def test_optimization_model(self):
num_cells = 20
model_type = pt_gs_k.PTGSKModel
opt_model_type = pt_gs_k.PTGSKOptModel
model = self.build_model(model_type, pt_gs_k.PTGSKParameter, num_cells)
cal = api.Calendar()
t0 = cal.time(2015, 1, 1, 0, 0, 0)
dt = api.deltahours(1)
n = 240
time_axis = api.TimeAxisFixedDeltaT(t0, dt, n)
model_interpolation_parameter = api.InterpolationParameter()
model.initialize_cell_environment(
time_axis
) # just show how we can split the run_interpolation into two calls(second one optional)
model.interpolate(
model_interpolation_parameter,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_snow_sca_swe_collection(-1, True)
model.set_states(
s0
) # at this point the intial state of model is established as well
model.run_cells()
cids = api.IntVector.from_numpy(
[1]) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
sum_discharge_value = model.statistics.discharge_value(
cids, 0) # at the first timestep
self.assertGreaterEqual(sum_discharge_value, 130.0)
# verify we can construct an optimizer
opt_model = model.create_opt_model_clone()
opt_model.set_snow_sca_swe_collection(
-1, True) # ensure to fill in swe/sca
opt_model.run_cells()
opt_sum_discharge = opt_model.statistics.discharge(cids)
opt_swe = opt_model.statistics.snow_swe(cids) # how to get out swe/sca
opt_swe_v = opt_model.statistics.snow_swe_value(cids, 0)
opt_sca = opt_model.statistics.snow_sca(cids)
for i in range(len(opt_model.time_axis)):
opt_sca_v = opt_model.statistics.snow_sca_value(cids, i)
self.assertTrue(0.0 <= opt_sca_v <= 1.0)
self.assertIsNotNone(opt_sum_discharge)
self.assertIsNotNone(opt_swe)
self.assertIsNotNone(opt_sca)
optimizer = opt_model_type.optimizer_t(
opt_model
) # notice that a model type know it's optimizer type, e.g. PTGSKOptimizer
self.assertIsNotNone(optimizer)
#
# create target specification
#
opt_model.revert_to_initial_state(
) # set_states(s0) # remember to set the s0 again, so we have the same initial condition for our game
tsa = api.TsTransform().to_average(t0, dt, n, sum_discharge)
t_spec_1 = api.TargetSpecificationPts(tsa, cids, 1.0, api.KLING_GUPTA,
1.0, 0.0, 0.0, api.DISCHARGE,
'test_uid')
target_spec = api.TargetSpecificationVector()
target_spec.append(t_spec_1)
upper_bound = model_type.parameter_t(model.get_region_parameter(
)) # the model_type know it's parameter_t
lower_bound = model_type.parameter_t(model.get_region_parameter())
upper_bound.kirchner.c1 = -1.9
lower_bound.kirchner.c1 = -3.0
upper_bound.kirchner.c2 = 0.99
lower_bound.kirchner.c2 = 0.80
optimizer.set_target_specification(target_spec, lower_bound,
upper_bound)
# Not needed, it will automatically get one.
# optimizer.establish_initial_state_from_model()
# s0_0 = optimizer.get_initial_state(0)
# optimizer.set_verbose_level(1000)
p0 = model_type.parameter_t(model.get_region_parameter())
orig_c1 = p0.kirchner.c1
orig_c2 = p0.kirchner.c2
# model.get_cells()[0].env_ts.precipitation.set(0, 5.1)
# model.get_cells()[0].env_ts.precipitation.set(1, 4.9)
goal_f0 = optimizer.calculate_goal_function(p0)
p0.kirchner.c1 = -2.4
p0.kirchner.c2 = 0.91
opt_param = optimizer.optimize(p0, 1500, 0.1, 1e-5)
goal_fx = optimizer.calculate_goal_function(opt_param)
p0.kirchner.c1 = -2.4
p0.kirchner.c2 = 0.91
# goal_fx1 = optimizer.calculate_goal_function(p0)
self.assertLessEqual(goal_fx, 10.0)
self.assertAlmostEqual(orig_c1, opt_param.kirchner.c1, 4)
self.assertAlmostEqual(orig_c2, opt_param.kirchner.c2, 4)
# verify the interface to the new optimize_global function
global_opt_param = optimizer.optimize_global(p0,
max_n_evaluations=1500,
max_seconds=3.0,
solver_eps=1e-5)
self.assertIsNotNone(
global_opt_param
) # just to ensure signature and results are covered
def test_model_initialize_and_run(self):
num_cells = 20
model_type = pt_gs_k.PTGSKModel
model = self.build_model(model_type,
pt_gs_k.PTGSKParameter,
num_cells,
num_catchments=1)
self.assertEqual(model.size(), num_cells)
self.verify_state_handler(model)
# demo of feature for threads
self.assertGreaterEqual(model.ncore,
1) # defaults to hardware concurrency
model.ncore = 4 # set it to 4, and
self.assertEqual(model.ncore, 4) # verify it works
# now modify snow_cv forest_factor to 0.1
region_parameter = model.get_region_parameter()
region_parameter.gs.snow_cv_forest_factor = 0.1
region_parameter.gs.snow_cv_altitude_factor = 0.0001
self.assertEqual(region_parameter.gs.snow_cv_forest_factor, 0.1)
self.assertEqual(region_parameter.gs.snow_cv_altitude_factor, 0.0001)
self.assertAlmostEqual(region_parameter.gs.effective_snow_cv(1.0, 0.0),
region_parameter.gs.snow_cv + 0.1)
