
|---|
| 113 | \begin{alltt} |
|---|
| 114 | \tiny |
|---|
| 115 | \begin{verbatim} |
|---|
| 116 | netcdf profiles_01 { |
|---|
| 117 | dimensions: |
|---|
| 118 | N_OBS = 603 ; |
|---|
| 119 | N_LEVELS = 150 ; |
|---|
| 120 | N_VARS = 2 ; |
|---|
| 121 | N_QCF = 2 ; |
|---|
| 122 | N_ENTRIES = 1 ; |
|---|
| 123 | N_EXTRA = 1 ; |
|---|
| 124 | STRINGNAM = 8 ; |
|---|
| 125 | STRINGGRID = 1 ; |
|---|
| 126 | STRINGWMO = 8 ; |
|---|
| 127 | STRINGTYP = 4 ; |
|---|
| 128 | STRINGJULD = 14 ; |
|---|
| 129 | variables: |
|---|
| 130 | char VARIABLES(N_VARS, STRINGNAM) ; |
|---|
| 131 | VARIABLES:long_name = "List of variables in feedback files" ; |
|---|
| 132 | char ENTRIES(N_ENTRIES, STRINGNAM) ; |
|---|
| 133 | ENTRIES:long_name = "List of additional entries for each variable in feedback files" ; |
|---|
| 134 | char EXTRA(N_EXTRA, STRINGNAM) ; |
|---|
| 135 | EXTRA:long_name = "List of extra variables" ; |
|---|
| 136 | char STATION_IDENTIFIER(N_OBS, STRINGWMO) ; |
|---|
| 137 | STATION_IDENTIFIER:long_name = "Station identifier" ; |
|---|
| 138 | char STATION_TYPE(N_OBS, STRINGTYP) ; |
|---|
| 139 | STATION_TYPE:long_name = "Code instrument type" ; |
|---|
| 140 | double LONGITUDE(N_OBS) ; |
|---|
| 141 | LONGITUDE:long_name = "Longitude" ; |
|---|
| 142 | LONGITUDE:units = "degrees_east" ; |
|---|
| 143 | LONGITUDE:_Fillvalue = 99999.f ; |
|---|
| 144 | double LATITUDE(N_OBS) ; |
|---|
| 145 | LATITUDE:long_name = "Latitude" ; |
|---|
| 146 | LATITUDE:units = "degrees_north" ; |
|---|
| 147 | LATITUDE:_Fillvalue = 99999.f ; |
|---|
| 148 | double DEPTH(N_OBS, N_LEVELS) ; |
|---|
| 149 | DEPTH:long_name = "Depth" ; |
|---|
| 150 | DEPTH:units = "metre" ; |
|---|
| 151 | DEPTH:_Fillvalue = 99999.f ; |
|---|
| 152 | int DEPTH_QC(N_OBS, N_LEVELS) ; |
|---|
| 153 | DEPTH_QC:long_name = "Quality on depth" ; |
|---|
| 154 | DEPTH_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 155 | DEPTH_QC:_Fillvalue = 0 ; |
|---|
| 156 | int DEPTH_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 157 | DEPTH_QC_FLAGS:long_name = "Quality flags on depth" ; |
|---|
| 158 | DEPTH_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 159 | double JULD(N_OBS) ; |
|---|
| 160 | JULD:long_name = "Julian day" ; |
|---|
| 161 | JULD:units = "days since JULD_REFERENCE" ; |
|---|
| 162 | JULD:Conventions = "relative julian days with decimal part (as parts of day)" ; |
|---|
| 163 | JULD:_Fillvalue = 99999.f ; |
|---|
| 164 | char JULD_REFERENCE(STRINGJULD) ; |
|---|
| 165 | JULD_REFERENCE:long_name = "Date of reference for julian days" ; |
|---|
| 166 | JULD_REFERENCE:Conventions = "YYYYMMDDHHMMSS" ; |
|---|
| 167 | int OBSERVATION_QC(N_OBS) ; |
|---|
| 168 | OBSERVATION_QC:long_name = "Quality on observation" ; |
|---|
| 169 | OBSERVATION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 170 | OBSERVATION_QC:_Fillvalue = 0 ; |
|---|
| 171 | int OBSERVATION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 172 | OBSERVATION_QC_FLAGS:long_name = "Quality flags on observation" ; |
|---|
| 173 | OBSERVATION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 174 | OBSERVATION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 175 | int POSITION_QC(N_OBS) ; |
|---|
| 176 | POSITION_QC:long_name = "Quality on position (latitude and longitude)" ; |
|---|
| 177 | POSITION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 178 | POSITION_QC:_Fillvalue = 0 ; |
|---|
| 179 | int POSITION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 180 | POSITION_QC_FLAGS:long_name = "Quality flags on position" ; |
|---|
| 181 | POSITION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 182 | POSITION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 183 | int JULD_QC(N_OBS) ; |
|---|
| 184 | JULD_QC:long_name = "Quality on date and time" ; |
|---|
| 185 | JULD_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 186 | JULD_QC:_Fillvalue = 0 ; |
|---|
| 187 | int JULD_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 188 | JULD_QC_FLAGS:long_name = "Quality flags on date and time" ; |
|---|
| 189 | JULD_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 190 | JULD_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 191 | int ORIGINAL_FILE_INDEX(N_OBS) ; |
|---|
| 192 | ORIGINAL_FILE_INDEX:long_name = "Index in original data file" ; |
|---|
| 193 | ORIGINAL_FILE_INDEX:_Fillvalue = -99999 ; |
|---|
| 194 | float POTM_OBS(N_OBS, N_LEVELS) ; |
|---|
| 195 | POTM_OBS:long_name = "Potential temperature" ; |
|---|
| 196 | POTM_OBS:units = "Degrees Celsius" ; |
|---|
| 197 | POTM_OBS:_Fillvalue = 99999.f ; |
|---|
| 198 | float POTM_Hx(N_OBS, N_LEVELS) ; |
|---|
| 199 | POTM_Hx:long_name = "Model interpolated potential temperature" ; |
|---|
| 200 | POTM_Hx:units = "Degrees Celsius" ; |
|---|
| 201 | POTM_Hx:_Fillvalue = 99999.f ; |
|---|
| 202 | int POTM_QC(N_OBS) ; |
|---|
| 203 | POTM_QC:long_name = "Quality on potential temperature" ; |
|---|
| 204 | POTM_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 205 | POTM_QC:_Fillvalue = 0 ; |
|---|
| 206 | int POTM_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 207 | POTM_QC_FLAGS:long_name = "Quality flags on potential temperature" ; |
|---|
| 208 | POTM_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 209 | POTM_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 210 | int POTM_LEVEL_QC(N_OBS, N_LEVELS) ; |
|---|
| 211 | POTM_LEVEL_QC:long_name = "Quality for each level on potential temperature" ; |
|---|
| 212 | POTM_LEVEL_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 213 | POTM_LEVEL_QC:_Fillvalue = 0 ; |
|---|
| 214 | int POTM_LEVEL_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 215 | POTM_LEVEL_QC_FLAGS:long_name = "Quality flags for each level on potential temperature" ; |
|---|
| 216 | POTM_LEVEL_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 217 | POTM_LEVEL_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 218 | int POTM_IOBSI(N_OBS) ; |
|---|
| 219 | POTM_IOBSI:long_name = "ORCA grid search I coordinate" ; |
|---|
| 220 | int POTM_IOBSJ(N_OBS) ; |
|---|
| 221 | POTM_IOBSJ:long_name = "ORCA grid search J coordinate" ; |
|---|
| 222 | int POTM_IOBSK(N_OBS, N_LEVELS) ; |
|---|
| 223 | POTM_IOBSK:long_name = "ORCA grid search K coordinate" ; |
|---|
