PV control variable code changes for ticket number 381

VarProg_CovPFstats

• USE variable PVcvScheme, line 133.
• INCLUDE Var_LSVtheta, line 210.
• INCLUDE Var_FillLSzm, line 211.
• INCLUDE Var_AllocateLSzm, line 212.
• INCLUDE Var_DeAllocateLSzm, line 213.
• Add code to compute zonal mean linearisation states, line 480.
• Add optional LSzmDealloc flag in call to Var_LSDeallocate to deallocate LSzm structure, lines 625, 693.

VarProg_AnalysePF

• USE variable PVcvScheme from VarMod_TransformInfo, line 188.
• INCLUDE Var_LSVtheta, line 296.
• INCLUDE Var_FillLSzm, line 297.
• INCLUDE Var_AllocateLSzm, line 298.
• Declare LocalFS and set to FieldStatus_global, line 357.
• Add code to call Var_LSVtheta, Var_AllocateLSzm Var_FillLSzm and Var_LSDeallocate if PVcvScheme is switched on, lines 498 to 513.
• Add optional LSzmDealloc flag in call to Var_LSDeallocate to deallocate LSzm structure, lines 844, 1341.
• Add PVcvScheme in call to Var_Utransforms, line 991.

VarMod_TransformInfo

• Declare logical variable PVcvScheme, line 271.
• Declare logical variable LZzmMeanOption, line 275.
• Declare integer variable LSzmIsoThermLevs, line 279.
• Declare integer variable PV_GCR_restart_bal, line 284.
• Declare integer variable PV_GCR_restart_unbal, line 285.
• Declare integer variable PV_GCR_maxiter_bal, line 286.
• Declare integer variable PV_GCR_maxiter_unbal, line 287.
• Declare integer variable PV_GCR_diagnostics, line 288.
• Declare real variable PV_GCR_Error_Tolerance, line 291.
• Declare real variable PV_GCR_Abs_Tolerance, line 292.
• Declare logical variable PV_GCR_use_Abs_Tol, line 293.
• Declare logical variable PV_GCR_use_residual_Tol, line 294.
• Declare integer variable PV_GCR_precon_option_bal, line 295.
• Declare integer variable PV_GCR_precon_option_unbal, line 296.
• Declare real variable PV_GCR_ADI_Mu_bal, line 299.
• Declare real variable PV_GCR_ADI_Mu_unbal, line 300.
• Declare integer variable PV_GCR_ADI_loops_bal, line 301.
• Declare integer variable PV_GCR_ADI_loops_unbal, line 302.
• Declare logical variable PV_GCR_MassWt_bal, line 303.
• Declare logical variable PV_GCR_MassWt_unbal, line 304.
• Add PVcvScheme, LZzmMeanOption, LSzmIsoThermLevs, PV_GCR_restart_bal, PV_GCR_restart_unbal, PV_GCR_maxiter_bal,PV_GCR_maxiter_unbal, PV_GCR_diagnostics, PV_GCR_Error_Tolerance, PV_GCR_Abs_Tolerance, PV_GCR_use_Abs_Tol, PV_GCR_use_residual_Tol, PV_GCR_precon_option_bal, PV_GCR_precon_option_unbal, PV_GCR_ADI_Mu_bal, PV_GCR_ADI_Mu_unbal, PV_GCR_ADI_loops_bal, PV_GCR_ADI_loops_unbal, PV_GCR_MassWt_bal, and PV_GCR_MassWt_unbal to namelist, line 704.

VarMod_LS

• Add new data type LSzm.
• Definition, line 69.
• Declaration inside LS_type, line 232.

• USE variable PVcvScheme, line 56.
• INCLUDE Var_AllocateLSzm, line 64.
• Add routine to call Var_AllocateLSzm, line 401.

• Add optional logical parameter LSzmDealloc, line 3.
• USE variable PVcvScheme, line 54.
• INCLUDE Var_DeAllocateLSzm, line 61.
• Declare LSzmDealloc, line 67.
• Declare LSzmDealloc_local, line 75.
• Allow for presence or non presence of LSzmDealloc, line 83.
• Add routine to call Var_DeAllocateLSzm, line 518.

• Allocation of LSzm structure.

• Deallocation of LSzm structure.

• Set 'calc' flags on each LSzm field to false to indicate that the zonal means are not calculated.

• Calculate LS zonal means from LS.

• Do zonal (or global) means, as required for LSzm fields.

• Do 3-2-1 smoother in latitude (done after zonal mean of LS).