self.assertAlmostEqual(
region_parameter.gs.effective_snow_cv(1.0, 1000.0),
region_parameter.gs.snow_cv + 0.1 + 0.1)
cal = api.Calendar()
time_axis = api.TimeAxisFixedDeltaT(cal.time(2015, 1, 1, 0, 0, 0),
api.deltahours(1), 240)
model_interpolation_parameter = api.InterpolationParameter()
# degC/m, so -0.5 degC/100m
model_interpolation_parameter.temperature_idw.default_temp_gradient = -0.005
# if possible use closest neighbor points and solve gradient using equation,(otherwise default min/max height)
model_interpolation_parameter.temperature_idw.gradient_by_equation = True
# Max number of temperature sources used for one interpolation
model_interpolation_parameter.temperature_idw.max_members = 6
# 20 km is max distance
model_interpolation_parameter.temperature_idw.max_distance = 20000
# zscale is used to discriminate neighbors at different elevation than target point
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 1.0)
model_interpolation_parameter.temperature_idw.zscale = 0.5
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 0.5)
# Pure linear interpolation
model_interpolation_parameter.temperature_idw.distance_measure_factor = 1.0
# This enables IDW with default temperature gradient.
model_interpolation_parameter.use_idw_for_temperature = True
self.assertAlmostEqual(
model_interpolation_parameter.precipitation.scale_factor,
1.02) # just verify this one is as before change to scale_factor
model.initialize_cell_environment(
time_axis
) # just show how we can split the run_interpolation into two calls(second one optional)
model.interpolate(
model_interpolation_parameter,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
m_ip_parameter = model.interpolation_parameter # illustrate that we can get back the passed interpolation parameter as a property of the model
self.assertEqual(
m_ip_parameter.use_idw_for_temperature,
True) # just to ensure we really did get back what we passed in
self.assertAlmostEqual(m_ip_parameter.temperature_idw.zscale, 0.5)
#
# Section to demo that we can ensure that model.cells[].env_ts.xxx
# have finite-values only
#
for env_ts_x in [
model.cells[0].env_ts.temperature,
model.cells[0].env_ts.precipitation,
model.cells[0].env_ts.rel_hum, model.cells[0].env_ts.radiation,
model.cells[0].env_ts.wind_speed
]:
self.assertTrue(
model.is_cell_env_ts_ok()
) # demon how to verify that cell.env_ts can be checked prior to run
vx = env_ts_x.value(0) # save value
env_ts_x.set(0, float('nan')) # insert a nan
self.assertFalse(model.is_cell_env_ts_ok()
) # demo it returns false if nan @cell.env_ts
env_ts_x.set(0, vx) # insert back original value
self.assertTrue(model.is_cell_env_ts_ok()) # ready to run again
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_states(s0)
model.set_state_collection(
-1, True) # enable state collection for all cells
model2 = model_type(
model
) # make a copy, so that we in the stepwise run below get a clean copy with all values zero.
opt_model = pt_gs_k.create_opt_model_clone(
model) # this is how to make a model suitable for optimizer
model.run_cells(
) # the default arguments applies: thread_cell_count=0,start_step=0,n_steps=0)
cids = api.IntVector(
) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
sum_discharge_value = model.statistics.discharge_value(
cids, 0) # at the first timestep
sum_charge = model.statistics.charge(cids)
sum_charge_value = model.statistics.charge_value(cids, 0)
self.assertAlmostEqual(sum_charge_value, -110.6998, places=2)
self.assertAlmostEqual(sum_charge.values[0],
sum_charge_value,
places=5)
cell_charge = model.statistics.charge_value(
api.IntVector([0, 1, 3]), 0, ix_type=api.stat_scope.cell)
self.assertAlmostEqual(cell_charge, -16.7138, places=2)
charge_sum_1_2_6 = model.statistics.charge(
api.IntVector([1, 2, 6]),
ix_type=api.stat_scope.cell).values.to_numpy().sum()
self.assertAlmostEqual(charge_sum_1_2_6, 107.3981, places=2)
ae_output = model.actual_evaptranspiration_response.output(cids)
ae_pot_ratio = model.actual_evaptranspiration_response.pot_ratio(cids)
self.assertIsNotNone(ae_output)
self.assertAlmostEqual(ae_output.values.to_numpy().max(),
0.189214067680088)
self.assertIsNotNone(ae_pot_ratio)
self.assertAlmostEqual(ae_pot_ratio.values.to_numpy().min(),
0.9995599424191931)
self.assertAlmostEqual(ae_pot_ratio.values.to_numpy().max(), 1.0)
opt_model.run_cells(
) # starting out with the same state, same interpolated values, and region-parameters, we should get same results
sum_discharge_opt_value = opt_model.statistics.discharge_value(cids, 0)
self.assertAlmostEqual(
sum_discharge_opt_value, sum_discharge_value,