| 224 | char POTM_GRID(STRINGGRID) ; |
|---|
| 225 | POTM_GRID:long_name = "ORCA grid search grid (T,U,V)" ; |
|---|
| 226 | float PSAL_OBS(N_OBS, N_LEVELS) ; |
|---|
| 227 | PSAL_OBS:long_name = "Practical salinity" ; |
|---|
| 228 | PSAL_OBS:units = "PSU" ; |
|---|
| 229 | PSAL_OBS:_Fillvalue = 99999.f ; |
|---|
| 230 | float PSAL_Hx(N_OBS, N_LEVELS) ; |
|---|
| 231 | PSAL_Hx:long_name = "Model interpolated practical salinity" ; |
|---|
| 232 | PSAL_Hx:units = "PSU" ; |
|---|
| 233 | PSAL_Hx:_Fillvalue = 99999.f ; |
|---|
| 234 | int PSAL_QC(N_OBS) ; |
|---|
| 235 | PSAL_QC:long_name = "Quality on practical salinity" ; |
|---|
| 236 | PSAL_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 237 | PSAL_QC:_Fillvalue = 0 ; |
|---|
| 238 | int PSAL_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 239 | PSAL_QC_FLAGS:long_name = "Quality flags on practical salinity" ; |
|---|
| 240 | PSAL_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 241 | PSAL_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 242 | int PSAL_LEVEL_QC(N_OBS, N_LEVELS) ; |
|---|
| 243 | PSAL_LEVEL_QC:long_name = "Quality for each level on practical salinity" ; |
|---|
| 244 | PSAL_LEVEL_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 245 | PSAL_LEVEL_QC:_Fillvalue = 0 ; |
|---|
| 246 | int PSAL_LEVEL_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 247 | PSAL_LEVEL_QC_FLAGS:long_name = "Quality flags for each level on practical salinity" ; |
|---|
| 248 | PSAL_LEVEL_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 249 | PSAL_LEVEL_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 250 | int PSAL_IOBSI(N_OBS) ; |
|---|
| 251 | PSAL_IOBSI:long_name = "ORCA grid search I coordinate" ; |
|---|
| 252 | int PSAL_IOBSJ(N_OBS) ; |
|---|
| 253 | PSAL_IOBSJ:long_name = "ORCA grid search J coordinate" ; |
|---|
| 254 | int PSAL_IOBSK(N_OBS, N_LEVELS) ; |
|---|
| 255 | PSAL_IOBSK:long_name = "ORCA grid search K coordinate" ; |
|---|
| 256 | char PSAL_GRID(STRINGGRID) ; |
|---|
| 257 | PSAL_GRID:long_name = "ORCA grid search grid (T,U,V)" ; |
|---|
| 258 | float TEMP(N_OBS, N_LEVELS) ; |
|---|
| 259 | TEMP:long_name = "Insitu temperature" ; |
|---|
| 260 | TEMP:units = "Degrees Celsius" ; |
|---|
| 261 | TEMP:_Fillvalue = 99999.f ; |
|---|
| 262 | |
|---|
| 263 | // global attributes: |
|---|
| 264 | :title = "NEMO observation operator output" ; |
|---|
| 265 | :Convention = "NEMO unified observation operator output" ; |
|---|
| 266 | } |
|---|
| 267 | \end{verbatim} |
|---|
| 268 | \end{alltt} |
|---|
| 269 | |
|---|
| 270 | \subsection{Sea level anomaly feedback type observation file header} |
|---|
| 271 | |
|---|
| 272 | \begin{alltt} |
|---|
| 273 | \tiny |
|---|
| 274 | \begin{verbatim} |
|---|
| 275 | netcdf sla_01 { |
|---|
| 276 | dimensions: |
|---|
| 277 | N_OBS = 41301 ; |
|---|
| 278 | N_LEVELS = 1 ; |
|---|
| 279 | N_VARS = 1 ; |
|---|
| 280 | N_QCF = 2 ; |
|---|
| 281 | N_ENTRIES = 1 ; |
|---|
| 282 | N_EXTRA = 1 ; |
|---|
| 283 | STRINGNAM = 8 ; |
|---|
| 284 | STRINGGRID = 1 ; |
|---|
| 285 | STRINGWMO = 8 ; |
|---|
| 286 | STRINGTYP = 4 ; |
|---|
| 287 | STRINGJULD = 14 ; |
|---|
| 288 | variables: |
|---|
| 289 | char VARIABLES(N_VARS, STRINGNAM) ; |
|---|
| 290 | VARIABLES:long_name = "List of variables in feedback files" ; |
|---|
| 291 | char ENTRIES(N_ENTRIES, STRINGNAM) ; |
|---|
| 292 | ENTRIES:long_name = "List of additional entries for each variable in feedback files" ; |
|---|
| 293 | char EXTRA(N_EXTRA, STRINGNAM) ; |
|---|
| 294 | EXTRA:long_name = "List of extra variables" ; |
|---|
| 295 | char STATION_IDENTIFIER(N_OBS, STRINGWMO) ; |
|---|
| 296 | STATION_IDENTIFIER:long_name = "Station identifier" ; |
|---|
| 297 | char STATION_TYPE(N_OBS, STRINGTYP) ; |
|---|
| 298 | STATION_TYPE:long_name = "Code instrument type" ; |
|---|
| 299 | double LONGITUDE(N_OBS) ; |
|---|
| 300 | LONGITUDE:long_name = "Longitude" ; |
|---|
| 301 | LONGITUDE:units = "degrees_east" ; |
|---|
| 302 | LONGITUDE:_Fillvalue = 99999.f ; |
|---|
| 303 | double LATITUDE(N_OBS) ; |
|---|
| 304 | LATITUDE:long_name = "Latitude" ; |
|---|
| 305 | LATITUDE:units = "degrees_north" ; |
|---|
| 306 | LATITUDE:_Fillvalue = 99999.f ; |
|---|
| 307 | double DEPTH(N_OBS, N_LEVELS) ; |
|---|
| 308 | DEPTH:long_name = "Depth" ; |
|---|
| 309 | DEPTH:units = "metre" ; |
|---|
| 310 | DEPTH:_Fillvalue = 99999.f ; |
|---|
| 311 | int DEPTH_QC(N_OBS, N_LEVELS) ; |
|---|
| 312 | DEPTH_QC:long_name = "Quality on depth" ; |
|---|
| 313 | DEPTH_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 314 | DEPTH_QC:_Fillvalue = 0 ; |
|---|
| 315 | int DEPTH_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 316 | DEPTH_QC_FLAGS:long_name = "Quality flags on depth" ; |
|---|
| 317 | DEPTH_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 318 | double JULD(N_OBS) ; |
|---|
| 319 | JULD:long_name = "Julian day" ; |
|---|
| 320 | JULD:units = "days since JULD_REFERENCE" ; |
|---|
| 321 | JULD:Conventions = "relative julian days with decimal part (as parts of day)" ; |
|---|
| 322 | JULD:_Fillvalue = 99999.f ; |
|---|
| 323 | char JULD_REFERENCE(STRINGJULD) ; |
|---|
| 324 | JULD_REFERENCE:long_name = "Date of reference for julian days" ; |
|---|
| 325 | JULD_REFERENCE:Conventions = "YYYYMMDDHHMMSS" ; |
|---|
| 326 | int OBSERVATION_QC(N_OBS) ; |
|---|
| 327 | OBSERVATION_QC:long_name = "Quality on observation" ; |
|---|
| 328 | OBSERVATION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 329 | OBSERVATION_QC:_Fillvalue = 0 ; |
|---|
| 330 | int OBSERVATION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 331 | OBSERVATION_QC_FLAGS:long_name = "Quality flags on observation" ; |
|---|
| 332 | OBSERVATION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 333 | OBSERVATION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 334 | int POSITION_QC(N_OBS) ; |
|---|
| 335 | POSITION_QC:long_name = "Quality on position (latitude and longitude)" ; |
|---|
| 336 | POSITION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 337 | POSITION_QC:_Fillvalue = 0 ; |
|---|
| 338 | int POSITION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 339 | POSITION_QC_FLAGS:long_name = "Quality flags on position" ; |
|---|
