252 | | === ENSRF/EAKF (filtertype=PDAF_DA_ENSRF=13) === |
253 | | |
254 | | {{{ |
255 | | PDAF Available options for ENSRF: |
256 | | PDAF --- Sub-types (Parameter subtype) --- |
257 | | PDAF 0: ENSRF with serial observation processing (cf. Houtekamer/Mitchell, 2002) |
258 | | PDAF 1: EAKF/2-step local least squares filter (cf. Anderson, 2003) |
259 | | PDAF --- Integer parameters (Array param_int) --- |
260 | | PDAF param_int(1): dim_p |
261 | | PDAF Dimension of state vector (>0), required |
262 | | PDAF param_int(2): dim_ens |
263 | | PDAF Ensemble size (>0), required |
264 | | PDAF param_int(3): not used |
265 | | PDAF param_int(4): not used |
266 | | PDAF param_int(5): not used |
267 | | PDAF param_int(6): not used |
268 | | PDAF param_int(7): not used |
269 | | PDAF param_int(8): observe_ens |
270 | | PDAF Application of observation operator H, optional |
271 | | PDAF 0: Apply H to ensemble mean to compute innovation |
272 | | PDAF 1: Apply H to ensemble states; then compute innovation from their mean (default) |
273 | | PDAF param_int(8)=1 is the recomended choice for nonlinear H |
274 | | PDAF param_int(9): type_obs_init |
275 | | PDAF Initialize observations before or after call to prepoststep_pdaf |
276 | | PDAF 0: Initialize observations before call to prepoststep_pdaf |
277 | | PDAF 1: Initialize observations after call to prepoststep_pdaf (default) |
278 | | PDAF --- Floating point parameters (Array param_real) --- |
279 | | PDAF param_real(1): forget |
280 | | PDAF Forgetting factor (usually >0 and <=1), required |
281 | | PDAF --- Further parameters --- |
282 | | PDAF n_modeltasks: Number of parallel model integration tasks |
283 | | PDAF (>=1; not larger than total number of processors) |
284 | | PDAF screen: Control verbosity of PDAF |
285 | | PDAF 0: no outputs |
286 | | PDAF 1: basic output (default) |
287 | | PDAF 2: 1 plus timing output |
288 | | PDAF 3: 2 plus debug output |
289 | | PDAF +++++++++ End of option overview for the ENSRF ++++++++++ |
290 | | }}} |
291 | | |
292 | | |
| 312 | }}} |
| 313 | |
| 314 | |
| 315 | |
| 316 | === ENSRF/EAKF (filtertype=PDAF_DA_ENSRF=13) === |
| 317 | |
| 318 | {{{ |
| 319 | PDAF Available options for ENSRF: |
| 320 | PDAF --- Sub-types (Parameter subtype) --- |
| 321 | PDAF 0: ENSRF with serial observation processing (cf. Houtekamer/Mitchell, 2002) |
| 322 | PDAF 1: EAKF/2-step local least squares filter (cf. Anderson, 2003) |
| 323 | PDAF --- Integer parameters (Array param_int) --- |
| 324 | PDAF param_int(1): dim_p |
| 325 | PDAF Dimension of state vector (>0), required |
| 326 | PDAF param_int(2): dim_ens |
| 327 | PDAF Ensemble size (>0), required |
| 328 | PDAF param_int(3): not used |
| 329 | PDAF param_int(4): not used |
| 330 | PDAF param_int(5): not used |
| 331 | PDAF param_int(6): not used |
| 332 | PDAF param_int(7): not used |
| 333 | PDAF param_int(8): observe_ens |
| 334 | PDAF Application of observation operator H, optional |
| 335 | PDAF 0: Apply H to ensemble mean to compute innovation |
| 336 | PDAF 1: Apply H to ensemble states; then compute innovation from their mean (default) |
| 337 | PDAF param_int(8)=1 is the recomended choice for nonlinear H |
| 338 | PDAF param_int(9): type_obs_init |
| 339 | PDAF Initialize observations before or after call to prepoststep_pdaf |
| 340 | PDAF 0: Initialize observations before call to prepoststep_pdaf |
| 341 | PDAF 1: Initialize observations after call to prepoststep_pdaf (default) |
| 342 | PDAF --- Floating point parameters (Array param_real) --- |
| 343 | PDAF param_real(1): forget |
| 344 | PDAF Forgetting factor (usually >0 and <=1), required |
| 345 | PDAF --- Further parameters --- |
| 346 | PDAF n_modeltasks: Number of parallel model integration tasks |
| 347 | PDAF (>=1; not larger than total number of processors) |
| 348 | PDAF screen: Control verbosity of PDAF |
| 349 | PDAF 0: no outputs |
| 350 | PDAF 1: basic output (default) |
| 351 | PDAF 2: 1 plus timing output |
| 352 | PDAF 3: 2 plus debug output |
| 353 | PDAF +++++++++ End of option overview for the ENSRF ++++++++++ |