3) # verify the opt_model clone gives same value
self.assertGreaterEqual(sum_discharge_value, 130.0)
opt_model.region_env.temperature[0].ts.set(
0, 23.2
) # verify that region-env is different (no aliasing, a true copy is required)
self.assertFalse(
abs(model.region_env.temperature[0].ts.value(0) -
opt_model.region_env.temperature[0].ts.value(0)) > 0.5)
#
# check values
#
self.assertIsNotNone(sum_discharge)
# now, re-run the process in 24-hours steps x 10
model.set_states(s0) # restore state s0
self.assertEqual(s0.size(), model.initial_state.size())
for do_collect_state in [False, True]:
model2.set_state_collection(
-1, do_collect_state
) # issue reported by Yisak, prior to 21.3, this would crash
model2.set_states(s0)
# now after fix, it works Ok
for section in range(10):
model2.run_cells(use_ncore=0,
start_step=section * 24,
n_steps=24)
section_discharge = model2.statistics.discharge(cids)
self.assertEqual(section_discharge.size(), sum_discharge.size(
)) # notice here that the values after current step are 0.0
stepwise_sum_discharge = model2.statistics.discharge(cids)
# assert stepwise_sum_discharge == sum_discharge
diff_ts = sum_discharge.values.to_numpy(
) - stepwise_sum_discharge.values.to_numpy()
self.assertAlmostEqual((diff_ts * diff_ts).max(), 0.0, 4)
# Verify that if we pass in illegal cids, then it raises exception(with first failing
try:
illegal_cids = api.IntVector([0, 4, 5])
model.statistics.discharge(illegal_cids)
self.assertFalse(
True, "Failed test, using illegal cids should raise exception")
except RuntimeError as rte:
pass
avg_temperature = model.statistics.temperature(cids)
avg_precipitation = model.statistics.precipitation(cids)
self.assertIsNotNone(avg_precipitation)
for time_step in range(time_axis.size()):
precip_raster = model.statistics.precipitation(
cids, time_step) # example raster output
self.assertEqual(precip_raster.size(), num_cells)
# example single value spatial aggregation (area-weighted) over cids for a specific timestep
avg_gs_sc_value = model.gamma_snow_response.sca_value(cids, 1)
self.assertGreaterEqual(avg_gs_sc_value, 0.0)
avg_gs_sca = model.gamma_snow_response.sca(cids) # swe output
self.assertIsNotNone(avg_gs_sca)
# lwc surface_heat alpha melt_mean melt iso_pot_energy temp_sw
avg_gs_albedo = model.gamma_snow_state.albedo(cids)
self.assertIsNotNone(avg_gs_albedo)
self.assertEqual(avg_temperature.size(), time_axis.size(),
"expect results equal to time-axis size")
copy_region_model = model.__class__(model)
self.assertIsNotNone(copy_region_model)
copy_region_model.run_cells(
) # just to verify we can copy and run the new model
#
# Play with routing and river-network
#
# 1st: add a river, with 36.000 meter hydro length, a UHGParameter with 1m/hour speed, alpha/beta suitable
model.river_network.add(
api.River(1, api.RoutingInfo(0, 3000.0),
api.UHGParameter(1 / 3.60, 7.0, 0.0))) # river id =1
# 2nd: let cells route to the river
model.connect_catchment_to_river(
1, 1) # now all cells in catchment 1 routes to river with id 1.
self.assertTrue(model.has_routing())
# 3rd: now we can have a look at water coming in and out
river_out_m3s = model.river_output_flow_m3s(
1) # should be delayed and reshaped
river_local_m3s = model.river_local_inflow_m3s(
1
) # should be equal to cell outputs (no routing stuff from cell to river)
river_upstream_inflow_m3s = model.river_upstream_inflow_m3s(
1
) # should be 0.0 in this case, since we do not have a routing network
self.assertIsNotNone(river_out_m3s)
self.assertAlmostEqual(river_out_m3s.value(8), 28.061248025828114, 0)
self.assertIsNotNone(river_local_m3s)
self.assertIsNotNone(river_upstream_inflow_m3s)
model.connect_catchment_to_river(1, 0)
self.assertFalse(model.has_routing())
#
# Test the state-adjustments interfaces
#
q_0 = model.cells[0].state.kirchner.q
model.adjust_q(2.0, cids)
q_1 = model.cells[0].state.kirchner.q
self.assertAlmostEqual(q_0 * 2.0, q_1)
model.revert_to_initial_state() # ensure we have a known state
model.run_cells(0, 10, 2) # just run step 10 and 11
q_avg = (model.statistics.discharge_value(cids, 10) +
model.statistics.discharge_value(cids, 11)
) / 2.0 # get out the discharge for step 10 and 11
x = 0.7 # we want x*q_avg as target
model.revert_to_initial_state(
) # important, need start state for the test here
adjust_result = model.adjust_state_to_target_flow(
x * q_avg,
cids,
start_step=10,
scale_range=3.0,
scale_eps=1e-3,
max_iter=350,
n_steps=2
) # This is how to adjust state to observed average flow for cids for tstep 10
self.assertEqual(len(adjust_result.diagnostics),
0) # diag should be len(0) if ok.