| 340 | POSITION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 341 | POSITION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 342 | int JULD_QC(N_OBS) ; |
|---|
| 343 | JULD_QC:long_name = "Quality on date and time" ; |
|---|
| 344 | JULD_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 345 | JULD_QC:_Fillvalue = 0 ; |
|---|
| 346 | int JULD_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 347 | JULD_QC_FLAGS:long_name = "Quality flags on date and time" ; |
|---|
| 348 | JULD_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 349 | JULD_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 350 | int ORIGINAL_FILE_INDEX(N_OBS) ; |
|---|
| 351 | ORIGINAL_FILE_INDEX:long_name = "Index in original data file" ; |
|---|
| 352 | ORIGINAL_FILE_INDEX:_Fillvalue = -99999 ; |
|---|
| 353 | float SLA_OBS(N_OBS, N_LEVELS) ; |
|---|
| 354 | SLA_OBS:long_name = "Sea level anomaly" ; |
|---|
| 355 | SLA_OBS:units = "metre" ; |
|---|
| 356 | SLA_OBS:_Fillvalue = 99999.f ; |
|---|
| 357 | float SLA_Hx(N_OBS, N_LEVELS) ; |
|---|
| 358 | SLA_Hx:long_name = "Model interpolated sea level anomaly" ; |
|---|
| 359 | SLA_Hx:units = "metre" ; |
|---|
| 360 | SLA_Hx:_Fillvalue = 99999.f ; |
|---|
| 361 | int SLA_QC(N_OBS) ; |
|---|
| 362 | SLA_QC:long_name = "Quality on sea level anomaly" ; |
|---|
| 363 | SLA_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 364 | SLA_QC:_Fillvalue = 0 ; |
|---|
| 365 | int SLA_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 366 | SLA_QC_FLAGS:long_name = "Quality flags on sea level anomaly" ; |
|---|
| 367 | SLA_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 368 | SLA_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 369 | int SLA_LEVEL_QC(N_OBS, N_LEVELS) ; |
|---|
| 370 | SLA_LEVEL_QC:long_name = "Quality for each level on sea level anomaly" ; |
|---|
| 371 | SLA_LEVEL_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 372 | SLA_LEVEL_QC:_Fillvalue = 0 ; |
|---|
| 373 | int SLA_LEVEL_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 374 | SLA_LEVEL_QC_FLAGS:long_name = "Quality flags for each level on sea level anomaly" ; |
|---|
| 375 | SLA_LEVEL_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 376 | SLA_LEVEL_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 377 | int SLA_IOBSI(N_OBS) ; |
|---|
| 378 | SLA_IOBSI:long_name = "ORCA grid search I coordinate" ; |
|---|
| 379 | int SLA_IOBSJ(N_OBS) ; |
|---|
| 380 | SLA_IOBSJ:long_name = "ORCA grid search J coordinate" ; |
|---|
| 381 | int SLA_IOBSK(N_OBS, N_LEVELS) ; |
|---|
| 382 | SLA_IOBSK:long_name = "ORCA grid search K coordinate" ; |
|---|
| 383 | char SLA_GRID(STRINGGRID) ; |
|---|
| 384 | SLA_GRID:long_name = "ORCA grid search grid (T,U,V)" ; |
|---|
| 385 | float MDT(N_OBS, N_LEVELS) ; |
|---|
| 386 | MDT:long_name = "Mean Dynamic Topography" ; |
|---|
| 387 | MDT:units = "metre" ; |
|---|
| 388 | MDT:_Fillvalue = 99999.f ; |
|---|
| 389 | |
|---|
| 390 | // global attributes: |
|---|
| 391 | :title = "NEMO observation operator output" ; |
|---|
| 392 | :Convention = "NEMO unified observation operator output" ; |
|---|
| 393 | } |
|---|
| 394 | \end{verbatim} |
|---|
| 395 | \end{alltt} |
|---|
| 396 | |
|---|
| 397 | The mean dynamic |
|---|
| 398 | topography (MDT) must be provided in a separate file defined on the model grid |
|---|
| 399 | called {\it slaReferenceLevel.nc}. The MDT is required in |
|---|
| 400 | order to produce the model equivalent sea level anomaly from the model sea |
|---|
| 401 | surface height. Below is an example header for this file (on the ORCA025 grid). |
|---|
| 402 | |
|---|
| 403 | \begin{alltt} |
|---|
| 404 | \tiny |
|---|
| 405 | \begin{verbatim} |
|---|
| 406 | dimensions: |
|---|
| 407 | x = 1442 ; |
|---|
| 408 | y = 1021 ; |
|---|
| 409 | variables: |
|---|
| 410 | float nav_lon(y, x) ; |
|---|
| 411 | nav_lon:units = "degrees_east" ; |
|---|
| 412 | float nav_lat(y, x) ; |
|---|
| 413 | nav_lat:units = "degrees_north" ; |
|---|
| 414 | float sossheig(y, x) ; |
|---|
| 415 | sossheig:_FillValue = -1.e+30f ; |
|---|
| 416 | sossheig:coordinates = "nav_lon nav_lat" ; |
|---|
| 417 | sossheig:long_name = "Mean Dynamic Topography" ; |
|---|
| 418 | sossheig:units = "metres" ; |
|---|
| 419 | sossheig:grid = "orca025T" ; |
|---|
| 420 | \end{verbatim} |
|---|
| 421 | \end{alltt} |
|---|
| 422 | |
|---|
| 423 | \subsection{Sea surface temperature feedback type observation file header} |
|---|
| 424 | |
|---|
| 425 | \begin{alltt} |
|---|
| 426 | \tiny |
|---|
| 427 | \begin{verbatim} |
|---|
| 428 | netcdf sst_01 { |
|---|
| 429 | dimensions: |
|---|
| 430 | N_OBS = 33099 ; |
|---|
| 431 | N_LEVELS = 1 ; |
|---|
| 432 | N_VARS = 1 ; |
|---|
| 433 | N_QCF = 2 ; |
|---|
| 434 | N_ENTRIES = 1 ; |
|---|
| 435 | STRINGNAM = 8 ; |
|---|
| 436 | STRINGGRID = 1 ; |
|---|
| 437 | STRINGWMO = 8 ; |
|---|
| 438 | STRINGTYP = 4 ; |
|---|
| 439 | STRINGJULD = 14 ; |
|---|
| 440 | variables: |
|---|
| 441 | char VARIABLES(N_VARS, STRINGNAM) ; |
|---|
| 442 | VARIABLES:long_name = "List of variables in feedback files" ; |
|---|
| 443 | char ENTRIES(N_ENTRIES, STRINGNAM) ; |
|---|
| 444 | ENTRIES:long_name = "List of additional entries for each variable in feedback files" ; |
|---|
| 445 | char STATION_IDENTIFIER(N_OBS, STRINGWMO) ; |
|---|
| 446 | STATION_IDENTIFIER:long_name = "Station identifier" ; |
|---|
| 447 | char STATION_TYPE(N_OBS, STRINGTYP) ; |
|---|
| 448 | STATION_TYPE:long_name = "Code instrument type" ; |
|---|
| 449 | double LONGITUDE(N_OBS) ; |
|---|
| 450 | LONGITUDE:long_name = "Longitude" ; |
|---|
| 451 | LONGITUDE:units = "degrees_east" ; |
|---|
| 452 | LONGITUDE:_Fillvalue = 99999.f ; |
|---|
| 453 | double LATITUDE(N_OBS) ; |
|---|
| 454 | LATITUDE:long_name = "Latitude" ; |
|---|
| 455 | LATITUDE:units = "degrees_north" ; |
|---|
| 456 | LATITUDE:_Fillvalue = 99999.f ; |
|---|
| 457 | double DEPTH(N_OBS, N_LEVELS) ; |
|---|
| 458 | DEPTH:long_name = "Depth" ; |
|---|
| 459 | DEPTH:units = "metre" ; |
|---|
| 460 | DEPTH:_Fillvalue = 99999.f ; |
|---|
| 461 | int DEPTH_QC(N_OBS, N_LEVELS) ; |
|---|
| 462 | DEPTH_QC:long_name = "Quality on depth" ; |
|---|
| 463 | DEPTH_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 464 | DEPTH_QC:_Fillvalue = 0 ; |
|---|
