#string that the fortran code automatically uses. Not sure how to
#change this at the moment.
#initializing variable to store output file name string
#will probably need to be incremented as model is run
outputfile_iter = 0
outputfile = 'workrslt.{:0>3d}'.format(outputfile_iter)
#need to write out outputfile_iter as a text file required by the
#fortran code
f = open('filenum.dat', 'w')
f.write('{:0>3d}'.format(outputfile_iter))
f.close()
#Evaluate model using fortran code to get synthetic observation
interface.exec_rtmod()
#Read result file into python
syn_obs = observation.Synthetic(filename = outputfile)
#below is testing ... remove when find_best_model.py is finished
initial_csq = compare.calc_chisq(datacuts, syn_obs)
print "Chi-square: {}".format(initial_csq)
#diagnostic:
real = observation.concat(datacuts)
syn = syn_obs.concat(3)
#remove temporary working model files
def calc_deriv(atm_model, synobs):
"""
Calculate derivatives (changes in synthetic observation) produced by changes introduced by adjusting the parameters.
"""
#determine number of parameters
nparams = atm_model.nflags
#create synthetic observation I/F data vector (i.e. 1 numpy array combining
#all spectral channels)
syn_vec = synobs.concat(synobs.nangles)
#initialize derivative array
deriv_array = np.array([])
#create copy of input model for testing
tmp_model = copy.deepcopy(atm_model)
epsilon = 0.01
#parse input atmospheric model for flags, change parameter one at a time
#for each atmospheric layer
for i in range(tmp_model.nlayers):
#first, extract said layer
l = getattr(tmp_model, 'l'+str(i+1))
#parse layer for flags
#determine index locations
flaglocs = [a for a in range(len(l.flaglist)) \
if type(l.flaglist[a]) == str]
#source
#http://stackoverflow.com/questions/7270321/finding-the-index-of-elements-based-on-a-condition-using-python-list-comprehensi
for j in flaglocs:
#when encountering flag:
#determine old value
old_param_val = getattr(l, l.param_names[j])
#adjust parameter by epsilon
#note, this is currently ALWAYS a positive change
#since this is only calculating derivative, not adjusting the model towards
#a best-fit.
l.adj_param_epsilon(l.param_names[j])
#determine change in parameter
new_param_val = getattr(l, l.param_names[j])
diff_param = new_param_val - old_param_val
#replace modified layer in atmosphere object
# i = layer number
# l = new layer object
tmp_model.replace_layer(i+1, l)
#run model with changed parameter
#set up temporary files for output / RT model execution
f = open('filenum.dat', 'w')
f.write('{:0<3d}'.format(999))
#999 = code for tmp/fleeting file in radtran model
f.close()
#write out temp model for RT code
tmp_model.write_ft_input(filename='workmodl999')
#Evaluate model using fortran code to get synthetic observation
interface.exec_rtmod()
#Read result file into python
tmp_syn = observation.Synthetic(filename = 'workrslt.999')
#create tmp synthetic observation concatenated vector
tmp_syn_vec = tmp_syn.concat(synobs.nangles)
#determine change in I/F
diff = (tmp_syn_vec - syn_vec)/(diff_param)
#store in derivative array
if deriv_array.size == 0:
deriv_array = np.append(deriv_array, diff)
else:
deriv_array = np.vstack((deriv_array, diff))
#restore tmp_model to the old state
tmp_model = copy.deepcopy(atm_model)
#delete temporary working model
os.system('rm -f workmodl999')
return deriv_array