self.assertAlmostEqual(adjust_result.q_r, q_avg * x,
2) # verify we reached target
self.assertAlmostEqual(adjust_result.q_0, q_avg, 2) # .q_0,
# now verify what happens if we put in bad values for observed value
adjust_result = model.adjust_state_to_target_flow(float('nan'),
cids,
start_step=10,
scale_range=3.0,
scale_eps=1e-3,
max_iter=300,
n_steps=2)
assert len(adjust_result.diagnostics
) > 0, 'expect diagnostics length be larger than 0'
# then verify what happens if we put in bad values on simulated result
model.cells[0].env_ts.temperature.set(10, float('nan'))
adjust_result = model.adjust_state_to_target_flow(30.0,
cids,
start_step=10,
scale_range=3.0,
scale_eps=1e-3,
max_iter=300,
n_steps=2)
assert len(adjust_result.diagnostics
) > 0, 'expect diagnostics length be larger than 0'
def test_model_initialize_and_run(self):
num_cells = 20
model_type = pt_gs_k.PTGSKModel
model = self.build_model(model_type, pt_gs_k.PTGSKParameter, num_cells)
self.assertEqual(model.size(), num_cells)
# now modify snow_cv forest_factor to 0.1
region_parameter = model.get_region_parameter()
region_parameter.gs.snow_cv_forest_factor = 0.1
region_parameter.gs.snow_cv_altitude_factor = 0.0001
self.assertEqual(region_parameter.gs.snow_cv_forest_factor, 0.1)
self.assertEqual(region_parameter.gs.snow_cv_altitude_factor, 0.0001)
self.assertAlmostEqual(region_parameter.gs.effective_snow_cv(1.0, 0.0),
region_parameter.gs.snow_cv + 0.1)
self.assertAlmostEqual(
region_parameter.gs.effective_snow_cv(1.0, 1000.0),
region_parameter.gs.snow_cv + 0.1 + 0.1)
cal = api.Calendar()
time_axis = api.Timeaxis(cal.time(api.YMDhms(2015, 1, 1, 0, 0, 0)),
api.deltahours(1), 240)
model_interpolation_parameter = api.InterpolationParameter()
# degC/m, so -0.5 degC/100m
model_interpolation_parameter.temperature_idw.default_temp_gradient = -0.005
# if possible use closest neighbor points and solve gradient using equation,(otherwise default min/max height)
model_interpolation_parameter.temperature_idw.gradient_by_equation = True
# Max number of temperature sources used for one interpolation
model_interpolation_parameter.temperature_idw.max_members = 6
# 20 km is max distance
model_interpolation_parameter.temperature_idw.max_distance = 20000
# Pure linear interpolation
model_interpolation_parameter.temperature_idw.distance_measure_factor = 1.0
# This enables IDW with default temperature gradient.
model_interpolation_parameter.use_idw_for_temperature = True
self.assertAlmostEqual(
model_interpolation_parameter.precipitation.scale_factor,
1.02) # just verify this one is as before change to scale_factor
model.run_interpolation(
model_interpolation_parameter, time_axis,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_states(s0)
model.set_state_collection(
-1, True) # enable state collection for all cells
model.run_cells()
cids = api.IntVector(
) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
self.assertIsNotNone(sum_discharge)
avg_temperature = model.statistics.temperature(cids)
avg_precipitation = model.statistics.precipitation(cids)
self.assertIsNotNone(avg_precipitation)
for time_step in range(time_axis.size()):
precip_raster = model.statistics.precipitation(
cids, time_step) # example raster output
self.assertEqual(precip_raster.size(), num_cells)
avg_gs_sca = model.gamma_snow_response.sca(cids) # swe output
self.assertIsNotNone(avg_gs_sca)
# lwc surface_heat alpha melt_mean melt iso_pot_energy temp_sw
avg_gs_albedo = model.gamma_snow_state.albedo(cids)
self.assertIsNotNone(avg_gs_albedo)
self.assertEqual(avg_temperature.size(), time_axis.size(),
"expect results equal to time-axis size")
copy_region_model = model.__class__(model)
self.assertIsNotNone(copy_region_model)
copy_region_model.run_cells(
) #just to verify we can copy and run the new model
def test_model_initialize_and_run(self):
num_cells = 20
model_type = pt_gs_k.PTGSKModel
model = self.build_model(model_type, pt_gs_k.PTGSKParameter, num_cells)
self.assertEqual(model.size(), num_cells)
self.verify_state_handler(model)
# demo of feature for threads
self.assertGreaterEqual(model.ncore,
1) # defaults to hardware concurrency
model.ncore = 4 # set it to 4, and
self.assertEqual(model.ncore, 4) # verify it works
# now modify snow_cv forest_factor to 0.1
region_parameter = model.get_region_parameter()
region_parameter.gs.snow_cv_forest_factor = 0.1
region_parameter.gs.snow_cv_altitude_factor = 0.0001
self.assertEqual(region_parameter.gs.snow_cv_forest_factor, 0.1)
self.assertEqual(region_parameter.gs.snow_cv_altitude_factor, 0.0001)
self.assertAlmostEqual(region_parameter.gs.effective_snow_cv(1.0, 0.0),
region_parameter.gs.snow_cv + 0.1)
self.assertAlmostEqual(
region_parameter.gs.effective_snow_cv(1.0, 1000.0),
region_parameter.gs.snow_cv + 0.1 + 0.1)
cal = api.Calendar()
time_axis = api.TimeAxisFixedDeltaT(cal.time(2015, 1, 1, 0, 0, 0),
api.deltahours(1), 240)
model_interpolation_parameter = api.InterpolationParameter()
# degC/m, so -0.5 degC/100m
model_interpolation_parameter.temperature_idw.default_temp_gradient = -0.005
# if possible use closest neighbor points and solve gradient using equation,(otherwise default min/max height)
model_interpolation_parameter.temperature_idw.gradient_by_equation = True
# Max number of temperature sources used for one interpolation
model_interpolation_parameter.temperature_idw.max_members = 6
# 20 km is max distance
model_interpolation_parameter.temperature_idw.max_distance = 20000
# zscale is used to discriminate neighbors at different elevation than target point
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 1.0)
model_interpolation_parameter.temperature_idw.zscale = 0.5
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 0.5)