| 465 | int DEPTH_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 466 | DEPTH_QC_FLAGS:long_name = "Quality flags on depth" ; |
|---|
| 467 | DEPTH_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 468 | double JULD(N_OBS) ; |
|---|
| 469 | JULD:long_name = "Julian day" ; |
|---|
| 470 | JULD:units = "days since JULD_REFERENCE" ; |
|---|
| 471 | JULD:Conventions = "relative julian days with decimal part (as parts of day)" ; |
|---|
| 472 | JULD:_Fillvalue = 99999.f ; |
|---|
| 473 | char JULD_REFERENCE(STRINGJULD) ; |
|---|
| 474 | JULD_REFERENCE:long_name = "Date of reference for julian days" ; |
|---|
| 475 | JULD_REFERENCE:Conventions = "YYYYMMDDHHMMSS" ; |
|---|
| 476 | int OBSERVATION_QC(N_OBS) ; |
|---|
| 477 | OBSERVATION_QC:long_name = "Quality on observation" ; |
|---|
| 478 | OBSERVATION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 479 | OBSERVATION_QC:_Fillvalue = 0 ; |
|---|
| 480 | int OBSERVATION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 481 | OBSERVATION_QC_FLAGS:long_name = "Quality flags on observation" ; |
|---|
| 482 | OBSERVATION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 483 | OBSERVATION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 484 | int POSITION_QC(N_OBS) ; |
|---|
| 485 | POSITION_QC:long_name = "Quality on position (latitude and longitude)" ; |
|---|
| 486 | POSITION_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 487 | POSITION_QC:_Fillvalue = 0 ; |
|---|
| 488 | int POSITION_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 489 | POSITION_QC_FLAGS:long_name = "Quality flags on position" ; |
|---|
| 490 | POSITION_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 491 | POSITION_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 492 | int JULD_QC(N_OBS) ; |
|---|
| 493 | JULD_QC:long_name = "Quality on date and time" ; |
|---|
| 494 | JULD_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 495 | JULD_QC:_Fillvalue = 0 ; |
|---|
| 496 | int JULD_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 497 | JULD_QC_FLAGS:long_name = "Quality flags on date and time" ; |
|---|
| 498 | JULD_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 499 | JULD_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 500 | int ORIGINAL_FILE_INDEX(N_OBS) ; |
|---|
| 501 | ORIGINAL_FILE_INDEX:long_name = "Index in original data file" ; |
|---|
| 502 | ORIGINAL_FILE_INDEX:_Fillvalue = -99999 ; |
|---|
| 503 | float SST_OBS(N_OBS, N_LEVELS) ; |
|---|
| 504 | SST_OBS:long_name = "Sea surface temperature" ; |
|---|
| 505 | SST_OBS:units = "Degree centigrade" ; |
|---|
| 506 | SST_OBS:_Fillvalue = 99999.f ; |
|---|
| 507 | float SST_Hx(N_OBS, N_LEVELS) ; |
|---|
| 508 | SST_Hx:long_name = "Model interpolated sea surface temperature" ; |
|---|
| 509 | SST_Hx:units = "Degree centigrade" ; |
|---|
| 510 | SST_Hx:_Fillvalue = 99999.f ; |
|---|
| 511 | int SST_QC(N_OBS) ; |
|---|
| 512 | SST_QC:long_name = "Quality on sea surface temperature" ; |
|---|
| 513 | SST_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 514 | SST_QC:_Fillvalue = 0 ; |
|---|
| 515 | int SST_QC_FLAGS(N_OBS, N_QCF) ; |
|---|
| 516 | SST_QC_FLAGS:long_name = "Quality flags on sea surface temperature" ; |
|---|
| 517 | SST_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 518 | SST_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 519 | int SST_LEVEL_QC(N_OBS, N_LEVELS) ; |
|---|
| 520 | SST_LEVEL_QC:long_name = "Quality for each level on sea surface temperature" ; |
|---|
| 521 | SST_LEVEL_QC:Conventions = "q where q =[0,9]" ; |
|---|
| 522 | SST_LEVEL_QC:_Fillvalue = 0 ; |
|---|
| 523 | int SST_LEVEL_QC_FLAGS(N_OBS, N_LEVELS, N_QCF) ; |
|---|
| 524 | SST_LEVEL_QC_FLAGS:long_name = "Quality flags for each level on sea surface temperature" ; |
|---|
| 525 | SST_LEVEL_QC_FLAGS:Conventions = "NEMOVAR flag conventions" ; |
|---|
| 526 | SST_LEVEL_QC_FLAGS:_Fillvalue = 0 ; |
|---|
| 527 | int SST_IOBSI(N_OBS) ; |
|---|
| 528 | SST_IOBSI:long_name = "ORCA grid search I coordinate" ; |
|---|
| 529 | int SST_IOBSJ(N_OBS) ; |
|---|
| 530 | SST_IOBSJ:long_name = "ORCA grid search J coordinate" ; |
|---|
| 531 | int SST_IOBSK(N_OBS, N_LEVELS) ; |
|---|
| 532 | SST_IOBSK:long_name = "ORCA grid search K coordinate" ; |
|---|
| 533 | char SST_GRID(STRINGGRID) ; |
|---|
| 534 | SST_GRID:long_name = "ORCA grid search grid (T,U,V)" ; |
|---|
| 535 | |
|---|
| 536 | // global attributes: |
|---|
| 537 | :title = "NEMO observation operator output" ; |
|---|
| 538 | :Convention = "NEMO unified observation operator output" ; |
|---|
| 539 | } |
|---|
| 540 | \end{verbatim} |
|---|
| 541 | \end{alltt} |
|---|
| 542 | |
|---|
| 543 | \section{Theoretical details} |
|---|
| 544 | \label{OBS_theory} |
|---|
| 545 | |
|---|
| 546 | \subsection{Horizontal interpolation methods} |
|---|
| 547 | |
|---|
| 548 | Consider an observation point ${\rm P}$ with |
|---|
| 549 | with longitude and latitude $({\lambda_{}}_{\rm P}, \phi_{\rm P})$ and the |
|---|
| 550 | four nearest neighbouring model grid points ${\rm A}$, ${\rm B}$, ${\rm C}$ |
|---|
| 551 | and ${\rm D}$ with longitude and latitude ($\lambda_{\rm A}$, $\phi_{\rm A}$), |
|---|
| 552 | ($\lambda_{\rm B}$, $\phi_{\rm B}$) etc. |
|---|
| 553 | All horizontal interpolation methods implemented in NEMO |
|---|
| 554 | estimate the value of a model variable $x$ at point $P$ as |
|---|
| 555 | a weighted linear combination of the values of the model |
|---|
| 556 | variables at the grid points ${\rm A}$, ${\rm B}$ etc.: |
|---|
| 557 | \begin{eqnarray} |
|---|
| 558 | {x_{}}_{\rm P} & \hspace{-2mm} = \hspace{-2mm} & |
|---|
| 559 | \frac{1}{w} \left( {w_{}}_{\rm A} {x_{}}_{\rm A} + |
|---|
| 560 | {w_{}}_{\rm B} {x_{}}_{\rm B} + |
|---|
| 561 | {w_{}}_{\rm C} {x_{}}_{\rm C} + |
|---|
| 562 | {w_{}}_{\rm D} {x_{}}_{\rm D} \right) |
|---|
| 563 | \end{eqnarray} |
|---|
| 564 | where ${w_{}}_{\rm A}$, ${w_{}}_{\rm B}$ etc. are the respective weights for the |
|---|
| 565 | model field at points ${\rm A}$, ${\rm B}$ etc., and |
|---|
| 566 | $w = {w_{}}_{\rm A} + {w_{}}_{\rm B} + {w_{}}_{\rm C} + {w_{}}_{\rm D}$. |
|---|
| 567 | |
|---|
| 568 | Four different possibilities are available for computing the weights. |
|---|
| 569 | |
|---|
| 570 | \begin{enumerate} |
|---|
| 571 | |
|---|
