This directory constains several tests:


1. swtc1ref.sh   Williamson et al. test 1.  advection of a cosine bell.
                 Run test with leapfrog, verify l1,l2,max errors have not 
                 changed.
                 cosine bell (k=1) errors: (leapfrog)
                   l1,l2,linf=  0.3080480E+00  0.9562703E-01  0.4883010E-01
                 cosine bell (k=1) errors: (RKSSP)
                   l1,l2,linf=  0.3251433E+00  0.9888528E-01  0.4721481E-01


           
2. swtc5ref.sh   Reference 1 degree resolution solution to test 5. 
                 Errors:  (computed by interpolating HOMME and NCAR T213 
                 solutions to 256x128 Gaussian grid, via plotgrid.ncl)
                    l1,l2,mx = 0.000489944 0.000693564 0.00961167   
                 In addition, you can compare the plots produced by plotgrid.ncl
                 with swtc5_homme.gif and swtc5_t213.gif

                 

3. swtc6ref.sh    A "reference" 1 degree resolution version of test 6,
                  used as a porting/validation test and descirbed 
                  in HOMME's toplevel README.homme file.
                  Plot and compare final height field to rossby_t15.gif
                  TODO:  compute l1,l2,max errors from NCAR T213 solutions

4. semi-implicit.sh   swtc5 semi-implicit run. 
                      TODO:  add plot from reference solution
                      TODO:  compute l1,l2,max errors from NCAR T213 solutions
                    
4. runtest.sh    short, low-res swtc5 and swtc6 conservation check
                 detailed documentation below:

**********************************************************************
runtest.sh:
**********************************************************************
Test cases for the conservative spectral element shallow water model,
advance_nonstag(), in advance_mod.F90

Params.inc should have:

NP    = 8
GRID_STAG =_NONSTAGGER

The code needs to be compiled with -DSPHEREW
(or add #define SPHEREW in advance_mod.F90)

All test cases have exact mass conservation.  Energy conservation
should balance dissipation from any diffusion added to the model, to
O( dt**2 ).


Test case 5 (flow over a mountain)

1. Verify semi-discrete energy conservation.  Run for 1 day, with
   smooth=0 and no filter or viscosity.  Very low resolution case
   (NE=2) so we are not obtaining conservation by over resolving the
   flow.  First run has dt=240, second run has dt=6


                    M-M0/M0    E-E0/E0    ENS-ENS0/ENS0         output
swtc5-ne2.nl      -2.10e-15   -5.68e-11      4.82e-7           swtc5-ne2.out   
swtc5-ne2-dt6.nl  -4.05e-15    2.75e-14      4.94e-7           swtc5-ne2-dt6.out 


2. Conservation with hyper-viscsoity.  Run for 15 days with NE=5,
   smooth=.0025, dt=120, with hyperviscosity (nu=nu_s=1e15) and no
   filter.

                    M-M0/M0    E-E0/E0    ENS-ENS0/ENS0         output
swtc5-ne5.nl       8.58e-15   -3.76e-6      6.91e-5             swtc5-ne5.out

dissipation rate comparison:  dE/dt vs Diffusion term:     -.555e-11,  -.576e-11



Test case 6 (Rossby-Haurwitz Wave)

1. Run for 14 days.  NE=5, smooth=.0025, dt=120, hyperviscsoity with
   nu=nu_s=1e15 and no filter, integration = explicit (leapfrog).  

                    M-M0/M0    E-E0/E0    ENS-ENS0/ENS0         output
swtc6.nl          -2.10e-15   -1.84e-5    -1.38e-3             swtc6.out

dissipation rate comparison:  dE/dt vs Diffusion term:    -.864e-11,  -.831e-11




