search

Eawag-AppliedSystemAnalysis / Simstrat

Simstrat - 1D lake model
http://www.eawag.ch/en/department/surf/projects/simstrat/
GNU General Public License v3.0
17 stars 8 forks source link

Pressure gradients not implemented #43

Closed f-baerenbold closed 1 year ago

f-baerenbold commented 3 years ago

Pressure gradient option 1 causes diverging horizontal water velocities and thus all state variables become NaN's pretty quickly (independent of latitude).

f-baerenbold commented 1 year ago

I think that pressure option 1 is the term presented in equation (10) in Svensson and Sahlberg (1989). In Goudsmit et al. (2002) a similar term is shown in equation (34). Goudsmit says that (34) can be added to equation (2) but forgot to mention that one has to divide it by "rho" before. Accordingly, the "rho" was still in the Simstrat code. Deleting "rho" and then using PressureGradients=1 leads to very similar results for temperature than for PressureGradients=0.

What I am not really sure about: In Svensson and Sahlberg (1989), equation (10) is the time derivative of dp/dx. The only time-dependent variable in equation (10) is the mean velocity (U). According to Goudsmit et al. (2008) equation (34), dp/dx = the integral between simulation start and current simulation time. However, in Simstrat we take the current value of "U" * delta_t.

The rho is deleted in commit d1794b5

f-baerenbold commented 1 year ago

The pressure gradient option is deleted now. Option 1 according to Svensson, 1978 was not correctly implemented (see equations A6 and A7 in "Probe An instruction manual" from Svensson (1998) for a possible implementation). Option 2 is actually not a pressure gradient but the addition of bottom friction in the momentum equations to avoid resonance phenomena in lakes close to the equator. Option 2 can now be used using the option "BottomFriction"