# Pure linear interpolation
model_interpolation_parameter.temperature_idw.distance_measure_factor = 1.0
# This enables IDW with default temperature gradient.
model_interpolation_parameter.use_idw_for_temperature = True
self.assertAlmostEqual(
model_interpolation_parameter.precipitation.scale_factor,
1.02) # just verify this one is as before change to scale_factor
model.initialize_cell_environment(
time_axis
) # just show how we can split the run_interpolation into two calls(second one optional)
model.interpolate(
model_interpolation_parameter,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
m_ip_parameter = model.interpolation_parameter # illustrate that we can get back the passed interpolation parameter as a property of the model
self.assertEqual(
m_ip_parameter.use_idw_for_temperature,
True) # just to ensure we really did get back what we passed in
self.assertAlmostEqual(m_ip_parameter.temperature_idw.zscale, 0.5)
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_states(s0)
model.set_state_collection(
-1, True) # enable state collection for all cells
model2 = model_type(
model
) # make a copy, so that we in the stepwise run below get a clean copy with all values zero.
opt_model = pt_gs_k.create_opt_model_clone(
model) # this is how to make a model suitable for optimizer
model.run_cells(
) # the default arguments applies: thread_cell_count=0,start_step=0,n_steps=0)
cids = api.IntVector(
) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
sum_discharge_value = model.statistics.discharge_value(
cids, 0) # at the first timestep
sum_charge = model.statistics.charge(cids)
sum_charge_value = model.statistics.charge_value(cids, 0)
ae_output = model.actual_evaptranspiration_response.output(cids)
ae_pot_ratio = model.actual_evaptranspiration_response.pot_ratio(cids)
self.assertIsNotNone(ae_output)
self.assertAlmostEqual(ae_output.values.to_numpy().max(),
0.189214067680088)
self.assertIsNotNone(ae_pot_ratio)
self.assertAlmostEqual(ae_pot_ratio.values.to_numpy().min(),
0.9999330003895371)
self.assertAlmostEqual(ae_pot_ratio.values.to_numpy().max(), 1.0)
opt_model.run_cells(
) # starting out with the same state, same interpolated values, and region-parameters, we should get same results
sum_discharge_opt_value = opt_model.statistics.discharge_value(cids, 0)
self.assertAlmostEqual(
sum_discharge_opt_value, sum_discharge_value,
3) # verify the opt_model clone gives same value
self.assertGreaterEqual(sum_discharge_value, 130.0)
opt_model.region_env.temperature[0].ts.set(
0, 23.2
) # verify that region-env is different (no aliasing, a true copy is required)
self.assertFalse(
abs(model.region_env.temperature[0].ts.value(0) -
opt_model.region_env.temperature[0].ts.value(0)) > 0.5)
#
# check values
#
self.assertIsNotNone(sum_discharge)
# now, re-run the process in 24-hours steps x 10
model.set_states(s0) # restore state s0
self.assertEqual(s0.size(), model.initial_state.size())
for do_collect_state in [False, True]:
model2.set_state_collection(
-1, do_collect_state
) # issue reported by Yisak, prior to 21.3, this would crash
model2.set_states(s0)
# now after fix, it works Ok
for section in range(10):
model2.run_cells(use_ncore=0,
start_step=section * 24,
n_steps=24)
section_discharge = model2.statistics.discharge(cids)
self.assertEqual(section_discharge.size(), sum_discharge.size(
)) # notice here that the values after current step are 0.0
stepwise_sum_discharge = model2.statistics.discharge(cids)
# assert stepwise_sum_discharge == sum_discharge
diff_ts = sum_discharge.values.to_numpy(
) - stepwise_sum_discharge.values.to_numpy()
self.assertAlmostEqual((diff_ts * diff_ts).max(), 0.0, 4)
# Verify that if we pass in illegal cids, then it raises exception(with first failing
try:
illegal_cids = api.IntVector([0, 4, 5])
model.statistics.discharge(illegal_cids)
self.assertFalse(
True, "Failed test, using illegal cids should raise exception")
except RuntimeError as rte:
pass
avg_temperature = model.statistics.temperature(cids)
avg_precipitation = model.statistics.precipitation(cids)
self.assertIsNotNone(avg_precipitation)
for time_step in range(time_axis.size()):
precip_raster = model.statistics.precipitation(
cids, time_step) # example raster output
self.assertEqual(precip_raster.size(), num_cells)
# example single value spatial aggregation (area-weighted) over cids for a specific timestep
avg_gs_sc_value = model.gamma_snow_response.sca_value(cids, 1)
self.assertGreaterEqual(avg_gs_sc_value, 0.0)
