rodrigo-caballero / climt-legacy Goto Github PK

I'm trying to install this library on my mac using the gnu95 compiler and running into this error:
ld: symbol(s) not found for architecture x86_64 collect2: error: ld returned 1 exit status error: Command "/usr/local/bin/gfortran -Wall -g -L/opt/boxen/homebrew/lib /var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/src.macosx-10.10-x86_64-2.7/_gridmodule.o /var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/src.macosx-10.10-x86_64-2.7/fortranobject.o /var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/src/grid/Driver.o /var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/src/grid/src/dimensions.o /var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/var/folders/r5/qd2yr2tx55gf5sn_29hkz63c0000gn/T/tmpuYUrO2/src.macosx-10.10-x86_64-2.7/_grid-f2pywrappers.o -L/usr/local/Cellar/gcc/5.3.0/lib/gcc/5/gcc/x86_64-apple-darwin14.5.0/5.3.0 -L/usr/local/lib -L/usr/local/opt/openssl/lib -L/usr/local/opt/sqlite/lib -lgfortran -o ./_grid.so" failed with exit status 1 +++ Compilation failed

The file crm.F90 converts the output from subroutines radcswmx and radclwmx from CGS to MKS units, e.g. on line 346:

qrl = qrl*1.e-3

Two problems:

1. I don't think this is the right conversion factor. The heating rates have dimensions energy/mass/time or (length)^2 / (time)^3. So to convert from CGS to MKS should be a factor of 1.E-4, not 1.E-3.
2. The heating rates qrl and qrs are actually already converted to MKS units (J/kg/s).
   See line 1041 of radlw.F90 and line 2422 of radsw.F90, where the factor 1.E-4 is applied.

So in the Python module, the heating rates swhr, lwhr and TdotRad seem to be a factor of 1000 too small. Also the Tinc temperature increment for the atmosphere, same problem.

If I'm right, then the fix is very simple -- just removing the lines that apply the incorrect conversion in crm.F90. I did this on my master branch of the CliMT code. Can't figure out how to submit a pull request for just this change without all the other stuff I've done recently.

I think there may be some issue with the Asselin filter, at least if an external time increment is imposed. I'm not sure what exactly is going on, but it may arise from the leapfrog step and filter being called for each component consecutively? (Sorry, I'm not good at python...)

Here is an example script which illustrates the issue. The results here are significantly dependent on the Asselin filter constant - even though a steady state is reached, in which case one would hope that the latter shouldn't do anything. Moreover, the radiative cooling rate does not agree with the externally applied heating rate (which it should in steady state), and there is a pretty clear relationship between the mismatch and the Asselin filter constant (seee comments in the script).

Maybe this is just an issue of understanding what the actual "effective" model time-step is. But it seems a bit funny that the latter should depend on the Asselin filter constant.

Cheers,
Malte

#!/usr/bin/env python

from numpy import *
import climt

### Initialize parameters
kwargs                          = {}
kwargs['lev']           = (arange(climt.get_nlev())+1)*(1000.-0.)/climt.get_nlev() ### Set pressure levels to what was modified above
kwargs['solin']         = 100 ### Solar insolation
kwargs['afc']                   = 0.2 
kwargs['dt']                    = 60.*60. ### Timestep
kwargs['Hslab']                 = 2. 
kwargs['tau_inf']       = 2.0 ### Tau
kwargs['Ts']                    = 250. ### Surface temperature init
kwargs['T']                     = ones((climt.get_nlev()))*250. ### Temperature profile init

NSteps = int( 1000*86400./kwargs['dt'] ) ## Number of steps (1000 days)

# Prescribe vertically constant heating rate of (supposedly) 1K/day
dTa = kwargs['dt']*2.*1./86400.+0.0*kwargs['lev'] 

rad = climt.radiation(scheme = 'grey', **kwargs) ### Radiation component
ocn = climt.ocean() ### ML Ocean
fed = climt.federation(rad,ocn,**kwargs) ### Coupler

### Run model
for i in range(NSteps): 

    # This formulation appears to give an effective heating rate of 1K *(1-afc) per day
    fed.step(Inc={'T':dTa})

    # This formulation appears to give an effective heating rate of 1K *(1-2*afc) per day
    #fed.step()
    #fed['T']=fed['T']+dTa

    print 'Ts: %10.5f' % (fed['Ts']-273.15)

# Print radiative temperature tendency by level - this should be -1K per day in equilibrium     
print(fed['TdotRad'])

In grid.py, shouldn't the dimensions for latitude and longitude be set as
self.value['nlat'] = self.Shape3D[1]
self.value['nlon'] = self.Shape3D[2]
At the moment they all get set to self.Shape3D[0].