def start_calibration(custom_crop, server, preprocessed_data=None):
#read params to be calibrated
params = []
with open('calibrator/calibratethese.csv') as paramscsv:
dialect = csv.Sniffer().sniff(paramscsv.read(), delimiters=';,\t')
paramscsv.seek(0)
reader = csv.reader(paramscsv, dialect)
next(reader, None) # skip the header
for row in reader:
p = {}
p["name"] = row[0]
p["array"] = row[1]
p["low"] = float(row[2])
p["high"] = float(row[3])
p["stepsize"] = float(row[4])
p["optguess"] = float(row[5])
p["minbound"] = float(row[6])
p["maxbound"] = float(row[7])
if len(row) == 9 and row[8] != "":
p["derive_function"] = make_lambda(row[8])
params.append(p)
spot_setup = spotpy_setup_MONICA.spot_setup(params, custom_crop, server,
preprocessed_data)
rep = 100
sampler = spotpy.algorithms.sceua(spot_setup,
dbname='SCEUA',
dbformat='ram')
#sampler.sample(rep, ngs=len(params)+1, kstop=2*len(params)+1, peps=0.0001, pcento=0.00001)
sampler.sample(rep, ngs=len(params) + 1, kstop=2)
best_params = sampler.status.params
with open('optimizedparams.csv', 'wb') as outcsvfile:
writer = csv.writer(outcsvfile)
for i in range(len(best_params)):
outrow = []
arr_pos = ""
if params[i]["array"].upper() != "FALSE":
arr_pos = params[i]["array"]
outrow.append(params[i]["name"] + arr_pos)
outrow.append(best_params[i])
writer.writerow(outrow)
if len(params) > len(best_params):
reminder = []
reminder.append(
"Don't forget to calculate and set derived params!")
writer.writerow(reminder)
text = 'optimized parameters saved!'
print(text)
next(reader, None) # skip the header
for row in reader:
p = {}
p["name"] = row[0]
p["array"] = row[1]
p["low"] = row[2]
p["high"] = row[3]
p["stepsize"] = row[4]
p["optguess"] = row[5]
p["minbound"] = row[6]
p["maxbound"] = row[7]
if len(row) == 9 and row[8] != "":
p["derive_function"] = make_lambda(row[8])
params.append(p)
spot_setup = spotpy_setup_MONICA.spot_setup(params, exp_maps, obslist)
rep = 10
results = []
sampler = spotpy.algorithms.sceua(spot_setup, dbname='SCEUA', dbformat='ram')
sampler.sample(rep, ngs=3, kstop=10)
#sampler.sample(rep, ngs=3, kstop=50, pcento=0.01, peps=0.05)
results.append(sampler.getdata())
best = sampler.status.params
with open('optimizedparams.csv', 'wb') as outcsvfile:
writer = csv.writer(outcsvfile)
for i in range(len(best)):
outrow = []
arr_pos = ""
"runexps": [1] # 1: Askov, 3: Grignon, 4: Kursk, 5: Rothamsted, 6: Ultuna, 7: Versailles, None: all
},
"muencheberg": {
"map_file": "crop_sim_site_MAP_mue.csv",
"observations": "observations_mue.csv",
"params": "calibratethese.csv",
"runexps": None #[13,25] #None: all
}
}
run = "muencheberg"
exp_maps = helper.read_exp_map(filename=datasets[run]["map_file"], exps=datasets[run]["runexps"])
observations = helper.read_obs(filename=datasets[run]["observations"], exps=datasets[run]["runexps"])
params = helper.read_params(datasets[run]["params"])
spot_setup = spotpy_setup_MONICA.spot_setup(params, exp_maps, observations, config, distrib="normal")
sampler = spotpy.algorithms.demcz(spot_setup, dbname='DEMCz', dbformat='csv')
sampler.sample(1000)
best = sampler.status.params
with open('optimizedparams.csv', 'wb') as outcsvfile:
writer = csv.writer(outcsvfile)
for i in range(len(best)):
outrow=[]
outrow.append(params[i]["name"])
outrow.append(best[i])
writer.writerow(outrow)
text='optimized parameters saved!'
print(text)
]
#customize environment:
for soil_p in soil_params:
set_param(env["params"]["userSoilOrganicParameters"]["DEFAULT"],
soil_p["name"], soil_p["value"])
for cm in env["cropRotations"][0]["cropRotation"]:
for ws in cm["worksteps"]:
if ws["type"] == "OrganicFertilization":
for fert_p in fert_params:
set_param(ws["parameters"], fert_p["name"],
fert_p["value"])
#with open("dumped_" + env_file, "wb") as _:
# _.write(json.dumps(env, indent=4))
spot_setup = spotpy_setup_MONICA.spot_setup(params, obslist, config,
env)
results = []
if config["cal-method"] == "SCE-UA":
sampler = spotpy.algorithms.sceua(spot_setup,
dbname='SCEUA',
dbformat='ram')
sampler.sample(repetitions=config["rep"],
ngs=len(params) + 1,
kstop=10,
pcento=10,
peps=10)
elif config["cal-method"] == "MLE":
sampler = spotpy.algorithms.mle(spot_setup,
dbname='MLE_CMF',
dbformat='ram')