avg_gs_sca = model.gamma_snow_response.sca(cids) # swe output
self.assertIsNotNone(avg_gs_sca)
# lwc surface_heat alpha melt_mean melt iso_pot_energy temp_sw
avg_gs_albedo = model.gamma_snow_state.albedo(cids)
self.assertIsNotNone(avg_gs_albedo)
self.assertEqual(avg_temperature.size(), time_axis.size(),
"expect results equal to time-axis size")
copy_region_model = model.__class__(model)
self.assertIsNotNone(copy_region_model)
copy_region_model.run_cells(
) # just to verify we can copy and run the new model
#
# Play with routing and river-network
#
# 1st: add a river, with 36.000 meter hydro length, a UHGParameter with 1m/hour speed, alpha/beta suitable
model.river_network.add(
api.River(1, api.RoutingInfo(0, 3000.0),
api.UHGParameter(1 / 3.60, 7.0, 0.0))) # river id =1
# 2nd: let cells route to the river
model.connect_catchment_to_river(
0, 1) # now all cells in catchment 0 routes to river with id 1.
self.assertTrue(model.has_routing())
# 3rd: now we can have a look at water coming in and out
river_out_m3s = model.river_output_flow_m3s(
1) # should be delayed and reshaped
river_local_m3s = model.river_local_inflow_m3s(
1
) # should be equal to cell outputs (no routing stuff from cell to river)
river_upstream_inflow_m3s = model.river_upstream_inflow_m3s(
1
) # should be 0.0 in this case, since we do not have a routing network
self.assertIsNotNone(river_out_m3s)
self.assertAlmostEqual(river_out_m3s.value(8), 31.57297, 0)
self.assertIsNotNone(river_local_m3s)
self.assertIsNotNone(river_upstream_inflow_m3s)
model.connect_catchment_to_river(0, 0)
self.assertFalse(model.has_routing())
#
# Test the state-adjustments interfaces
#
q_0 = model.cells[0].state.kirchner.q
model.adjust_q(2.0, cids)
q_1 = model.cells[0].state.kirchner.q
self.assertAlmostEqual(q_0 * 2.0, q_1)
model.revert_to_initial_state() # ensure we have a known state
model.run_cells(0, 10, 1) # just run step 10
q_avg = model.statistics.discharge_value(
cids, 10) # get out the discharge for step 10
x = 0.7 # we want x*q_avg as target
model.revert_to_initial_state(
) # important, need start state for the test here
adjust_result = model.adjust_state_to_target_flow(
x * q_avg,
cids,
start_step=10,
scale_range=3.0,
scale_eps=1e-3,
max_iter=350
) # This is how to adjust state to observed average flow for cids for tstep 10
self.assertEqual(len(adjust_result.diagnostics),
0) # diag should be len(0) if ok.
self.assertAlmostEqual(adjust_result.q_r, q_avg * x,
3) # verify we reached target
self.assertAlmostEqual(adjust_result.q_0, q_avg, 3) # .q_0,
def test_optimization_model(self):
num_cells = 20
model_type = pt_gs_k.PTGSKOptModel
model = self.build_model(model_type, pt_gs_k.PTGSKParameter,
num_cells)
cal = api.Calendar()
t0 = cal.time(2015, 1, 1, 0, 0, 0)
dt = api.deltahours(1)
n = 240
time_axis = api.TimeAxisFixedDeltaT(t0, dt, n)
model_interpolation_parameter = api.InterpolationParameter()
model.initialize_cell_environment(
time_axis
) # just show how we can split the run_interpolation into two calls(second one optional)
model.interpolate(
model_interpolation_parameter,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_states(
s0
) # at this point the intial state of model is established as well
model.run_cells()
cids = api.IntVector.from_numpy(
[0]) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
sum_discharge_value = model.statistics.discharge_value(
cids, 0) # at the first timestep
self.assertGreaterEqual(sum_discharge_value, 130.0)
# verify we can construct an optimizer
optimizer = model_type.optimizer_t(
model
) # notice that a model type know it's optimizer type, e.g. PTGSKOptimizer
self.assertIsNotNone(optimizer)
#
# create target specification
#
model.revert_to_initial_state(
) # set_states(s0) # remember to set the s0 again, so we have the same initial condition for our game
tsa = api.TsTransform().to_average(t0, dt, n, sum_discharge)
t_spec_1 = api.TargetSpecificationPts(tsa, cids, 1.0,
api.KLING_GUPTA, 1.0, 0.0,
0.0, api.DISCHARGE,
'test_uid')
target_spec = api.TargetSpecificationVector()
target_spec.append(t_spec_1)
upper_bound = model_type.parameter_t(model.get_region_parameter(
)) # the model_type know it's parameter_t
lower_bound = model_type.parameter_t(model.get_region_parameter())
upper_bound.kirchner.c1 = -1.9
lower_bound.kirchner.c1 = -3.0
upper_bound.kirchner.c2 = 0.99
lower_bound.kirchner.c2 = 0.80
optimizer.set_target_specification(target_spec, lower_bound,
upper_bound)