| 572 | \item[1.] {\bf Great-Circle distance-weighted interpolation.} The weights |
|---|
| 573 | are computed as a function of the great-circle distance $s(P, \cdot)$ |
|---|
| 574 | between $P$ and the model grid points $A$, $B$ etc. For example, |
|---|
| 575 | the weight given to the field ${x_{}}_{\rm A}$ is specified as the |
|---|
| 576 | product of the distances from ${\rm P}$ to the other points: |
|---|
| 577 | \begin{eqnarray} |
|---|
| 578 | {w_{}}_{\rm A} = s({\rm P}, {\rm B}) \, s({\rm P}, {\rm C}) \, s({\rm P}, {\rm D}) |
|---|
| 579 | \nonumber |
|---|
| 580 | \end{eqnarray} |
|---|
| 581 | where |
|---|
| 582 | \begin{eqnarray} |
|---|
| 583 | s\left ({\rm P}, {\rm M} \right ) |
|---|
| 584 | & \hspace{-2mm} = \hspace{-2mm} & |
|---|
| 585 | \cos^{-1} \! \left\{ |
|---|
| 586 | \sin {\phi_{}}_{\rm P} \sin {\phi_{}}_{\rm M} |
|---|
| 587 | + \cos {\phi_{}}_{\rm P} \cos {\phi_{}}_{\rm M} |
|---|
| 588 | \cos ({\lambda_{}}_{\rm M} - {\lambda_{}}_{\rm P}) |
|---|
| 589 | \right\} |
|---|
| 590 | \end{eqnarray} |
|---|
| 591 | and $M$ corresponds to $B$, $C$ or $D$. |
|---|
| 592 | A more stable form of the great-circle distance formula for |
|---|
| 593 | small distances ($x$ near 1) involves the arcsine function |
|---|
| 594 | ($e.g.$ see p.~101 of \citet{Daley_Barker_Bk01}: |
|---|
| 595 | \begin{eqnarray} |
|---|
| 596 | s\left( {\rm P}, {\rm M} \right) |
|---|
| 597 | & \hspace{-2mm} = \hspace{-2mm} & |
|---|
| 598 | \sin^{-1} \! \left\{ \sqrt{ 1 - x^2 } \right\} |
|---|
| 599 | \nonumber |
|---|
| 600 | \end{eqnarray} |
|---|
| 601 | where |
|---|
| 602 | \begin{eqnarray} |
|---|
| 603 | x & \hspace{-2mm} = \hspace{-2mm} & |
|---|
| 604 | {a_{}}_{\rm M} {a_{}}_{\rm P} + {b_{}}_{\rm M} {b_{}}_{\rm P} + {c_{}}_{\rm M} {c_{}}_{\rm P} |
|---|
| 605 | \nonumber |
|---|
| 606 | \end{eqnarray} |
|---|
| 607 | and |
|---|
| 608 | \begin{eqnarray} |
|---|
| 609 | {a_{}}_{\rm M} & \hspace{-2mm} = \hspace{-2mm} & \sin {\phi_{}}_{\rm M}, |
|---|
| 610 | \nonumber \\ |
|---|
| 611 | {a_{}}_{\rm P} & \hspace{-2mm} = \hspace{-2mm} & \sin {\phi_{}}_{\rm P}, |
|---|
| 612 | \nonumber \\ |
|---|
| 613 | {b_{}}_{\rm M} & \hspace{-2mm} = \hspace{-2mm} & \cos {\phi_{}}_{\rm M} \cos {\phi_{}}_{\rm M}, |
|---|
| 614 | \nonumber \\ |
|---|
| 615 | {b_{}}_{\rm P} & \hspace{-2mm} = \hspace{-2mm} & \cos {\phi_{}}_{\rm P} \cos {\phi_{}}_{\rm P}, |
|---|
| 616 | \nonumber \\ |
|---|
| 617 | {c_{}}_{\rm M} & \hspace{-2mm} = \hspace{-2mm} & \cos {\phi_{}}_{\rm M} \sin {\phi_{}}_{\rm M}, |
|---|
| 618 | \nonumber \\ |
|---|
| 619 | {c_{}}_{\rm P} & \hspace{-2mm} = \hspace{-2mm} & \cos {\phi_{}}_{\rm P} \sin {\phi_{}}_{\rm P}. |
|---|
| 620 | \nonumber |
|---|
| 621 | \nonumber |
|---|
| 622 | \end{eqnarray} |
|---|
| 623 | |
|---|
| 624 | \item[2.] {\bf Great-Circle distance-weighted interpolation with small angle |
|---|
| 625 | approximation.} Similar to the previous interpolation but with the |
|---|
| 626 | distance $s$ computed as |
|---|
| 627 | \begin{eqnarray} |
|---|
| 628 | s\left( {\rm P}, {\rm M} \right) |
|---|
| 629 | & \hspace{-2mm} = \hspace{-2mm} & |
|---|
| 630 | \sqrt{ \left( {\phi_{}}_{\rm M} - {\phi_{}}_{\rm P} \right)^{2} |
|---|
| 631 | + \left( {\lambda_{}}_{\rm M} - {\lambda_{}}_{\rm P} \right)^{2} |
|---|
| 632 | \cos^{2} {\phi_{}}_{\rm M} } |
|---|
| 633 | \end{eqnarray} |
|---|
| 634 | where $M$ corresponds to $A$, $B$, $C$ or $D$. |
|---|
| 635 | |
|---|
| 636 | \item[3.] {\bf Bilinear interpolation for a regular spaced grid.} The |
|---|
| 637 | interpolation is split into two 1D interpolations in the longitude |
|---|
| 638 | and latitude directions, respectively. |
|---|
| 639 | |
|---|
| 640 | \item[4.] {\bf Bilinear remapping interpolation for a general grid.} An |
|---|
| 641 | iterative scheme that involves first mapping a quadrilateral cell |
|---|
| 642 | into a cell with coordinates (0,0), (1,0), (0,1) and (1,1). This |
|---|
| 643 | method is based on the SCRIP interpolation package \citep{Jones_1998}. |
|---|
| 644 | |
|---|
| 645 | \end{enumerate} |
|---|
| 646 | |
|---|
| 647 | \subsection{Grid search} |
|---|
| 648 | |
|---|
| 649 | For many grids used by the NEMO model, such as the ORCA family, |
|---|
| 650 | the horizontal grid coordinates $i$ and $j$ are not simple functions |
|---|
| 651 | of latitude and longitude. Therefore, it is not always straightforward |
|---|
| 652 | to determine the grid points surrounding any given observational position. |
|---|
| 653 | Before the interpolation can be performed, a search |
|---|
| 654 | algorithm is then required to determine the corner points of |
|---|
| 655 | the quadrilateral cell in which the observation is located. |
|---|
| 656 | This is the most difficult and time consuming part of the |
|---|
| 657 | 2D interpolation procedure. |
|---|
| 658 | A robust test for determining if an observation falls |
|---|
| 659 | within a given quadrilateral cell is as follows. Let |
|---|
| 660 | ${\rm P}({\lambda_{}}_{\rm P} ,{\phi_{}}_{\rm P} )$ denote the observation point, |
|---|
| 661 | and let ${\rm A}({\lambda_{}}_{\rm A} ,{\phi_{}}_{\rm A} )$, |
|---|
| 662 | ${\rm B}({\lambda_{}}_{\rm B} ,{\phi_{}}_{\rm B} )$, |
|---|
| 663 | ${\rm C}({\lambda_{}}_{\rm C} ,{\phi_{}}_{\rm C} )$ |
|---|
| 664 | and |
|---|
| 665 | ${\rm D}({\lambda_{}}_{\rm D} ,{\phi_{}}_{\rm D} )$ denote |
|---|
| 666 | the bottom left, bottom right, top left and top right |
|---|
| 667 | corner points of the cell, respectively. |
|---|
| 668 | To determine if P is inside |
|---|
| 669 | the cell, we verify that the cross-products |
|---|
| 670 | \begin{eqnarray} |
|---|
| 671 | \begin{array}{lllll} |
|---|
| 672 | {{\bf r}_{}}_{\rm PA} \times {{\bf r}_{}}_{\rm PC} |
|---|
| 673 | & = & [({\lambda_{}}_{\rm A}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 674 | ({\phi_{}}_{\rm C} \; -\; {\phi_{}}_{\rm P} ) |
|---|
| 675 | - ({\lambda_{}}_{\rm C}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 676 | ({\phi_{}}_{\rm A} \; -\; {\phi_{}}_{\rm P} )] \; \widehat{\bf k} \\ |
|---|
| 677 | {{\bf r}_{}}_{\rm PB} \times {{\bf r}_{}}_{\rm PA} |
|---|
| 678 | & = & [({\lambda_{}}_{\rm B}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 679 | ({\phi_{}}_{\rm A} \; -\; {\phi_{}}_{\rm P} ) |
|---|
