-
Notifications
You must be signed in to change notification settings - Fork 118
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
fix(lak): fix issue with outlet observations (#1646)
* in cases with multiple outlets in a lake the outlet observation used the incorrect index when evaluating the package ibound * added test that compares outlet observations to known values
- Loading branch information
1 parent
61b0883
commit 577b14d
Showing
2 changed files
with
228 additions
and
4 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,227 @@ | ||
""" | ||
Test for lake package outlet observations. | ||
The test evaluates the total outlet flow calculated using lake boundname | ||
with EXT-OUTFLOW and OUTLET flow for all three outlets. | ||
""" | ||
|
||
|
||
import os | ||
|
||
import flopy | ||
import numpy as np | ||
import pathlib as pl | ||
import pytest | ||
|
||
from framework import DNODATA, TestFramework | ||
|
||
cases = ["lakoutlet_obs"] | ||
|
||
|
||
def build_models(idx, test): | ||
nlay = 2 | ||
nrow = 3 | ||
ncol = 3 | ||
nper = 1 | ||
delc = 1000.0 | ||
delr = 1000.0 | ||
top = 5.0 | ||
botm = [-5.0, -10.] | ||
|
||
perlen = [1.0] | ||
nstp = [1] | ||
tsmult = [1.0] | ||
|
||
Kh = 1.0 | ||
|
||
tdis_rc = [(1.0, 1, 1.0)] | ||
|
||
name = cases[idx] | ||
|
||
# build MODFLOW 6 files | ||
sim = flopy.mf6.MFSimulation( | ||
sim_name=name, | ||
sim_ws=test.workspace, | ||
) | ||
|
||
# create tdis package | ||
tdis = flopy.mf6.ModflowTdis( | ||
sim, | ||
time_units="DAYS", | ||
nper=nper, | ||
perioddata=tdis_rc, | ||
) | ||
|
||
# create gwf model | ||
gwf = flopy.mf6.ModflowGwf(sim, modelname=name) #, newtonoptions="newton") | ||
|
||
imsgwf = flopy.mf6.ModflowIms( | ||
sim, | ||
print_option="SUMMARY", | ||
complexity="simple", | ||
inner_dvclose=1e-9, | ||
outer_dvclose=1e-8, | ||
# linear_acceleration="bicgstab", | ||
) | ||
|
||
# number of columns to be a lake for layer 1, 2, , ... len(lakend) | ||
idomain = np.ones((nlay, nrow, ncol), dtype=int) | ||
idomain[0, 1, 1] = 0 | ||
dis = flopy.mf6.ModflowGwfdis( | ||
gwf, | ||
nlay=nlay, | ||
nrow=nrow, | ||
ncol=ncol, | ||
delr=delr, | ||
delc=delc, | ||
top=top, | ||
botm=botm, | ||
idomain=idomain, | ||
) | ||
|
||
# initial conditions | ||
strt = 0.0 | ||
ic = flopy.mf6.ModflowGwfic(gwf, strt=strt) | ||
|
||
# node property flow | ||
npf = flopy.mf6.ModflowGwfnpf( | ||
gwf, | ||
icelltype=1, | ||
k=Kh, | ||
) | ||
|
||
sy = 0.3 | ||
ss = 0.0 | ||
sto = flopy.mf6.ModflowGwfsto(gwf, sy=sy, ss=ss, iconvert=1) | ||
|
||
lake_connect = [ | ||
(0, 1), | ||
(1, 0), | ||
(1, 2), | ||
(2, 1), | ||
] | ||
nlakeconn = len(lake_connect) + 1 | ||
|
||
# pak_data = [ifno, strt, nlakeconn] | ||
pak_data = [(0, strt, nlakeconn, "LAKE1")] | ||
|
||
bedleak = DNODATA | ||
belev = botm[0] | ||
con_data = [ | ||
( | ||
0, | ||
idx, | ||
(0, i, j), | ||
"HORIZONTAL", | ||
bedleak, | ||
belev, | ||
top, | ||
delr / 2.0, | ||
delr, | ||
) | ||
for idx, (i, j) in enumerate(lake_connect) | ||
] | ||
con_data.append( | ||
( | ||
0, | ||
4, | ||
(1, 1, 1), | ||
"VERTICAL", | ||
bedleak, | ||
belev, | ||
top, | ||
0.0, | ||
0.0, | ||
) | ||
|
||
) | ||
|
||
# outlet data | ||
outlet_data = [ | ||
(0, 0, -1, "specified", 0.0, 0.0, 0.0, 0.0), | ||
(1, 0, -1, "specified", 0.0, 0.0, 0.0, 0.0), | ||
(2, 0, -1, "specified", 0.0, 0.0, 0.0, 0.0), | ||
] | ||
noutlets = len(outlet_data) | ||
|
||
# period data | ||
p_data = [ | ||
(0, "RATE", -3.0), | ||
(1, "RATE", -2.0), | ||
(2, "RATE", -1.0), | ||
# (0, "INFLOW", 6.0), | ||
] | ||
|
||
lak_obs = { | ||
f"{name}.lak.obs.csv": [ | ||
("total", "ext-outflow", "LAKE1"), | ||
("o1", "outlet", (0,)), | ||
("o2", "outlet", (1,)), | ||
("o3", "outlet", (2,)), | ||
], | ||
} | ||
|
||
lak = flopy.mf6.modflow.ModflowGwflak( | ||
gwf, | ||
surfdep=0.0, | ||
save_flows=True, | ||
print_input=True, | ||
print_flows=True, | ||
print_stage=True, | ||
boundnames=True, | ||
nlakes=len(pak_data), | ||
ntables=0, | ||
noutlets=noutlets, | ||
outlets=outlet_data, | ||
packagedata=pak_data, | ||
pname="LAK-1", | ||
connectiondata=con_data, | ||
perioddata=p_data, | ||
observations=lak_obs, | ||
) | ||
|
||
# idx = np.where(idomain == 1) | ||
# chdspd = [] | ||
# for k, i, j in zip(*idx): | ||
# chdspd.append(((k, i, j), 0.0)) | ||
# chd = flopy.mf6.modflow.ModflowGwfchd(gwf, stress_period_data=chdspd) | ||
|
||
rech = 6.0 / (8.0 * delr * delc) | ||
rch = flopy.mf6.modflow.ModflowGwfrcha( | ||
gwf, print_flows=True, save_flows=True, recharge=rech, | ||
) | ||
|
||
# output control | ||
oc = flopy.mf6.ModflowGwfoc( | ||
gwf, | ||
printrecord=[("HEAD", "ALL"), ("BUDGET", "ALL")], | ||
) | ||
|
||
return sim, None | ||
|
||
|
||
def check_output(idx, test): | ||
name = cases[idx] | ||
obs_values = flopy.utils.Mf6Obs(test.workspace / f"{name}.lak.obs.csv") | ||
test_values = { | ||
"TOTAL": [-6.0], | ||
"O1": [-3.0], | ||
"O2": [-2.0], | ||
"O3": [-1.0], | ||
} | ||
for key, value in test_values.items(): | ||
assert np.array_equal( | ||
obs_values.get_data(obsname=key)[key], value | ||
), f"failed comparison for '{key}' observation" | ||
|
||
|
||
@pytest.mark.parametrize("idx, name", enumerate(cases)) | ||
def test_mf6model(idx, name, function_tmpdir, targets): | ||
test = TestFramework( | ||
name=name, | ||
workspace=function_tmpdir, | ||
targets=targets, | ||
build=lambda t: build_models(idx, t), | ||
check=lambda t: check_output(idx, t), | ||
) | ||
test.run() |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters