Low temperature NA issue phydro_analytical()

[foxeswithdata]

Hello,

P-hydro encounters an issue when running low temperatures.

Set of environmental inputs with issue:

tc  = -13.2327
tg = -13.2327
ppfd = 79.7121
netrad = 250
vpd = 29.9155
co2 = 368.9
pa = 101325
fapar = 0.59343
kphio = 0.055
psi_soil = -0.0410445
rdark = 0.011
vwind = 3

Issue appears with jmax where jmax gets a -nan result in the calculation due to a negative term in a square root.

Thanks!

[jaideep777]

tl;dr:

Change this line in hyd_analytical_solver.h : calc_dpsi_bound() :
double Iabs_bound = pn::zero(use_bound*0.001, use_bound*0.99, f1, 1e-6).root;
to
double Iabs_bound = pn::zero(use_bound*1e-6, use_bound*0.99, f1, 1e-6).root;

---

Detailed explanation:

In this function, I calculate valid bounds for dpsi (valid in the sense that none of the dpsi-dependent terms should go NA, which happens when dpsi is very large or extremely small). This involves constraining dpsi in 2 steps. The upper bound obtained from the 1st step is used to calculate a further constrained upper bound. So the lower bound should technically be 0, but 0 is also numerically unstable, so I had used the upper bound*0.001. However, in extreme cases (very low temp, very negative psi_soil), dpsi needs to be even lower, so this lower bound doesnt work. Setting it to even lower helps. There should ideally be a more robust bounds checking needed here... Happy to discuss if you're interested.