345 | | |
346 | | |
| 346 | == NEW VALIDATION TESTS (OCT-NOV 2011) == |
| 347 | |
| 348 | After merging the tag 1.9.5.2 into the externalized version, I launched some new validation tests. [[BR]] |
| 349 | |
| 350 | There are three kind of tests : Fluxnet (local scale), regional and global. [[BR]] |
| 351 | |
| 352 | These tests are not part of a scientific validation of the model but only technical ones. As I have only reorganized the code, I have to check if |
| 353 | I obtain the same results in both cases. |
| 354 | |
| 355 | I used the same forcing files for all tests : ncc forcing ( for the year 1982) except for the test with the weather generator (CRU for the year 1985).[[BR]] |
| 356 | |
| 357 | My referenced version was the TAG 1.9.5.2 with corrections of the two following bugs : black_carbon and herbivorory. |
| 358 | [[BR]] |
| 359 | The tools used are cdo (cdo diffv command), a validation script (ksh + ferret) and ferret (difference maps).[[BR]] |
| 360 | |
| 361 | === FLUXNET sites === |
| 362 | |
| 363 | I launched the following tests : |
| 364 | 1. FLUXNET_REF_OCT2011 (launched with the referenced version) |
| 365 | 2. VAL_FLUXNET_2PFTS_OCT |
| 366 | 3. VAL_FLUXNET_C3crop_to_C3grass |
| 367 | 4. VAL_FLUXNET_C3GRASS_in_14 |
| 368 | 5. VAL_FLUXNET_C3GRASS_SPLIT |
| 369 | The four last ones were launched with the externalized version. The setup of these jobs are the same defined in March (see above) : |
| 370 | ||VAL_FLUXNET_2PFTS_OCT ||setup : 2 PFTS |
| 371 | ||VAL_FLUXNET_C3crop_to_C3grass|| setup : 13 PFTs with PFT_12 reparametrized as PFT 10 |
| 372 | ||VAL_FLUXNET_C3GRASS_in_14 || setup : 14 PFTs with PFT 14 as C3 grass |
| 373 | ||VAL_FLUXNET_C3GRASS_SPLIT || setup : 14 PFTs with C3grass shared between PFT 10 and PFT 14 |
| 374 | |
| 375 | In order to check in teststomate was correctly adapted to the externalized version, I launched and compared the two following jobs : |
| 376 | 1. FLUXNET_TESTSTOMATE (referenced version) : 13 PFTs with vegetation only on PFT1 and PFT 10 |
| 377 | 2. EXT_TESTSTOMATE_FLUXNET : 2PFTS |
| 378 | Both have the following setup : |
| 379 | {{{ |
| 380 | # Initialisation for spin-up : |
| 381 | # orchidee with sechiba alone (!!! if ok_stomate == n !!!) |
| 382 | duree_nostomate=0 |
| 383 | # orchidee with stomate |
| 384 | duree_inistomate=1 |
| 385 | # teststomate (only if duree_inistomate > 0) |
| 386 | duree_offlineini=20 |
| 387 | |
| 388 | # Loop over ORCHIDEE runs (used for spin-up) |
| 389 | # The whole job is restarted n_iter times |
| 390 | n_iter=1 |
| 391 | # orchidee with sechiba (and stomate if ok_stomate=y below) |
| 392 | duree_sechiba=20 |
| 393 | # teststomate |
| 394 | duree_stomate=0 |
| 395 | # forcesoil |
| 396 | duree_carbonsol=0 |
| 397 | |
| 398 | # Final run (full ORCHIDEE) |
| 399 | # This last parameter must be non-zero. |
| 400 | duree_final=20 |
| 401 | |
| 402 | }}} |
| 403 | |
| 404 | The last configuration to test was FORCESOIL. So I launched the following jobs : |
| 405 | 1. TAG_FLUX_FOR (referenced version) : 13 PFTs |
| 406 | 2. EXT_FLUX_FOR : 13 PFTs |
| 407 | The setup was : |
| 408 | |
| 409 | {{{ |
| 410 | # Initialisation for spin-up : |
| 411 | # orchidee with sechiba alone (!!! if ok_stomate == n !!!) |
| 412 | duree_nostomate=0 |
| 413 | # orchidee with stomate |
| 414 | duree_inistomate=1 |
| 415 | # teststomate (only if duree_inistomate > 0) |
| 416 | duree_offlineini=0 |
| 417 | |
| 418 | # Loop over ORCHIDEE runs (used for spin-up) |
| 419 | # The whole job is restarted n_iter times |
| 420 | n_iter=1 |
| 421 | # orchidee with sechiba (and stomate if ok_stomate=y below) |
| 422 | duree_sechiba=20 |
| 423 | # teststomate |
| 424 | duree_stomate=0 |
| 425 | # forcesoil |
| 426 | duree_carbonsol=10000 |
| 427 | |
| 428 | # Final run (full ORCHIDEE) |
| 429 | # This last parameter must be non-zero. |
| 430 | duree_final=20 |
| 431 | }}} |
| 432 | |
| 433 | |
| 434 | |
| 435 | |
| 436 | === Comparisons === |
| 437 | |
| 438 | The comparisons by the three tools were all OK. |
| 439 | All results are available in a directory called VALIDATE_OCTOBRE_2011 on obelix. |
| 440 | |
| 441 | |
| 442 | |
| 443 | == Regional scale == |
| 444 | |
| 445 | Only one test was made over a region of Europe (France I think). |
| 446 | |