• Print-out zonal means from each LSzm variable. Calls Var_ShowZM.

• Print-out zonal means for a given LSzm variable. Calls Var_ShowZM.

VarMod_Transforms

• USE variable FieldStatus_absent, line 54.
• USE variable StatusFatal, line 58.
• USE variable PVcvScheme, line 84.
• INCLUDE Var_FillLSzm, line 120.
• CALL Var_FillLSzm, line 317, including possibility of terminating program if LS % Vtheta is not precalculated.

• USE variable PVcvScheme, line 137.
• Don't do inverse tests if using PV-based transforms (it takes FAR too long for a routine assimilation run), add IF lines 964-967, END IF, line 1120 and IF, lines 1125-1128 and END IF, line 1337.
• Call extra Var_Utransform if PV scheme used to create a state PF2 which is needed by the adjoint test section, lines 1340-1350.

• USE PVcvScheme, line 76.
• Add optional PV diagnostics parameter in call to Var_TpTransform, line 143.
• Add optional PV diagnostics parameter in call to Var_UpTransform, line 232.

• Add optional argument PV_Diagnostics, line 9.
• Declare PV_Diagnostics, line 104.
• Declare local variable PV_Diagnostics1, line 122.
• Set value of PV_Diagnostics1, lines 129-133.
• Add PV_Diagnostics1 in call to Var_UpTransform, lines 228.

• Add optional argument PV_Diagnostics, line 8.
• Declare PV_Diagnostics, line 81.
• Declare local variable PV_Diagnostics1, line 93.
• Set value of PV_Diagnostics1, lines 99-103.
• Add PV_Diagnostics1 in call to Var_Utransform, lines 117.

VarMod_Vp

• Calculation of balanced streamfunction by PV inversion.

• Calculation of unbalanced pressure by antiPV inversion.

VarMod_TpTransform

• Add optional argument PV_Residuals, line 9.
• USE variable PVcvScheme, line 109.
• INCLUDE Var_VpPsi_B, line 131.
• INCLUDE Var_VpPr_UnB, line 132.
• Declare parameter PV_Residuals, line 141.
• Declare variable PV_Residuals1, line 157.
• Set value of PV_Residuals1, lines 171-175.
• Corrected message section numbers passed to Gen_Trace.
• Add call to Var_VpPsi_B if PVcvScheme switch is set, line 217.
• Add call to Var_VpPr_UnB if PVcvScheme switch is set, line 277.

• Calculation of PV increment from PF (or equivalent) fields.

• To do vertical regression (to remove small vertical modes from first guess streamfunction). This has been removed, 01/05/07.

• Calculation of anti-PV increment from PF (or equivalent) fields.

• GCR solver tailored to the PV problem.

• Calculates either PV or anti-PV.

• Performs crude inverse PV or anti-PV operators with varying degrees of approximation. Options include the identity matrix (ie no preconditioning), diagonal inverse (diagonal preconditioning), vertical inverse (vertical-only preconditioning) and an iterative approximate inverse (ADI preconditioning).

• Initialises structures necessary to perform preconditioning.

• Diagnostic print of absolute error norms (for use in PV GCR solver).

• Calculate maximum value of error over all processors (for use in PV GCR solver).

• Diagnostic print of residual error norms (for use in PV GCR solver).

• Calculates the two norm of the input field divided by the number of points in the field - no mass weighting (for use in PV GCR solver).

• Calculates the two norm of the input field divided by the number of points in the field - with mass weighting (for use in PV GCR solver).

• Calculates inner products - with mass weighting (for use in PV GCR solver).

• Calculates inner products - no mass weighting (for use in PV GCR solver).

VarMod_Transform_p

• USE variable PVcvScheme, line 42.
• Specify Ap halos of 1 (1 is set for psi) if using PV scheme, line 75.

VarMod_UpPF

• Calculation of balanced pressure on psi-points and using a zonal mean linearisation state.

• Calculation of unbalanced streamfunction on p-points and using a zonal mean linearisation state.

• Change arguments from Vp to Psi, Chi fields and their headers, line 6.
• Use Header_type from VarMod_PFInfo , line 57.
• Removed Vp_type from VarMod_TransformInfo, line 63.
• Change declaration of arguments, line 80.
• Change Vp % psi to Psi and Vp % chi to Chi, lines 146, 147, 152, 153, 165, 166, 171, 172.

