Changes between Version 63 and Version 64 of ticket/0927_Energy_diag
- Timestamp:
- 2012-06-28T21:58:43+02:00 (9 years ago)
fabien.roquet
Legend:
- Unmodified
- Added
- Removed
- Modified
-
ticket/0927_Energy_diag
v63 v64 12 12 ''' Motivation: ''' output 3D trends of tracers, momentum, kinetic energy and potential energy.[[BR]] ''' Status :''' the extraction of trends terms exists, but not the 3D output of the trends [[BR]] ''' Main tasks :''' [[BR]] 13 13 14 (1) implement the 3D output of tracers and momentum trends using iom_put [[BR]] (2) compute and output the 3D trends of PE and KE [[BR]](3) validatation + documentation [[BR]]14 (1) implement the 3D output of tracers and momentum trends using iom_put [[BR]] (2) compute and output the 3D trends of PE and KE [[BR]] (3) validatation + documentation [[BR]] 15 15 16 16 ''' Science Reviewer:''' NOCS guy? [[BR]] ''' System Reviewer:''' NOCS guy? [[BR]] ''' Deadline:''' spring 2012 [[BR]] ''' Priority:''' high [[BR]] ''' Depends on:''' gurvan disponibilities [[BR]] ''' Principal Investigator : ''' Gurvan Madec and Fabien Roquet (gurvan.madec@locean-ipsl.upmc.fr) [[BR]] [[BR]] … … 24 24 see [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3316 revision 3316] 25 25 26 ''' trdmod_oce''' module and ''' namtrd namelist'''[[BR]]26 ''' trdmod_oce''' module and ''' namtrd namelist''' [[BR]] 27 27 28 28 logical flags added in namlist namtrd which now controls what is done with the trends. [[BR]] All the types of treatment of a given trend are available at the same time. The memory requirement will only increase due to the time averaged arrays defined in IOM. … … 40 40 Add these new logical in the namelist. '''==>>> CAUTION only in the ORCA2_LIM directory''' [[BR]] NB: here is the new name set in revision number 3318[[BR]] 41 41 42 ''' trdtra'''module[[BR]]42 ''' trdtra''' module[[BR]] 43 43 44 44 Add a systematic mask of the trend.[[BR]] Change the comments to better describe the purpose of this module. Its purpose is: [[BR]] 45 45 46 'TRA' case: to regroup T & S trends and send them to trd_mod, with, in case of advection, transform the incoming advective fluxes into advctive trend (U.grad[T])[[BR]] 'TRC' case: send trend to ted_mod_trc, with, in case of advection, transform the incoming advective fluxes into advective trend [[BR]] all cases : mask the trend (''' ===>>> PROBABLY add in the module a lbc_lnk so that the trend is defined everywhere''')47 48 ''' dynadv_cen2 and _ubs'''modules[[BR]]46 'TRA' case: to regroup T & S trends and send them to trd_mod, with, in case of advection, transform the incoming advective fluxes into advctive trend (U.grad[T])[[BR]] 'TRC' case: send trend to ted_mod_trc, with, in case of advection, transform the incoming advective fluxes into advective trend [[BR]] all cases : mask the trend (''' ===>>> PROBABLY add in the module a lbc_lnk so that the trend is defined everywhere''' ) 47 48 ''' dynadv_cen2 and _ubs''' modules[[BR]] 49 49 50 50 change jpdyn_trd_had into jpdyn_trd_keg. Now in flux form _keg corresponds to the horizontal advection trends and _rvo to the metric terms[[BR]] [[BR]] 51 51 52 ''' dynnxt'''module[[BR]]52 ''' dynnxt''' module[[BR]] 53 53 54 54 add the output using sum of the total dyn trend (except asselin time filter) ("utrd_tot", "vtrd_tot") and of the asselin time filter trend ("utrd_atf", "vtrd_atf") but with a shift by one time step[[BR]] [[BR]] 55 55 56 ''' dynvor, trdvor and trdmod_oce'''modules[[BR]]56 ''' dynvor, trdvor and trdmod_oce''' modules[[BR]] 57 57 58 58 suppress the call to trd_mod in the jpdyn_trd_dat case (computation of beta.V) add add the calculation of beta.V term in 'trdvor' in jpvor_pvo case. And obviously suppress jpdyn_trd_dat from trdmod_oce [[BR]] Also suppress the jpdyn_trd_had case horizontal advection for the dynamics is 'keg' + 'vor' ; in case of flux form, 'had' is put in 'keg' and the metric terms is put in 'vor' [[BR]] there is now only 10 trends on the dynamics instead of 12[[BR]] [[BR]] 59 59 60 ''' trdmod'''module[[BR]]60 ''' trdmod''' module[[BR]] 61 61 62 62 1- introduce the new logical namelist parameters[[BR]] 2- introduce new subroutines : '''trd_budget''' : computation of the domain averaged T,T^2^, PE, KE trends formerly computes in trd_mod routine)[[BR]] ''' trd_3Diom''': output of the 3D trends using IOM [[BR]] [[BR]] 63 63 64 ''' trdicp'''module[[BR]]64 ''' trdicp''' module[[BR]] 65 65 66 66 add in trd_twr routine the computation of the vertical diffusive trend on T & S in case of iso-neutral diffusion (ln_traldf_iso=T). These trends ("ttrd_zdfp", "strd_zdfp") name zdfp for "PURE" vertical diffusion trends are output so that by difference with "zdf" trends we can access to the vertical contribution of the iso-neutral operator [[BR]] [[BR]] 67 67 68 ''' iodef.xml'''file[[BR]]68 ''' iodef.xml''' file[[BR]] 69 69 70 70 add all the trends nick name : '''==>>> CAUTION only in the ORCA2_LIM directory''' [[BR]] [[BR]] … … 73 73 ''' Step II : simplification of the structure ''' 74 74 75 ''' trdicp and trdicp_oce'''modules[[BR]]75 ''' trdicp and trdicp_oce''' modules[[BR]] 76 76 77 77 see [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3317 revision 3317] … … 79 79 1- suppress trdicp_oce module (put required variables in trdmod_oce parameters)[[BR]] 2- use the jptra_trd_... instead of jpicp_... 3- move trd_budget subroutine from trdmod to trdicp.F90. Suppress trd_icp routine (i.e. trd_2d, trd_3d) as the work is now simply done in trd_budget.[[BR]] 4- rescan all dyn/tra trend indices. add jptra_trd_zdfp for "PURE Kz dissusive trend when ln_traldf_iso=T (see also changes in tranxt where a call to trd_tra is done just before the swap, so that PURE Kz trends can be diagnosed in tra_trd. See also xml file). 5- jptra_trd_nsr and jptra_trd_cdt are 3D trends as their incorporate both surface forcing AND runoff, the later being possibly spread in depth ('''==>>> probably to be changed''')[[BR]] 6- suppress the key_trddyn key trdtra from trdicp[[BR]] 80 80 81 ''' trdtra - trddyn - trdmod - trdvor''' modules : impact on ''' TRA''' , ''' DYN''' , ''' TRD''' and almost all ''' TOP_SRC'''modules[[BR]]81 ''' trdtra - trddyn - trdmod - trdvor''' modules : impact on ''' TRA''' , ''' DYN''' , ''' TRD''' and almost all ''' TOP_SRC''' modules[[BR]] 82 82 83 83 see [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3318 revision 3318] … … 94 94 ''' Step III : addition of 3D KE diagnostics ''' 95 95 96 ''' trdken'''module[[BR]]96 ''' trdken''' module[[BR]] 97 97 98 98 see [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3325 revision 3325] … … 109 109 ''' Step IV : addition of 3D PE diagnostics ''' 110 110 111 ''' trdpen'''module[[BR]]111 ''' trdpen''' module[[BR]] 112 112 113 113 see [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3326 revision 3326] and [https://forge.ipsl.jussieu.fr/nemo/log/branches/2012/dev_r3309_LOCEAN12_Ediag/NEMOGCM/NEMO/OPA_SRC?rev=3327 revision 3327] 0- correct a bug in trdtra and trdken … … 125 125 [[BR]] [[BR]] [[BR]] [[BR]] [[BR]] [[BR]] [[BR]] [[BR]] [[BR]] '''Pending issues''' : 126 126 127 atmospheric pressure gradient trend not taken into account (see dynspg.F90 ''' To be done !!!!''' [[BR]]'''128 129 kpp non-local trend put in zdf trends !!! this will not work ! a additional trend term should be add ''' To be done !!!!''' [[BR]]'''127 atmospheric pressure gradient trend not taken into account (see dynspg.F90 ''' To be done !!!!''' [[BR]] ''' 128 129 kpp non-local trend put in zdf trends !!! this will not work ! a additional trend term should be add ''' To be done !!!!''' [[BR]] ''' 130 130 131 131 problems to be solved: vvl case for tracer sad trends ; flux form case for had (keg) and zad momentum trends[[BR]] … … 135 135 create the momentum diag over the ML 136 136 137 reshape trdtra so that T and S are treated separately in all ted routine (including mld diag...) create umask_i and vmask_i (2D) fields that mirror task_i field but for the velocity points 138 '''''' 137 reshape trdtra so that T and S are treated separately in all ted routine (including mld diag...) create umask_i and vmask_i (2D) fields that mirror task_i field but for the velocity points ''''' '''''' 139 138 140 139 '''Changes done by Fabien R.''' … … 142 141 In oce.F90: 143 142 144 * Added 3D variables alpha and beta, thermal and haline expansion coefs.