def do2_dofst_volt(voltage_counts, Voffset, Soc, A, B, C, E, P, T, SP, lat, lon):
"""do2_dofst_volt
Takes voltage counts measured from a DOFST-A (SBE 43) Oxygen sensor
attached to a CTDPF-A (SBE 16+ V2) CTD, and converts the counts to a
voltage and then voltage to dissolved oxygen in units of
micromoles/kg for the OOI level 2 data product DOCONCF L2 (fast
response oxygen) in combination with salinity, temperature, and
pressure from the CTD.
A Wrapper function for "dofst_calc".
Usage:
DO = do2_dofst_volt(volt_counts,Voffset,Soc,A,B,C,E,P,T,SP,lat,lon)
where
DO = corrected dissolved oxygen [micro-mole/kg], DOCONCF_L2
volt_counts = Oxygen sensor voltage counts [counts], DOCONCF_L0
Voffset = Voltage offset [V].
Soc = Oxygen signal slope [units are inherently inverse volts]
A = Residual temperature correction factor A
B = Residual temperature correction factor B
C = Residual temperature correction factor C
E = Pressure correction factor
P = PRESWAT water pressure [dbar]. (see
1341-00020_Data_Product_Spec_PRESWAT)
T = TEMPWAT water temperature [deg C]. (see
1341-00010_Data_Product_Spec_TEMPWAT)
SP = PRACSAL practical salinity [unitless]. (see
1341-00040_Data_Product_Spec_PRACSAL)
lat, lon = latitude and longitude of the instrument [degrees].
Example:
v_counts = 16384
P = 201.2
T = 30.3
SP = 31.2
lat,lon = 39.0, -70.5
A = -3.1867e-3, B = 1.7749e-4, C = -3.5718e-6
E = 0.036, Voffset = -0.5186, Soc = 0.4396
DO = do2_dofst_volt(v_counts,Voffset,Soc,A,B,C,E,P,T,SP,lat,lon)
print DO
> 61.89990653
See Also: dofst_calc
Implemented by:
2013-08-20: Stuart Pearce. Initial Code.
2015-08-05: Russell Desiderio. Added fillvalue conversion to Nan.
"""
# replace fill values with nan
voltage_counts = replace_fill_with_nan(None, voltage_counts)
# convert voltage counts to voltss
volts = voltage_counts / 13107.
do, do_int = dofst_calc(volts, Voffset, Soc, A, B, C, E, P, T, SP, lat, lon)
return do
def do2_dofst_frequency(frequency, Foffset, Soc, A, B, C, E, P, T, SP, lat,
lon):
"""do2_dofst_frequency
Takes a frequency measured from a DOFST-K (SBE 43F) Oxygen sensor
connected to a CTDPF-CKL (SBE 52-MP) profiling CTD, and converts the
frequency to dissolved oxygen in units of micromoles/kg for the OOI
level 2 data product DOCONCF L2 (fast response oxygen) in
combination with salinity, temperature, and pressure from the CTD.
A Wrapper function for "dofst_calc".
Usage:
DO = do2_dofst_frequency(frequency,Foffset,Soc,A,B,C,E,P,T,SP,lat,lon)
where
DO = corrected dissolved oxygen [micro-mole/kg], DOCONCF_L2.
frequency = Oxygen sensor frequency [Hz], DOCONCF_L0.
Foffset = Frequency offset [Hz].
Soc = Oxygen signal slope [units are inherently inverse Hz = seconds]
A = Residual temperature correction factor A
B = Residual temperature correction factor B
C = Residual temperature correction factor C
E = Pressure correction factor
P = PRESWAT water pressure [dbar]. (see
1341-00020_Data_Product_Spec_PRESWAT)
T = TEMPWAT water temperature [deg C]. (see
1341-00010_Data_Product_Spec_TEMPWAT)
SP = PRACSAL practical salinity [unitless]. (see
1341-00040_Data_Product_Spec_PRACSAL)
lat, lon = latitude and longitude of the instrument [degrees].