# Not needed, it will automatically get one.
# optimizer.establish_initial_state_from_model()
# s0_0 = optimizer.get_initial_state(0)
# optimizer.set_verbose_level(1000)
p0 = model_type.parameter_t(model.get_region_parameter())
orig_c1 = p0.kirchner.c1
orig_c2 = p0.kirchner.c2
# model.get_cells()[0].env_ts.precipitation.set(0, 5.1)
# model.get_cells()[0].env_ts.precipitation.set(1, 4.9)
p0.kirchner.c1 = -2.4
p0.kirchner.c2 = 0.91
opt_param = optimizer.optimize(p0, 1500, 0.1, 1e-5)
goal_fx = optimizer.calculate_goal_function(opt_param)
p0.kirchner.c1 = -2.4
p0.kirchner.c2 = 0.91
# goal_fx1 = optimizer.calculate_goal_function(p0)
self.assertLessEqual(goal_fx, 10.0)
self.assertAlmostEqual(orig_c1, opt_param.kirchner.c1, 4)
self.assertAlmostEqual(orig_c2, opt_param.kirchner.c2, 4)
def test_model_initialize_and_run(self):
num_cells = 20
model_type = pt_gs_k.PTGSKModel
model = self.build_model(model_type, pt_gs_k.PTGSKParameter, num_cells)
self.assertEqual(model.size(), num_cells)
self.verify_state_handler(model)
# demo of feature for threads
self.assertGreaterEqual(model.ncore,
1) # defaults to hardware concurrency
model.ncore = 4 # set it to 4, and
self.assertEqual(model.ncore, 4) # verify it works
# now modify snow_cv forest_factor to 0.1
region_parameter = model.get_region_parameter()
region_parameter.gs.snow_cv_forest_factor = 0.1
region_parameter.gs.snow_cv_altitude_factor = 0.0001
self.assertEqual(region_parameter.gs.snow_cv_forest_factor, 0.1)
self.assertEqual(region_parameter.gs.snow_cv_altitude_factor, 0.0001)
self.assertAlmostEqual(region_parameter.gs.effective_snow_cv(1.0, 0.0),
region_parameter.gs.snow_cv + 0.1)
self.assertAlmostEqual(
region_parameter.gs.effective_snow_cv(1.0, 1000.0),
region_parameter.gs.snow_cv + 0.1 + 0.1)
cal = api.Calendar()
time_axis = api.TimeAxisFixedDeltaT(cal.time(2015, 1, 1, 0, 0, 0),
api.deltahours(1), 240)
model_interpolation_parameter = api.InterpolationParameter()
# degC/m, so -0.5 degC/100m
model_interpolation_parameter.temperature_idw.default_temp_gradient = -0.005
# if possible use closest neighbor points and solve gradient using equation,(otherwise default min/max height)
model_interpolation_parameter.temperature_idw.gradient_by_equation = True
# Max number of temperature sources used for one interpolation
model_interpolation_parameter.temperature_idw.max_members = 6
# 20 km is max distance
model_interpolation_parameter.temperature_idw.max_distance = 20000
# zscale is used to discriminate neighbors at different elevation than target point
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 1.0)
model_interpolation_parameter.temperature_idw.zscale = 0.5
self.assertAlmostEqual(
model_interpolation_parameter.temperature_idw.zscale, 0.5)
# Pure linear interpolation
model_interpolation_parameter.temperature_idw.distance_measure_factor = 1.0
# This enables IDW with default temperature gradient.
model_interpolation_parameter.use_idw_for_temperature = True
self.assertAlmostEqual(
model_interpolation_parameter.precipitation.scale_factor,
1.02) # just verify this one is as before change to scale_factor
model.initialize_cell_environment(
time_axis
) # just show how we can split the run_interpolation into two calls(second one optional)
model.interpolate(
model_interpolation_parameter,
self.create_dummy_region_environment(
time_axis,
model.get_cells()[int(num_cells / 2)].geo.mid_point()))
m_ip_parameter = model.interpolation_parameter # illustrate that we can get back the passed interpolation parameter as a property of the model
self.assertEqual(
m_ip_parameter.use_idw_for_temperature,
True) # just to ensure we really did get back what we passed in
self.assertAlmostEqual(m_ip_parameter.temperature_idw.zscale, 0.5)
s0 = pt_gs_k.PTGSKStateVector()
for i in range(num_cells):
si = pt_gs_k.PTGSKState()
si.kirchner.q = 40.0
s0.append(si)
model.set_states(s0)
model.set_state_collection(
-1, True) # enable state collection for all cells
model2 = model_type(
model
) # make a copy, so that we in the stepwise run below get a clean copy with all values zero.