• Add optional argument PV_Diagnostics, line 8.
• USE FieldsOutput, lines 39-40.
• USE Vp_header from VarMod_TransformInfo , line 60.
• USE PVcvScheme from VarMod_TransformInfo, line 62.
• INCLUDE Var_UpPF_Balp_lszm, line 72.
• INCLUDE Var_FieldOutput, line 73.
• Declare PV_Diagnostics, line 81.
• Declare integer x and y loop variables, line 99.
• Declare p_psi, allocatable real array, line 100.
• Declare PV_Diagnostics1, line 101.
• Set value of PV_Diagnostics1, lines 109-113.
• Modify code to allow PV scheme (including CALL Var_UpPF_Balp_lszm if using PV scheme), lines 140-195.
• Output optional balanced and unbalanced pressures, lines 197-205.

VarMod_UpPF_Adj

• Adjoint of Var_UpPF_Balp_lszm.

• Adjoint of Var_UpPF_Unbalpsi_lszm.

• Change arguments from Vp_hat to Psi_hat, Chi_hat and their headers, line 6.
• Use Header_type from VarMod_PFInfo , line 54.
• Removed Vp_type from VarMod_TransformInfo, line 63.
• Change declaration of arguments, line 89.
• Change Vp_hat % psi to Psi_hat and Vp_hat % chi to Chi_hat, lines 352, 355, 361, 364, 377, 380, 386, 389.

• USE Vp_header from VarMod_TransformInfo , line 52.
• USE PVcvScheme from VarMod_TransformInfo, line 54.
• INCLUDE Var_UpPF_Balp_lszm_Adj, line 73.
• Declare integer x, y, lowz and topz variables, line 96.
• Declare p_psi_adj, allocatable real array, line 97.
• CALL Var_UpPF_Balp_lszm_Adj if using PV scheme, line 169.

• Modify call to Var_UpPFu_v, due to change of arguments, line 1127.
• Modify call to Var_UpPFu_v_Adj, due to change of arguments, line 1193.

VarMod_UpTransform

• Add optional parameter PV_Diagnostics, line 9.
• USE PVcvScheme, line 70.
• USE SwapScalar, line 79-80.
• INCLUDE Var_UpPF_Unbalpsi_lszm, line 107.
• INCLUDE Var_SwapBounds, line 108.
• INCLUDE Var_PFdiagnostics_forPV, line 109.
• Declare PV_Diagnostics, line 120.
• Declare allocatable psi field, line 132.
• Declare allocatable unbalpsi_p field, line 133.
• Declare logical variable Use_normal_modes, line 134.
• Declare integers x, y, LowLev and TopLev loop variables, line 135.
• Declare PV_Diagnostics1, line 136.
• Set value of PV_Diagnostics1, lines 149-153.
• Add code to output optional u,v diagnostics associated with the velocity potential, lines 193-200.
• Add code to output optional u,v diagnostics associated with the balanced streamfunction, lines 211-218.
• Allocate psi (the total streamfunction), lines 220-224.
• Add code for PV scheme (to allow calculation of the unbalanced streamfunction, including allocation and deallocation of unbalpsi_p and code to output optional u,v diagnostics associated with the unbalanced streamfunction), lines 226-274.
• Modify call to Var_UpPFu_v, due to change of arguments, line 280.
• Deallocate psi, line 285.
• Add optional argument PV_Diagnostics1 to call to Var_UpPFHp, line 313.

• To output the u,v associated with either a streamfunction or velocity potential field. Called by Var_UpTransform.

VarMod_UpTransform_Adj

• USE PVcvScheme, line 62.
• USE SwapScalar, line 58.
• INCLUDE Var_UpPF_Unbalpsi_lszm_Adj, line 86.
• INCLUDE Var_SwapBounds_Adj, line 87.
• Declare allocatable psi_hat field, line 107.
• Declare allocatable unbalpsi_p_hat field, line 108.
• Declare logical variable Use_normal_modes, line 109.
• Declare integer x, y, LowLev and TopLev loop variables, line 110.
• Allocate psi_hat, line 314.
• Modify call to Var_UpPFu_v_Adj, due to change of arguments, line 319.
• Add code for PV scheme, line 328.

VarMod_CovVertical

• Modify call to Var_UpPFu_v, due to change of arguments, lines 398 and 440.

VarMod_DelSquaredFFT

• Add new optional parameter Mu_in (screening term), line 12.
• Declare Mu_in, line 143.
• Declare Mu, line 258.
• Deal with possibility of Mu_in being present or not, line 266.
• Add Mu to call to Var_TriDiag_p, line 1036.
• Add Mu to call to Var_TriDiag_psi, line 1053.
• Add Mu to call to Var_TriDiag_LAM_psi, line 1070.
• Add Mu to call to Var_TriDiag_LAM_p, line 1087.
• Add Mu to call to Var_TriDiag_LAM_chi, line 1104.

• Add new optional parameter Mu_in (screening term), line 17.
• Declare Mu_in, line 81.
• Declare Mu, line 98.
• Deal with possibility of Mu_in being present or not, line 104.
• Add Mu to diagonal terms, line 127.

• Add new optional parameter Mu_in (screening term), line 17.
• Declare Mu_in, line 80.
• Declare Mu, line 97.
• Deal with possibility of Mu_in being present or not, line 103.
• Add Mu to diagonal terms, line 127.

• Add new optional parameter Mu_in (screening term), line 17.
• Declare Mu_in, line 80.
• Declare Mu, line 97.
• Deal with possibility of Mu_in being present or not, line 103.
• Add Mu to diagonal terms, line 127.

• Add new optional parameter Mu_in (screening term), line 17.
• Declare Mu_in, line 77.
• Declare Mu, line 94.
• Deal with possibility of Mu_in being present or not, line 100.
• Add Mu to diagonal terms, lines 121, 125, 136.

• Add new optional parameter Mu_in (screening term), line 17.
• Declare Mu_in, line 77.
• Declare Mu, line 95.
• Deal with possibility of Mu_in being present or not, line 107.
• Add Mu to diagonal terms, line 125.

VarMod_PseudoOb

• PseudoObNmax = 10 (was 5), line 79.

VarProg_TestCov

• USE variable PVcvScheme from VarMod_TransformInfo, line 115.
• INCLUDE Var_LSVtheta, line 149.
• INCLUDE Var_FillLSzm, line 150.
• INCLUDE Var_AllocateLSzm, line 151.
• Declare LocalFS and set to FieldStatus_global, line 179.
• Add code to call Var_LSVtheta, Var_AllocateLSzm Var_FillLSzm and Var_LSDeallocate if PVcvScheme is switched on, lines 269 to 281.
• Add optional LSzmDealloc flag in call to Var_LSDeallocate to deallocate LSzm structure, line 405.

Temporary code written / added

• Routine Var_RoughOutput added. I have undone this change.
• Diagnostic code added to Var_TpTransform to output first guess streamfunction, balanced streamfunction, first guess unbalanced pressure, unbalanced pressure. I have undone this change.
• Diagnostic code added to Var_GCR_k_PV to output the residual in PV or antiPV. I have undone this change.

• Diagnostic code added to Var_VpPsi_B to output RHS PV. I have undone this change and put in diagnostic code that is in-line with the standard code.
• Diagnostic code added to Var_VpPr_UnB to output RHS antiPV. I have undone this change and put in diagnostic code that is in-line with the standard code.

• Removed condition on Diagnostic mode to be selected to call Var_ShowZM. I have undone this change.

• Borrowed Marek's code Var_InitTrans_PFs_setup. This is introduced in later versions of Var., but I introduce it now to (hopefully) fix a bug. Need to remove this before merging - removed 30/04/07.
• Add thee lines also in VarProg_CovPFstats to call this routine. Need to remove this before merging - removed 30/04/07.

• Add extra condition MainVariant \= TestTransforms to VarProg_AnalysePF at lines 779, 812, 859 and 896. This is to prevent unnecessary calls to transforms when doing a transform test run. Need to remove this before merging (actually kept this for now - need to check that it will not interfere with normal operation.).
• INCLUDE Var_CheckHeaders_SameDate, line 234. To allow LS and PF to have the same date. Need to remove this before merging - removed 30/04/07.
• Change Gen_CheckHeaders to Var_CheckHeaders_SameDate, line 579. To allow LS and PF to have the same date. Need to remove this before merging - removed 30/04/07.

• Add routine Var_CheckHeaders_SameDate to set the PF header date to the same value as the LS header date. Need to remove this before merging - removed 30/04/07.

• Add printing comments in Var_PFGp to state whether vertical regression on Gp is performed. Need to remove this before merging - removed 30/04/07.

• Commented-out lines detailed in main Var_InitTransforms entry above. Need to remove these before merging if they are not needed - removed 30/04/07.

• Replaced with latest version (Marek's code). This is introduced in later versions of Var., but I introduce it now to (hopefully) fix a bug. Need to replace this before merging - replaced 30/04/07.

Issues to sort out