Example:
f = 4354
P = 60.5200
T = 15.5257
SP = 34.1145
lat,lon = 45.0, -125.0
A = -4.1168e-3, B = 2.4818e-4, C = -3.8820e-6
E = 0.036, Foffset = -839.55, Soc = 2.9968e-4
DO = do2_dofst_frequency(f,Foffset,Soc,A,B,C,E,P,T,SP,lat,lon)
print DO
> 256.97434863158
See Also: dofst_calc
Implemented by:
2013-08-20: Stuart Pearce. Initial Code.
2015-08-05: Russell Desiderio. Added fillvalue conversion to Nan.
"""
# replace fill values with nan
frequency = replace_fill_with_nan(None, frequency)
do, do_int = dofst_calc(frequency, Foffset, Soc, A, B, C, E, P, T, SP, lat,
lon)
return do
def do2_dofst_frequency(frequency, Foffset, Soc, A, B, C, E, P, T, SP, lat, lon):
"""do2_dofst_frequency
Takes a frequency measured from a DOFST-K (SBE 43F) Oxygen sensor
connected to a CTDPF-CKL (SBE 52-MP) profiling CTD, and converts the
frequency to dissolved oxygen in units of micromoles/kg for the OOI
level 2 data product DOCONCF L2 (fast response oxygen) in
combination with salinity, temperature, and pressure from the CTD.
A Wrapper function for "dofst_calc".
Usage:
DO = do2_dofst_frequency(frequency,Foffset,Soc,A,B,C,E,P,T,SP,lat,lon)
where
DO = corrected dissolved oxygen [micro-mole/kg], DOCONCF_L2.
frequency = Oxygen sensor frequency [Hz], DOCONCF_L0.
Foffset = Frequency offset [Hz].
Soc = Oxygen signal slope [units are inherently inverse Hz = seconds]
A = Residual temperature correction factor A
B = Residual temperature correction factor B
C = Residual temperature correction factor C
E = Pressure correction factor
P = PRESWAT water pressure [dbar]. (see
1341-00020_Data_Product_Spec_PRESWAT)
T = TEMPWAT water temperature [deg C]. (see
1341-00010_Data_Product_Spec_TEMPWAT)
SP = PRACSAL practical salinity [unitless]. (see
1341-00040_Data_Product_Spec_PRACSAL)
lat, lon = latitude and longitude of the instrument [degrees].
Example:
f = 4354
P = 60.5200
T = 15.5257
SP = 34.1145
lat,lon = 45.0, -125.0
A = -4.1168e-3, B = 2.4818e-4, C = -3.8820e-6
E = 0.036, Foffset = -839.55, Soc = 2.9968e-4
DO = do2_dofst_frequency(f,Foffset,Soc,A,B,C,E,P,T,SP,lat,lon)
print DO
> 256.97434863158
See Also: dofst_calc
Implemented by:
2013-08-20: Stuart Pearce. Initial Code.
2015-08-05: Russell Desiderio. Added fillvalue conversion to Nan.
"""
# replace fill values with nan
frequency = replace_fill_with_nan(None, frequency)
do, do_int = dofst_calc(frequency, Foffset, Soc, A, B, C, E, P, T, SP, lat, lon)
return do
def o2_counts_to_uM(o2_counts):
"""
Description:
Computes the DOCONCS_L1 data product from a DOSTA Aanderaa Optode
connected to a SBE 16+ V2 CTD via RS-232.
Usage:
DO = o2_counts_to_uM(o2_counts)
where
DO = DOCONCS_L1, dissolved oxygen concentration uncorrected for salinity
and pressure effects [micro-mole/L]
o2_counts = DOCONCS-CNT_L0, counts from the CTD
Implemented by:
2013-04-26: Stuart Pearce. Initial Code.
2015-04-10: Russell Desiderio. Added documentation and fill value handling.
Notes:
The DOCONCS_L1 data product has units of micromole/liter; SAF incorrectly
lists the units for this L1 product as micromole/kg.