[[BR]]To avoid name conflict, replace variables alph and beta in sol_oce.F90 and solpcg.F90 by sol_alph and sol_beta resp.145 * rhd and rhop are kept for backward compatibility, but they are meant to become obsolete[[BR]](especially rhop which could alreadybe easily removed, with minor changes in zdfkpp, zdfmxl and zdfric)143 * Added 4D variable alpbet: alpbet(:,:,:,jp_tem) is the thermal expansion coef, and alpbet(:,:,:,jp_sal) the haline one. 144 * rhd and rhop are kept for backward compatibility, but they are meant to become obsolete[[BR]](especially rhop which could be easily removed, with minor changes in zdfkpp, zdfmxl and zdfric) 146 145 147 146 In step.F90: … … 171 170 In trdglo.F90: 172 171 173 * In trd_glo_init, corrected definition for tvolt:[[BR]] 172 * In trd_glo_init, corrected definition for tvolt:[[BR]] tvolt = tvolt + SUM( e1e2t(:,:) * fse3t(:,:,jk) * tmask(:,:,jk) * tmask_i(:,:) ) 174 173 * Treatment of bfr has been clarified: if bfr is implicit (ln_bfrimp=.TRUE.), the bottom friction term remain 0, as bottom friction is implicitly included in the vertical diffusion term. 175 174 * Correction of outputs of dynamic trends and hke. … … 223 222 In eosbn2.F90 224 223 225 * Add Vallis equation of state (Vallis 2006, p34-35): most simple form of EOS which accounts for thermobaricity, cabelling and compressibility. 224 * Add Vallis equation of state (Vallis 2006, p34-35): most simple form of EOS which accounts for thermobaricity, cabelling and compressibility.[[BR]]Generalization of linear equation of state, so T-linear and TS-linear cases have been removed. Vallis EOS: nn_eos=1. 226 225 * STRANGE: nn_eos=0 case is a modified version of the Jackett and McDougall (1995) EOS!! Numerically close though... 227 226 * Add original [wiki:McDougall Jackett and McDougall] (1995) EOS (nn_eos=-1) 228 * Buoyancy frequency calculation: exact formulation. 229 * Same for alpha and beta 230 * eos_alpbet modified, which now provides alpha and beta instead of alpha/beta and beta0=0 or 1.[[BR]]ldf_slp_grif in ldfslp.F90 modified accordingly (only place eos_alpbet was used). 231 * eos_bn2 modified, calling first eos_alpbet to update alpha and beta coefficients, then interpolating alpha/beta on w-points and finally calculating pn2. 232 * compute_pen added to provide PE anomaly, and its partial derivatives with respect to T and S, needed in PE diagnostic (trdpen).[[BR]]PE anomaly is the anomaly of PE: (PE - rau0*gz)/(rau0*gz). For a linear case, PE anomaly is equal to density anomaly (nice isn't it?)[[BR]]Done for linear case, and Vallis case.[[BR]]Not ready yet for realistic cases... (discussion with Trevor is needed!)[[BR]]When z=0, PE anomaly is always defined asymptotically, converging toward the density anomaly value. 227 * eos_alpbet modified, which now provides alpha and beta (as a 4d variable) instead of alpha/beta and beta0=0 or 1.[[BR]]ldf_slp_grif in ldfslp.F90 modified accordingly (only place eos_alpbet was used).[[BR]]Now, there is no problem if beta=0, because its inverse is never used. 228 * bn2 modified: 229 * use an exact formulation based on each eos 230 * eos_bn2_alpbet: call first eos_alpbet to compute alpha and beta coefficients, then interpolating alpha/beta on w-points and finally calculating pn2. 231 * eos_bn2: alpha and beta coefficients are directly given in arguments, then interpolated to compute bn2. 232 * eos_pen added to compute PE anomaly, and equivalents os alpha and beta for the PE state variable (called alphaPE and betaPE).[[BR]]function is called at each time step trdpen.F90 (PE diagnostic).[[BR]]PE anomaly is defined as: (PE - rau0*gz)/(rau0*gz). For a linear case, PE anomaly is equal to density anomaly (nice isn't it?)[[BR]]When z=0, PE anomaly is defined asymptotically, converging toward the density anomaly value.[[BR]]Waiting for a new EOS from Trevor McDougall that would allow a more efficient computation of each terms. 233 233 * TODO: add global arrays for alpha and beta, computed at the same time than bn2? Indeed, they are used in several places, and are needed to compute bn2 in any case... 234 * BUG: I don't understand why, but the linear case 1 is very slow, contrary to the others. Could it be coming from eos_alpbet?234 * TODO: remove references to rhop, then remove eos_insitu_pot which calculates it. 235 235 236 236 In trdtra.F90 … … 249 249 Bug during compilation: 250 250 251 * make: Circular dynhpg.o <- trddyn.o dependency dropped 252 253 patched by compiling a first time with the line 'USE trddyn' commented in dynhpg.F90, then recompiling with 'USE trddyn' decommented. 251 * make: Circular dynhpg.o <- trddyn.o dependency dropped[[BR]]patched by compiling a first time with the line 'USE trddyn' commented in dynhpg.F90, then recompiling with 'USE trddyn' decommented. 254 252 255 253 Documentation:
