# Bug in computation of potential temp from conservative temp (\`eos\_pt\_from\_ct\`) **URL:** \ **Category:** v4.2.x\ **Created:** [11 October 2025 13:34 UTC](https://nemo-ocean.discourse.group/t/bug-in-computation-of-potential-temp-from-conservative-temp-eos-pt-from-ct/953 "2025-10-11T13:34:53Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![rabbit](https://avatars.discourse-cdn.com/v4/letter/r/919ad9/32.png) [@rabbit](https://nemo-ocean.discourse.group/u/rabbit)\ **Post date:** [11 October 2025 13:34 UTC](https://nemo-ocean.discourse.group/t/bug-in-computation-of-potential-temp-from-conservative-temp-eos-pt-from-ct/953/1 "2025-10-11T13:34:53Z")
We are running nemo with TEOS-10 and open cavities. We came across the bug that the output variable `toce_pot`(thetao) is masked with surface tmask In `src/OCE/TRA/eosbn2.F90` on line 1086: ```auto ztm = tmask(ji,jj,1) ``` Which is then applied to the derived potential temperature ```auto ptmp(ji,jj) = ( zt / z1_T0 + zn / zd ) * ztm ``` I think that there is no need to perform multiplication by `ztm` at all, which erroneously masks open cavities. `ztm` can be removed from the routine. I checked that the bug is present in v5 as well. Could you add the fix on nemo forge, @smasson so that this is not forgotten? And thank you for fixing hfbasin, we recently pulled in the patch to nemo4.2 ---
**Author:** ![Dorothee](https://avatars.discourse-cdn.com/v4/letter/d/edb3f5/32.png) [@Dorothee](https://nemo-ocean.discourse.group/u/Dorothee)\ **Post date:** [16 October 2025 20:47 UTC](https://nemo-ocean.discourse.group/t/bug-in-computation-of-potential-temp-from-conservative-temp-eos-pt-from-ct/953/2 "2025-10-16T20:47:03Z")
Hi! Here is a fix. I have added a 3d version of eos\_pt\_from\_ct In `src/OCE/TRA/eosbn2.F90:` ```auto FUNCTION eos_pt_from_ct_3d( ctmp, psal ) RESULT( ptmp ) !!---------------------------------------------------------------------- !! ***ROUTINE eos_pt_from_ct*** !! !! ** Purpose : Compute pot.temp. from cons. temp. [Celsius] !! !! ** Method : rational approximation (5/3th order) of TEOS-10 algorithm !! checkvalue: pt=20.02391895 Celsius for sa=35.7g/kg, ct=20degC !! !! Reference : TEOS-10, UNESCO !! Rational approximation to TEOS10 algorithm (rms error on WOA13 values: 4.0e-5 degC) !!---------------------------------------------------------------------- REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(in ) :: ctmp ! Cons. Temp [Celsius] REAL(wp), DIMENSION(jpi,jpj,jpk), INTENT(in ) :: psal ! salinity [psu] ! Leave result array automatic rather than making explicitly allocated REAL(wp), DIMENSION(jpi,jpj,jpk) :: ptmp ! potential temperature [Celsius] ! INTEGER :: ji, jj, jk ! dummy loop indices REAL(wp) :: zt , zs , ztm ! local scalars REAL(wp) :: zn , zd ! local scalars REAL(wp) :: zdeltaS , z1_S0 , z1_T0 !!---------------------------------------------------------------------- ! IF( ln_timing ) CALL timing_start('eos_pt_from_ct_3d') ! zdeltaS = 5._wp z1_S0 = 0.875_wp/35.16504_wp z1_T0 = 1._wp/40._wp ! DO_3D( nn_hls, nn_hls, nn_hls, nn_hls, 1, jpkm1 ) ! zt = ctmp (ji,jj,jk) * z1_T0 zs = SQRT( ABS( psal(ji,jj,jk) + zdeltaS ) * z1_S0 ) !jkt = mikt(ji,jj) ztm = tmask(ji,jj,jk) ! zn = ((((-2.1385727895e-01_wp*zt & & - 2.7674419971e-01_wp*zs+1.0728094330_wp)*zt & & + (2.6366564313_wp*zs+3.3546960647_wp)*zs-7.8012209473_wp)*zt & & + ((1.8835586562_wp*zs+7.3949191679_wp)*zs-3.3937395875_wp)*zs-5.6414948432_wp)*zt & & + (((3.5737370589_wp*zs-1.5512427389e+01_wp)*zs+2.4625741105e+01_wp)*zs & & +1.9912291000e+01_wp)*zs-3.2191146312e+01_wp)*zt & & + ((((5.7153204649e-01_wp*zs-3.0943149543_wp)*zs+9.3052495181_wp)*zs & & -9.4528934807_wp)*zs+3.1066408996_wp)*zs-4.3504021262e-01_wp ! zd = (2.0035003456_wp*zt & & -3.4570358592e-01_wp*zs+5.6471810638_wp)*zt & & + (1.5393993508_wp*zs-6.9394762624_wp)*zs+1.2750522650e+01_wp ! ptmp(ji,jj,jk) = ( zt / z1_T0 + zn / zd ) * ztm ! END_3D ! IF( ln_timing ) CALL timing_stop('eos_pt_from_ct_3d') ! END FUNCTION eos_pt_from_ct_3d ``` In `src/OCE/DIA/diaar5.F90`: ```auto IF( ln_teos10 ) THEN ! ! potential temperature (TEOS-10 case) IF( iom_use( 'toce_pot') .OR. iom_use( 'temptot_pot' ) .OR. iom_use( 'sst_pot' ) & .OR. iom_use( 'ssttot' ) .OR. iom_use( 'tosmint_pot' ) ) THEN ! ALLOCATE( ztpot(jpi,jpj,jpk) ) ztpot(:,:,jpk) = 0._wp ! ztpot(:,:,:) =eos_pt_from_ct_3d( ts(:,:,:,jp_tem,Kmm), ts(:,:,:,jp_sal,Kmm) ) ! CALL iom_put( 'toce_pot', ztpot(:,:,:) ) ! potential temperature (TEOS-10 case) CALL iom_put( 'sst_pot' , ztpot(:,:,1) ) ! surface temperature ``` ---
**Author:** ![Dorothee](https://avatars.discourse-cdn.com/v4/letter/d/edb3f5/32.png) [@Dorothee](https://nemo-ocean.discourse.group/u/Dorothee)\ **Post date:** [17 October 2025 20:10 UTC](https://nemo-ocean.discourse.group/t/bug-in-computation-of-potential-temp-from-conservative-temp-eos-pt-from-ct/953/3 "2025-10-17T20:10:43Z")
More elegant way: ```auto FUNCTION eos_pt_from_ct( ctmp, psal, pjk ) RESULT( ptmp ) !!---------------------------------------------------------------------- !! ***ROUTINE eos_pt_from_ct*** !! !! ** Purpose : Compute pot.temp. from cons. temp. [Celsius] !! !! ** Method : rational approximation (5/3th order) of TEOS-10 algorithm !! checkvalue: pt=20.02391895 Celsius for sa=35.7g/kg, ct=20degC !! !! Reference : TEOS-10, UNESCO !! Rational approximation to TEOS10 algorithm (rms error on WOA13 values: 4.0e-5 degC) !!---------------------------------------------------------------------- REAL(wp), DIMENSION(jpi,jpj), INTENT(in ) :: ctmp ! Cons. Temp [Celsius] REAL(wp), DIMENSION(jpi,jpj), INTENT(in ) :: psal ! salinity [psu] INTEGER, OPTIONAL, INTENT(in ) :: pjk ! level for tmask ! Leave result array automatic rather than making explicitly allocated REAL(wp), DIMENSION(jpi,jpj) :: ptmp ! potential temperature [Celsius] ! INTEGER :: ji, jj, zjk ! dummy loop indices REAL(wp) :: zt , zs , ztm ! local scalars REAL(wp) :: zn , zd ! local scalars REAL(wp) :: zdeltaS , z1_S0 , z1_T0 !!---------------------------------------------------------------------- ! IF( ln_timing ) CALL timing_start('eos_pt_from_ct') ! zdeltaS = 5._wp z1_S0 = 0.875_wp/35.16504_wp z1_T0 = 1._wp/40._wp ! IF (PRESENT(pjk)) THEN zjk = pjk ELSE zjk = 1 ! default value END IF ! DO_2D( nn_hls, nn_hls, nn_hls, nn_hls ) ! zt = ctmp (ji,jj) * z1_T0 zs = SQRT( ABS( psal(ji,jj) + zdeltaS ) * z1_S0 ) ztm = tmask(ji,jj,zjk) ! zn = ((((-2.1385727895e-01_wp*zt & & - 2.7674419971e-01_wp*zs+1.0728094330_wp)*zt & & + (2.6366564313_wp*zs+3.3546960647_wp)*zs-7.8012209473_wp)*zt & & + ((1.8835586562_wp*zs+7.3949191679_wp)*zs-3.3937395875_wp)*zs-5.6414948432_wp)*zt & & + (((3.5737370589_wp*zs-1.5512427389e+01_wp)*zs+2.4625741105e+01_wp)*zs & & +1.9912291000e+01_wp)*zs-3.2191146312e+01_wp)*zt & & + ((((5.7153204649e-01_wp*zs-3.0943149543_wp)*zs+9.3052495181_wp)*zs & & -9.4528934807_wp)*zs+3.1066408996_wp)*zs-4.3504021262e-01_wp ! zd = (2.0035003456_wp*zt & & -3.4570358592e-01_wp*zs+5.6471810638_wp)*zt & & + (1.5393993508_wp*zs-6.9394762624_wp)*zs+1.2750522650e+01_wp ! ptmp(ji,jj) = ( zt / z1_T0 + zn / zd ) * ztm ! END_2D ! IF( ln_timing ) CALL timing_stop('eos_pt_from_ct') ! END FUNCTION eos_pt_from_ct ```