References:
OOI (2012). Data Product Specification for Oxygen Concentration
from "Stable" Instruments. Document Control Number
1341-00520. https://alfresco.oceanobservatories.org/ (See:
Company Home >> OOI >> Controlled >> 1000 System Level
>> 1341-00520_Data_Product_SPEC_DOCONCS_OOI.pdf)
"""
# replace fill values with nan
o2_counts = replace_fill_with_nan(None, o2_counts)
DO = (o2_counts / 10000.0) - 10.0
return DO
def o2_counts_to_uM(o2_counts):
"""
Description:
Computes the DOCONCS_L1 data product from a DOSTA Aanderaa Optode
connected to a SBE 16+ V2 CTD via RS-232.
Usage:
DO = o2_counts_to_uM(o2_counts)
where
DO = DOCONCS_L1, dissolved oxygen concentration uncorrected for salinity
and pressure effects [micro-mole/L]
o2_counts = DOCONCS-CNT_L0, counts from the CTD
Implemented by:
2013-04-26: Stuart Pearce. Initial Code.
2015-04-10: Russell Desiderio. Added documentation and fill value handling.
Notes:
The DOCONCS_L1 data product has units of micromole/liter; SAF incorrectly
lists the units for this L1 product as micromole/kg.
References:
OOI (2012). Data Product Specification for Oxygen Concentration
from "Stable" Instruments. Document Control Number
1341-00520. https://alfresco.oceanobservatories.org/ (See:
Company Home >> OOI >> Controlled >> 1000 System Level
>> 1341-00520_Data_Product_SPEC_DOCONCS_OOI.pdf)
"""
# replace fill values with nan
o2_counts = replace_fill_with_nan(None, o2_counts)
DO = (o2_counts / 10000.0) - 10.0
return DO
def do2_dofst_volt(voltage_counts, Voffset, Soc, A, B, C, E, P, T, SP, lat,
lon):
"""do2_dofst_volt
Takes voltage counts measured from a DOFST-A (SBE 43) Oxygen sensor
attached to a CTDPF-A (SBE 16+ V2) CTD, and converts the counts to a
voltage and then voltage to dissolved oxygen in units of
micromoles/kg for the OOI level 2 data product DOCONCF L2 (fast
response oxygen) in combination with salinity, temperature, and
pressure from the CTD.
A Wrapper function for "dofst_calc".
Usage:
DO = do2_dofst_volt(volt_counts,Voffset,Soc,A,B,C,E,P,T,SP,lat,lon)
where
DO = corrected dissolved oxygen [micro-mole/kg], DOCONCF_L2
volt_counts = Oxygen sensor voltage counts [counts], DOCONCF_L0
Voffset = Voltage offset [V].
Soc = Oxygen signal slope [units are inherently inverse volts]
A = Residual temperature correction factor A
B = Residual temperature correction factor B
C = Residual temperature correction factor C
E = Pressure correction factor
P = PRESWAT water pressure [dbar]. (see
1341-00020_Data_Product_Spec_PRESWAT)
T = TEMPWAT water temperature [deg C]. (see
1341-00010_Data_Product_Spec_TEMPWAT)
SP = PRACSAL practical salinity [unitless]. (see
1341-00040_Data_Product_Spec_PRACSAL)
lat, lon = latitude and longitude of the instrument [degrees].
Example:
v_counts = 16384
P = 201.2
T = 30.3
SP = 31.2
lat,lon = 39.0, -70.5
A = -3.1867e-3, B = 1.7749e-4, C = -3.5718e-6
E = 0.036, Voffset = -0.5186, Soc = 0.4396
DO = do2_dofst_volt(v_counts,Voffset,Soc,A,B,C,E,P,T,SP,lat,lon)
print DO
> 61.89990653
See Also: dofst_calc
Implemented by:
2013-08-20: Stuart Pearce. Initial Code.
2015-08-05: Russell Desiderio. Added fillvalue conversion to Nan.