| 680 | - ({\lambda_{}}_{\rm A}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 681 | ({\phi_{}}_{\rm B} \; -\; {\phi_{}}_{\rm P} )] \; \widehat{\bf k} \\ |
|---|
| 682 | {{\bf r}_{}}_{\rm PC} \times {{\bf r}_{}}_{\rm PD} |
|---|
| 683 | & = & [({\lambda_{}}_{\rm C}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 684 | ({\phi_{}}_{\rm D} \; -\; {\phi_{}}_{\rm P} ) |
|---|
| 685 | - ({\lambda_{}}_{\rm D}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 686 | ({\phi_{}}_{\rm C} \; -\; {\phi_{}}_{\rm P} )] \; \widehat{\bf k} \\ |
|---|
| 687 | {{\bf r}_{}}_{\rm PD} \times {{\bf r}_{}}_{\rm PB} |
|---|
| 688 | & = & [({\lambda_{}}_{\rm D}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 689 | ({\phi_{}}_{\rm B} \; -\; {\phi_{}}_{\rm P} ) |
|---|
| 690 | - ({\lambda_{}}_{\rm B}\; -\; {\lambda_{}}_{\rm P} ) |
|---|
| 691 | ({\phi_{}}_{\rm D} \; - \; {\phi_{}}_{\rm P} )] \; \widehat{\bf k} \\ |
|---|
| 692 | \end{array} |
|---|
| 693 | \label{eq:cross} |
|---|
| 694 | \end{eqnarray} |
|---|
| 695 | point in the opposite direction to the unit normal |
|---|
| 696 | $\widehat{\bf k}$ (i.e., that the coefficients of |
|---|
| 697 | $\widehat{\bf k}$ are negative), |
|---|
| 698 | where ${{\bf r}_{}}_{\rm PA}$, ${{\bf r}_{}}_{\rm PB}$, |
|---|
| 699 | etc. correspond to the vectors between points P and A, |
|---|
| 700 | P and B, etc.. The method used is |
|---|
| 701 | similar to the method used in |
|---|
| 702 | the SCRIP interpolation package \citep{Jones_1998}. |
|---|
| 703 | |
|---|
| 704 | In order to speed up the grid search, there is the possibility to construct |
|---|
| 705 | a lookup table for a user specified resolution. This lookup |
|---|
| 706 | table contains the lower and upper bounds on the $i$ and $j$ indices |
|---|
| 707 | to be searched for on a regular grid. For each observation position, |
|---|
| 708 | the closest point on the regular grid of this position is computed and |
|---|
| 709 | the $i$ and $j$ ranges of this point searched to determine the precise |
|---|
| 710 | four points surrounding the observation. |
|---|
| 711 | |
|---|
| 712 | \subsection{Parallel aspects of horizontal interpolation} |
|---|
| 713 | \label{OBS_parallel} |
|---|
| 714 | |
|---|
| 715 | For horizontal interpolation, there is the basic problem that the |
|---|
| 716 | observations are unevenly distributed on the globe. In numerical |
|---|
| 717 | models, it is common to divide the model grid into subgrids (or |
|---|
| 718 | domains) where each subgrid is executed on a single processing element |
|---|
| 719 | with explicit message passing for exchange of information along the |
|---|
| 720 | domain boundaries when running on a massively parallel processor (MPP) |
|---|
| 721 | system. This approach is used by \NEMO. |
|---|
| 722 | |
|---|
| 723 | For observations there is no natural distribution since the |
|---|
| 724 | observations are not equally distributed on the globe. |
|---|
| 725 | Two options have been made available: 1) geographical distribution; |
|---|
| 726 | and 2) round-robin. |
|---|
| 727 | |
|---|
| 728 | \subsubsection{Geographical distribution of observations among processors} |
|---|
| 729 | |
|---|
| 730 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 731 | \begin{figure} \begin{center} |
|---|
| 732 | \includegraphics[width=10cm,height=12cm,angle=-90.]{./TexFiles/Figures/Fig_ASM_obsdist_local} |
|---|
| 733 | \caption{ \label{fig:obslocal} |
|---|
| 734 | Example of the distribution of observations with the geographical distribution of observational data.} |
|---|
| 735 | \end{center} \end{figure} |
|---|
| 736 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 737 | |
|---|
| 738 | This is the simplest option in which the observations are distributed according |
|---|
| 739 | to the domain of the grid-point parallelization. Figure~\ref{fig:obslocal} |
|---|
| 740 | shows an example of the distribution of the {\em in situ} data on processors |
|---|
| 741 | with a different colour for each observation |
|---|
| 742 | on a given processor for a 4 $\times$ 2 decomposition with ORCA2. |
|---|
| 743 | The grid-point domain decomposition is clearly visible on the plot. |
|---|
| 744 | |
|---|
| 745 | The advantage of this approach is that all |
|---|
| 746 | information needed for horizontal interpolation is available without |
|---|
| 747 | any MPP communication. Of course, this is under the assumption that |
|---|
| 748 | we are only using a $2 \times 2$ grid-point stencil for the interpolation |
|---|
| 749 | (e.g., bilinear interpolation). For higher order interpolation schemes this |
|---|
| 750 | is no longer valid. A disadvantage with the above scheme is that the number of |
|---|
| 751 | observations on each processor can be very different. If the cost of |
|---|
| 752 | the actual interpolation is expensive relative to the communication of |
|---|
| 753 | data needed for interpolation, this could lead to load imbalance. |
|---|
| 754 | |
|---|
| 755 | \subsubsection{Round-robin distribution of observations among processors} |
|---|
| 756 | |
|---|
| 757 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 758 | \begin{figure} \begin{center} |
|---|
| 759 | \includegraphics[width=10cm,height=12cm,angle=-90.]{./TexFiles/Figures/Fig_ASM_obsdist_global} |
|---|
| 760 | \caption{ \label{fig:obsglobal} |
|---|
| 761 | Example of the distribution of observations with the round-robin distribution of observational data.} |
|---|
| 762 | \end{center} \end{figure} |
|---|
| 763 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 764 | |
|---|
| 765 | An alternative approach is to distribute the observations equally |
|---|
| 766 | among processors and use message passing in order to retrieve |
|---|
| 767 | the stencil for interpolation. The simplest distribution of the observations |
|---|
| 768 | is to distribute them using a round-robin scheme. Figure~\ref{fig:obsglobal} |
|---|
| 769 | shows the distribution of the {\em in situ} data on processors for the |
|---|
| 770 | round-robin distribution of observations with a different colour for |
|---|
| 771 | each observation on a given processor for a 4 $\times$ 2 decomposition |
|---|
| 772 | with ORCA2 for the same input data as in Fig.~\ref{fig:obslocal}. |
|---|
| 773 | The observations are now clearly randomly distributed on the globe. |
|---|
| 774 | In order to be able to perform horizontal interpolation in this case, |
|---|
| 775 | a subroutine has been developed that retrieves any grid points in the |
|---|
| 776 | global space. |
|---|
| 777 | |
|---|
| 778 | \subsection{Vertical interpolation operator} |
|---|
| 779 | |
|---|
| 780 | Vertical interpolation is achieved using either a cubic spline or |
|---|
| 781 | linear interpolation. For the cubic spline, the top and |
|---|
| 782 | bottom boundary conditions for the second derivative of the |
|---|
| 783 | interpolating polynomial in the spline are set to zero. |
|---|
| 784 | At the bottom boundary, this is done using the land-ocean mask. |
|---|
| 785 | |
|---|
| 786 | \newpage |
|---|
| 787 | |
|---|
| 788 | \section{Observation Utilities} |
|---|
| 789 | \label{OBS_obsutils} |
|---|
| 790 | |
|---|
| 791 | Some tools for viewing and processing of observation and feedback files are provided in the |
|---|
| 792 | NEMO repository for convenience. These include OBSTOOLS which are a collection of Fortran |
|---|
| 793 | programs which are helpful to deal with feedback files. They do such tasks as observation file |
|---|
| 794 | conversion, printing of file contents, some basic statistical analysis of feedback files. The |
|---|
| 795 | other tool is an IDL program called dataplot which uses a graphical interface to visualise |
|---|
| 796 | observations and feedback files. OBSTOOLS and dataplot are described in more detail below. |
|---|
| 797 | |
|---|
| 798 | \subsection{Obstools} |
|---|
| 799 | |
|---|
| 800 | A series of Fortran utilities is provided with NEMO called OBSTOOLS. This are helpful in |
|---|
| 801 | handling observation files and the feedback file output from the NEMO observation operator. |
|---|
| 802 | The utilities are as follows |
|---|
| 803 | |
|---|
| 804 | \subsubsection{corio2fb} |
|---|
| 805 | |
|---|
| 806 | The program corio2fb converts profile observation files from the Coriolis format to the |
|---|
| 807 | standard feedback format. The program is called in the following way: |
|---|
| 808 | |
|---|
| 809 | \begin{alltt} |
|---|
| 810 | \footnotesize |
|---|
| 811 | \begin{verbatim} |
|---|
| 812 | corio2fb.exe outputfile inputfile1 inputfile2 ... |
|---|
| 813 | \end{verbatim} |
|---|
| 814 | \end{alltt} |
|---|
| 815 | |
|---|
| 816 | \subsubsection{enact2fb} |
|---|
| 817 | |
|---|
| 818 | The program enact2fb converts profile observation files from the ENACT format to the standard |
|---|
| 819 | feedback format. The program is called in the following way: |
|---|
| 820 | |
|---|
| 821 | \begin{alltt} |
|---|
| 822 | \footnotesize |
|---|
| 823 | \begin{verbatim} |
|---|
| 824 | enact2fb.exe outputfile inputfile1 inputfile2 ... |
|---|
| 825 | \end{verbatim} |
|---|
| 826 | \end{alltt} |
|---|
| 827 | |
|---|
| 828 | \subsubsection{fbcomb} |
|---|
| 829 | |
|---|
| 830 | The program fbcomb combines multiple feedback files produced by individual processors in an |
|---|
| 831 | MPI run of NEMO into a single feedback file. The program is called in the following way: |
|---|
| 832 | |
|---|
| 833 | \begin{alltt} |
|---|
| 834 | \footnotesize |
|---|
| 835 | \begin{verbatim} |
|---|
| 836 | fbcomb.exe outputfile inputfile1 inputfile2 ... |
|---|
| 837 | \end{verbatim} |
|---|
| 838 | \end{alltt} |
|---|
| 839 | |
|---|
| 840 | \subsubsection{fbmatchup} |
|---|
| 841 | |
|---|
| 842 | The program fbmatchup will match observations from two feedback files. The program is called |
|---|
| 843 | in the following way: |
|---|
| 844 | |
|---|
| 845 | \begin{alltt} |
|---|
| 846 | \footnotesize |
|---|
| 847 | \begin{verbatim} |
|---|
| 848 | fbmatchup.exe outputfile inputfile1 varname1 inputfile2 varname2 ... |
|---|
| 849 | \end{verbatim} |
|---|
| 850 | \end{alltt} |
|---|
| 851 | |
|---|
| 852 | |
|---|
| 853 | \subsubsection{fbprint} |
|---|
| 854 | |
|---|
| 855 | The program fbprint will print the contents of a feedback file or files to standard output. |
|---|
| 856 | Selected information can be output using optional arguments. The program is called in the |
|---|
| 857 | following way: |
|---|
| 858 | |
|---|
| 859 | \begin{alltt} |
|---|
| 860 | \footnotesize |
|---|
| 861 | \begin{verbatim} |
|---|
| 862 | fbprint.exe [options] inputfile |
|---|
| 863 | |
|---|
| 864 | options: |
|---|
| 865 | -b shorter output |
|---|
| 866 | -q Select observations based on QC flags |
|---|
| 867 | -Q Select observations based on QC flags |
|---|
| 868 | -B Select observations based on QC flags |
|---|
| 869 | -u unsorted |
|---|
| 870 | -s ID select station ID |
|---|
| 871 | -t TYPE select observation type |
|---|
| 872 | -v NUM1-NUM2 select variable range to print by number |
|---|
| 873 | (default all) |
|---|
| 874 | -a NUM1-NUM2 select additional variable range to print by number |
|---|
| 875 | (default all) |
|---|
| 876 | -e NUM1-NUM2 select extra variable range to print by number |
|---|
| 877 | (default all) |
|---|
| 878 | -d output date range |
|---|
| 879 | -D print depths |
|---|
| 880 | -z use zipped files |
|---|
| 881 | \end{verbatim} |
|---|
| 882 | \end{alltt} |
|---|
| 883 | |
|---|
| 884 | \subsubsection{fbsel} |
|---|
| 885 | |
|---|
| 886 | The program fbsel will select or subsample observations. The program is called in the |
|---|
| 887 | following way: |
|---|
| 888 | |
|---|
| 889 | \begin{alltt} |
|---|
| 890 | \footnotesize |
|---|
| 891 | \begin{verbatim} |
|---|
| 892 | fbsel.exe <input filename> <output filename> |
|---|
| 893 | \end{verbatim} |
|---|
| 894 | \end{alltt} |
|---|
| 895 | |
|---|
| 896 | \subsubsection{fbstat} |
|---|
| 897 | |
|---|
| 898 | The program fbstat will output summary statistics in different global areas into a number of |
|---|
| 899 | files. The program is called in the following way: |
|---|
| 900 | |
|---|
| 901 | \begin{alltt} |
|---|
| 902 | \footnotesize |
|---|
| 903 | \begin{verbatim} |
|---|
| 904 | fbstat.exe [-nmlev] <filenames> |
|---|
| 905 | \end{verbatim} |
|---|
| 906 | \end{alltt} |
|---|
| 907 | |
|---|
| 908 | \subsubsection{fbthin} |
|---|
| 909 | |
|---|
| 910 | The program fbthin will thin the data to 1 degree resolution. The code could easily be |
|---|
| 911 | modified to thin to a different resolution. The program is called in the following way: |
|---|
| 912 | |
|---|
| 913 | \begin{alltt} |
|---|
| 914 | \footnotesize |
|---|
| 915 | \begin{verbatim} |
|---|
| 916 | fbthin.exe inputfile outputfile |
|---|
| 917 | \end{verbatim} |
|---|
| 918 | \end{alltt} |
|---|
| 919 | |
|---|
| 920 | \subsubsection{sla2fb} |
|---|
| 921 | |
|---|
| 922 | The program sla2fb will convert an AVISO SLA format file to feedback format. The program is |
|---|
| 923 | called in the following way: |
|---|
| 924 | |
|---|
| 925 | \begin{alltt} |
|---|
| 926 | \footnotesize |
|---|
| 927 | \begin{verbatim} |
|---|
| 928 | sla2fb.exe [-s type] outputfile inputfile1 inputfile2 ... |
|---|
| 929 | |
|---|
| 930 | Option: |
|---|
| 931 | -s Select altimeter data_source |
|---|
| 932 | \end{verbatim} |
|---|
| 933 | \end{alltt} |
|---|
| 934 | |
|---|
| 935 | \subsubsection{vel2fb} |
|---|
| 936 | |
|---|
| 937 | The program vel2fb will convert TAO/PIRATA/RAMA currents files to feedback format. The program |
|---|
| 938 | is called in the following way: |
|---|
| 939 | |
|---|
| 940 | \begin{alltt} |
|---|
| 941 | \footnotesize |
|---|
| 942 | \begin{verbatim} |
|---|
| 943 | vel2fb.exe outputfile inputfile1 inputfile2 ... |
|---|
| 944 | \end{verbatim} |
|---|
| 945 | \end{alltt} |
|---|
| 946 | |
|---|
| 947 | \subsection{building the obstools} |
|---|
| 948 | |
|---|
| 949 | To build the obstools use in the tools directory use ./maketools -n OBSTOOLS -m [ARCH]. |
|---|
| 950 | |
|---|
| 951 | \subsection{Dataplot} |
|---|
| 952 | |
|---|
| 953 | An IDL program called dataplot is included which uses a graphical interface to visualise |
|---|
| 954 | observations and feedback files. It is possible to zoom in, plot individual profiles and |
|---|
| 955 | calculate some basic statistics. To plot some data run IDL and then: |
|---|
| 956 | \begin{alltt} |
|---|
| 957 | \footnotesize |
|---|
| 958 | \begin{verbatim} |
|---|
| 959 | IDL> dataplot, "filename" |
|---|
| 960 | \end{verbatim} |
|---|
| 961 | \end{alltt} |
|---|
| 962 | |
|---|
| 963 | To read multiple files into dataplot, for example multiple feedback files from different |
|---|
| 964 | processors or from different days, the easiest method is to use the spawn command to generate |
|---|
| 965 | a list of files which can then be passed to dataplot. |
|---|
| 966 | \begin{alltt} |
|---|
| 967 | \footnotesize |
|---|
| 968 | \begin{verbatim} |
|---|
| 969 | IDL> spawn, 'ls profb*.nc', files |
|---|
| 970 | IDL> dataplot, files |
|---|
| 971 | \end{verbatim} |
|---|
| 972 | \end{alltt} |
|---|
| 973 | |
|---|
| 974 | Fig~\ref{fig:obsdataplotmain} shows the main window which is launched when dataplot starts. |
|---|
| 975 | This is split into three parts. At the top there is a menu bar which contains a variety of |
|---|
| 976 | drop down menus. Areas - zooms into prespecified regions; plot - plots the data as a |
|---|
| 977 | timeseries or a T-S diagram if appropriate; Find - allows data to be searched; Config - sets |
|---|
| 978 | various configuration options. |
|---|
| 979 | |
|---|
| 980 | The middle part is a plot of the geographical location of the observations. This will plot the |
|---|
| 981 | observation value, the model background value or observation minus background value depending |
|---|
| 982 | on the option selected in the radio button at the bottom of the window. The plotting colour |
|---|
| 983 | range can be changed by clicking on the colour bar. The title of the plot gives some basic |
|---|
| 984 | information about the date range and depth range shown, the extreme values, and the mean and |
|---|
| 985 | rms values. It is possible to zoom in using a drag-box. You may also zoom in or out using the |
|---|
| 986 | mouse wheel. |
|---|
| 987 | |
|---|
| 988 | The bottom part of the window controls what is visible in the plot above. There are two bars |
|---|
| 989 | which select the level range plotted (for profile data). The other bars below select the date |
|---|
| 990 | range shown. The bottom of the figure allows the option to plot the mean, root mean square, |
|---|
| 991 | standard deviation or mean square values. As mentioned above you can choose to plot the |
|---|
| 992 | observation value, the model background value or observation minus background value. The next |
|---|
| 993 | group of radio buttons selects the map projection. This can either be regular latitude |
|---|
| 994 | longitude grid, or north or south polar stereographic. The next group of radio buttons will |
|---|
| 995 | plot bad observations, switch to salinity and plot density for profile observations. The |
|---|
| 996 | rightmost group of buttons will print the plot window as a postscript, save it as png, or exit |
|---|
| 997 | from dataplot. |
|---|
| 998 | |
|---|
| 999 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 1000 | \begin{figure} \begin{center} |
|---|
| 1001 | %\includegraphics[width=10cm,height=12cm,angle=-90.]{./TexFiles/Figures/Fig_OBS_dataplot_main} |
|---|
| 1002 | \includegraphics[width=9cm,angle=-90.]{./TexFiles/Figures/Fig_OBS_dataplot_main} |
|---|
| 1003 | \caption{ \label{fig:obsdataplotmain} |
|---|
| 1004 | Main window of dataplot.} |
|---|
| 1005 | \end{center} \end{figure} |
|---|
| 1006 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 1007 | |
|---|
| 1008 | If a profile point is clicked with the mouse button a plot of the observation and background |
|---|
| 1009 | values as a function of depth (Fig~\ref{fig:obsdataplotprofile}). |
|---|
| 1010 | |
|---|
| 1011 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 1012 | \begin{figure} \begin{center} |
|---|
| 1013 | %\includegraphics[width=10cm,height=12cm,angle=-90.]{./TexFiles/Figures/Fig_OBS_dataplot_prof} |
|---|
| 1014 | \includegraphics[width=7cm,angle=-90.]{./TexFiles/Figures/Fig_OBS_dataplot_prof} |
|---|
| 1015 | \caption{ \label{fig:obsdataplotprofile} |
|---|
| 1016 | Profile plot from dataplot produced by right clicking on a point in the main window.} |
|---|
| 1017 | \end{center} \end{figure} |
|---|
| 1018 | %>>>>>>>>>>>>>>>>>>>>>>>>>>>> |
|---|
| 1019 | |
|---|
| 1020 | |
|---|
| 1021 | |
|---|
| 1022 | |
|---|
