gettau.f

GETTAU computes the convective overturn timescale $\tau_\mathrm{cz}$ one pressure scale height above the base of the surface convection zone ("G Somers, 3/17"; its comment calls it "a new, systematic method"). Depending on where that point lies, it uses the interior model, the envelope integration, or the two stitched together. In the new method (TAUINTNEW) the timescale is local, $\tau_\mathrm{cz} = H_p/v_\mathrm{conv}$ (PSCA/CVEL), evaluated one $H_p$ above the base; for a zone thinner than $H_p$ it is the crossing time $\Delta r/v_\mathrm{conv}$ (DELR/CVEL, with $\Delta r$ the zone's width and $v_\mathrm{conv}$ at its middle), and with no surface convection zone it is 0.

• Who calls it: GETW, STARIN and WRTOUT.
• Inputs: the model arrays (HCOMP, HR, HP, HD, HS1, HT, LC, M), the form factors FP, FT, the surface TEFFL, BL, HSTOT; the mixing-length parameter CMIXL (an argument since rp_1, passed on to ENVINT); and from SCRTCH_MOD, the interior's convective velocities SVEL and gradients SDEL, left there by an earlier physics pass.
• What it calls: ENVINT (the envelope, which also sets the photospheric pressure PPHOT and, in the old scheme, TAUCZ), then TAUINTNEW or TAUINT.
• Outputs: TAUCZ and PPHOT in OVRTRN_MOD, read by about twenty routines (wind and magnetic braking among them: MWIND, COWIND, DADCOEFT, …); RBCZ (argument), the radius of the convection zone's base in solar radii, set by TAUINTNEW (0 if it finds none); the old method (TAUINT) leaves it unchanged.
• Modules (COMMON blocks until rp_1): OVRTRN_MOD (TAUCZ, PPHOT); ENVSTRUCT_MOD (the envelope solution); SCRTCH_MOD (interior velocities and gradients); NMLPAR_MOD (LNEWTCZ, which selects the new method, and LCALCENV, which tells ENVINT whether to store the envelope's structure, both formerly in /OVRTRN/; the starspot parameters SPOTF, SPOTX, formerly /SPOTS/); DIMENSIONS_MOD and CONSTANTS_MOD.

1. Find the base of the surface convection zone, IMAX, scanning inward from the surface. A fully convective star sets LALLCZ, which nothing reads (unused?, line 91↗).

2. If $\log P_\mathrm{base} - \log P_\mathrm{surf} < 1$ (or the old method is chosen), the point one scale height up may be in the envelope, so LCALCENV is set and the envelope's structure kept.

3. Integrate the envelope with ENVINT from the surface conditions (radius and gravity from $L$ and $T_\mathrm{eff}$, as in GETNEWENV, with the starspot correction to $T_\mathrm{eff}$). In the old method, ENVINT sets TAUCZ itself and the routine ends there.

4. Copy the interior into work arrays, with $g = Gm/r^2$, the velocities and gradients from SCRTCH_MOD: DELF1 = $\nabla_\mathrm{rad}$ (SDEL(1,·)) and DELF2 = $\nabla_\mathrm{ad}$ (SDEL(3,·), as PHYSIC stores them). With the new method and LCALCENV, append the envelope points 2 to NUMENV (the first is the last interior point), taking the same two gradients from EDELS(1,·) and EDELS(2,·), which QENV stores in the order $\nabla_\mathrm{rad}$, $\nabla_\mathrm{ad}$, $\nabla$. The composition is copied too, into HCOMPF, for the envelope points from HCOMP(J,M+1), one point past the model; but since the calls no longer pass HCOMPF, it is read nowhere (unused?, lines 144–147↗, lines 166–169↗).

5. TAUINTNEW (new method) evaluates the overturn time from these arrays, as above; TAUINT (old) computes it its own way.

6. LCALCENV is reset to true.