"""
# replace fill values with nan
voltage_counts = replace_fill_with_nan(None, voltage_counts)
# convert voltage counts to voltss
volts = voltage_counts / 13107.
do, do_int = dofst_calc(volts, Voffset, Soc, A, B, C, E, P, T, SP, lat,
lon)
return do
def test_replace_fill_with_nan_with_one_data_argument(self):
"""
Description:
Tests replace_fill_with_nan when called with only one data argument.
Coding replace_fill_with_nan for this was initially not straightforward
due to the core python\numpy "interface".
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-23: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up a list of 8 input integer arrays to be tested one at a time
v4 = np.array([[1, 2, 3, 4, 5]])
v5 = np.array([[11, ifill, 13, zfill, sfill]])
v = [np.array([500]),
np.array([ifill]),
np.array([zfill]),
np.array([sfill]),
v4,
v5,
np.tile(v4, (3, 1))[np.newaxis, :],
np.vstack((v5, v4, v5))[np.newaxis, :]
]
### CASE(1): no instrument fill values
# expected values
x = [np.array([500.0]),
np.array([ifill]).astype('float'),
np.array([zfill]).astype('float'),
np.array([np.nan]),
np.array([[1.0, 2.0, 3.0, 4.0, 5.0]]),
np.array([[11, ifill, 13, zfill, np.nan]]).astype('float'),
v[6].astype(float),
np.copy(v[7].astype(float))
]
x[7][0, 0, 4] = np.nan
x[7][0, 2, 4] = np.nan
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan(None, v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x[0], x[1], x[3], x[4] and x[6] are unchanged
x[2] = np.array([np.nan])
x[5][0, 3] = np.nan # x[5] is now reset
x[7][0, 0, 3] = np.nan
x[7][0, 2, 3] = np.nan # x[7] is now reset
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan(zfill, v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; only list elements 5 and 7 will change
x[1] = np.array([np.nan])
x[5][0, 1] = np.nan # x[5] is now reset
x[7][0, 0, 1] = np.nan
x[7][0, 2, 1] = np.nan # x[7] is now reset
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan([ifill, zfill], v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
def test_replace_fill_with_nan_with_vectortime_variables(self):
"""
Description:
Tests replace_fill_with_nan for variables that are time-vectorized
with more than one time point. Functionally this means that the lead
dimension of these test variables will have a shape greater than 1.
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-11: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up 6 integer arrays to be tested, v1-v6.
# this will also designate their ordering on input and output
v1_good_1D = np.array([1, 2, 3, 4, 5])
v2_good_2D = np.tile(v1_good_1D, (3, 1))
v3_good_3D = np.tile(v2_good_2D, (3, 1, 1))
v4_fill_1D = np.array([11, ifill, 13, zfill, sfill])
v5_fill_2D = np.vstack((v4_fill_1D, v1_good_1D, v4_fill_1D))
v6_fill_3D = np.copy(v3_good_3D)
v6_fill_3D[0, 0, 0] = zfill
v6_fill_3D[1, 1, 1] = sfill
v6_fill_3D[2, 2, 2:5] = ifill # last 3 elements in row '2'
# and a float array which should pass through unchanged
v7_float_array = v3_good_3D.astype(float)
v7_float_array[0, 1, 2] = np.nan
#SET INPUTS
v1, v2, v3 = np.copy(v1_good_1D), np.copy(v2_good_2D), np.copy(v3_good_3D)
v4, v5, v6 = np.copy(v4_fill_1D), np.copy(v5_fill_2D), np.copy(v6_fill_3D)
v7 = np.copy(v7_float_array)
### CASE 1: no instrument fill values, only system fill values
# set expected outputs
x1 = v1_good_1D.astype(float)
x2 = v2_good_2D.astype(float)
x3 = v3_good_3D.astype(float)
x4 = np.array([11.0, float(ifill), 13.0, float(zfill), np.nan])
x5 = np.vstack((x4, x1, x4))
x6 = v6_fill_3D.astype(float)
x6[1, 1, 1] = np.nan
x7 = np.copy(v7_float_array)
# (1a): instrument_fillvalue = None
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
None, v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (1b) instrument_fillvalue = empty list
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
[], v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (1c) instrument_fillvalue: empty ndarray
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
np.array([]), v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x1-x3 and x7 are unchanged
x4[3] = np.nan # x4 is now reset
x5 = np.vstack((x4, x1, x4))
x6[0, 0, 0] = np.nan # x6 is now reset
# (2a) instrument_fillvalue is a scalar
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
zfill, v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (2b) instrument_fillvalue is a one-element list
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
[zfill], v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (2c) instrument_fillvalue is a shape (1,) ndarray
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
np.array([zfill]), v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; x1-x3 and x7 are unchanged
x4[1] = np.nan # x4 is now reset
x5 = np.vstack((x4, x1, x4))
x6[2, 2, 2:5] = np.nan # x6 is now reset
# (3a) instrument_fillvalues are contained in a list
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
[ifill, zfill], v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (3b) instrument_fillvalues are contained in an ndarray
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan(
np.array([ifill, zfill]), v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
def test_replace_fill_with_nan_with_scalartime_variables(self):
"""
Description:
Tests replace_fill_with_nan for variables that are time-vectorized
with one time point. Functionally this means that the lead dimension
of these test variables will have shape 1.