opt_model = pt_gs_k.create_opt_model_clone(
model) # this is how to make a model suitable for optimizer
model.run_cells(
) # the default arguments applies: thread_cell_count=0,start_step=0,n_steps=0)
cids = api.IntVector(
) # optional, we can add selective catchment_ids here
sum_discharge = model.statistics.discharge(cids)
sum_discharge_value = model.statistics.discharge_value(
cids, 0) # at the first timestep
sum_charge = model.statistics.charge(cids)
sum_charge_value = model.statistics.charge_value(cids, 0)
opt_model.run_cells(
) # starting out with the same state, same interpolated values, and region-parameters, we should get same results
sum_discharge_opt_value = opt_model.statistics.discharge_value(cids, 0)
self.assertAlmostEqual(
sum_discharge_opt_value, sum_discharge_value,
3) # verify the opt_model clone gives same value
self.assertGreaterEqual(sum_discharge_value, 130.0)
opt_model.region_env.temperature[0].ts.set(
0, 23.2
) # verify that region-env is different (no aliasing, a true copy is required)
self.assertFalse(
abs(model.region_env.temperature[0].ts.value(0) -
opt_model.region_env.temperature[0].ts.value(0)) > 0.5)
#
# check values
#
self.assertIsNotNone(sum_discharge)
# now, re-run the process in 24-hours steps x 10
model.set_states(s0) # restore state s0
self.assertEqual(s0.size(), model.initial_state.size())
for do_collect_state in [False, True]:
model2.set_state_collection(
-1, do_collect_state
) # issue reported by Yisak, prior to 21.3, this would crash
model2.set_states(s0)
# now after fix, it works Ok
for section in range(10):
model2.run_cells(use_ncore=0,
start_step=section * 24,
n_steps=24)
section_discharge = model2.statistics.discharge(cids)
self.assertEqual(section_discharge.size(), sum_discharge.size(
)) # notice here that the values after current step are 0.0
stepwise_sum_discharge = model2.statistics.discharge(cids)
# assert stepwise_sum_discharge == sum_discharge
diff_ts = sum_discharge.values.to_numpy(
) - stepwise_sum_discharge.values.to_numpy()
self.assertAlmostEqual((diff_ts * diff_ts).max(), 0.0, 4)
# Verify that if we pass in illegal cids, then it raises exception(with first failing
try:
illegal_cids = api.IntVector([0, 4, 5])
model.statistics.discharge(illegal_cids)
self.assertFalse(
True, "Failed test, using illegal cids should raise exception")
except RuntimeError as rte:
pass
avg_temperature = model.statistics.temperature(cids)
avg_precipitation = model.statistics.precipitation(cids)
self.assertIsNotNone(avg_precipitation)
for time_step in range(time_axis.size()):
precip_raster = model.statistics.precipitation(
cids, time_step) # example raster output
self.assertEqual(precip_raster.size(), num_cells)
# example single value spatial aggregation (area-weighted) over cids for a specific timestep
avg_gs_sc_value = model.gamma_snow_response.sca_value(cids, 1)
self.assertGreaterEqual(avg_gs_sc_value, 0.0)
avg_gs_sca = model.gamma_snow_response.sca(cids) # swe output
self.assertIsNotNone(avg_gs_sca)
# lwc surface_heat alpha melt_mean melt iso_pot_energy temp_sw
avg_gs_albedo = model.gamma_snow_state.albedo(cids)
self.assertIsNotNone(avg_gs_albedo)
self.assertEqual(avg_temperature.size(), time_axis.size(),
"expect results equal to time-axis size")
copy_region_model = model.__class__(model)
self.assertIsNotNone(copy_region_model)
copy_region_model.run_cells(
) # just to verify we can copy and run the new model
#
# Play with routing and river-network
#
# 1st: add a river, with 36.000 meter hydro length, a UHGParameter with 1m/hour speed, alpha/beta suitable
model.river_network.add(
api.River(1, api.RoutingInfo(0, 3000.0),
api.UHGParameter(1 / 3.60, 7.0, 0.0))) # river id =1
# 2nd: let cells route to the river
model.connect_catchment_to_river(
0, 1) # now all cells in catchment 0 routes to river with id 1.
self.assertTrue(model.has_routing())
# 3rd: now we can have a look at water coming in and out
river_out_m3s = model.river_output_flow_m3s(
1) # should be delayed and reshaped
river_local_m3s = model.river_local_inflow_m3s(
1
) # should be equal to cell outputs (no routing stuff from cell to river)
river_upstream_inflow_m3s = model.river_upstream_inflow_m3s(
1
) # should be 0.0 in this case, since we do not have a routing network
self.assertIsNotNone(river_out_m3s)
self.assertAlmostEqual(river_out_m3s.value(8), 31.57297, 0)
self.assertIsNotNone(river_local_m3s)
self.assertIsNotNone(river_upstream_inflow_m3s)
model.connect_catchment_to_river(0, 0)
self.assertFalse(model.has_routing())