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-12: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up 8 integer arrays to be tested, v1-v8.
# this will also designate their ordering on input and output
v1 = np.array([500])
v2 = np.array([ifill])
v3 = np.array([zfill])
v4 = np.array([sfill])
v5 = np.array([[1, 2, 3, 4, 5]])
v6 = np.array([[11, ifill, 13, zfill, sfill]])
v7 = np.tile(v5, (3, 1))[np.newaxis, :]
v8 = np.vstack((v6, v5, v6))[np.newaxis, :]
### CASE 1: no instrument fill values, only system fill values
# set expected outputs
x1, x2, x3 = v1.astype(float), v2.astype(float), v3.astype(float)
x4 = np.array([np.nan])
x5 = v5.astype(float)
x6 = np.array([[11.0, float(ifill), 13.0, float(zfill), np.nan]])
x7 = v7.astype(float)
x8 = np.copy(v8.astype(float))
x8[0, 0, 4] = np.nan
x8[0, 2, 4] = np.nan
# (1a) instrument_fillvalue = None
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
None, v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (1b) instrument_fillvalue: empty list
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
[], v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (1c) instrument_fillvalue: empty ndarray
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
np.array([]), v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x1, x2, x4, x5 and x7 are unchanged
x3 = np.array([np.nan])
x6[0, 3] = np.nan # x6 is now reset
x8[0, 0, 3] = np.nan
x8[0, 2, 3] = np.nan # x8 is now reset
# (2a) instrument_fillvalue is a scalar
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
zfill, v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (2b) instrument_fillvalue is a one element list
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
[zfill], v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (2c) instrument_fillvalue is a one element ndarray
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
np.array([zfill]), v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; x1,x3-x5 and x7 are unchanged
x2 = np.array([np.nan])
x6[0, 1] = np.nan # x6 is now reset
x8[0, 0, 1] = np.nan
x8[0, 2, 1] = np.nan # x8 is now reset
# (3a) instrument_fillvalues are contained in a list
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
[ifill, zfill], v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (3b) instrument_fillvalues are contained in an ndarray
(c1, c2, c3, c4, c5, c6, c7, c8) = gfunc.replace_fill_with_nan(
np.array([ifill, zfill]), v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
def test_replace_fill_with_nan_with_one_data_argument(self):
"""
Description:
Tests replace_fill_with_nan when called with only one data argument.
Coding replace_fill_with_nan for this was initially not straightforward
due to the core python\numpy "interface".
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-23: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up a list of 8 input integer arrays to be tested one at a time
v4 = np.array([[1, 2, 3, 4, 5]])
v5 = np.array([[11, ifill, 13, zfill, sfill]])
v = [
np.array([500]),
np.array([ifill]),
np.array([zfill]),
np.array([sfill]), v4, v5,
np.tile(v4, (3, 1))[np.newaxis, :],
np.vstack((v5, v4, v5))[np.newaxis, :]
]
### CASE(1): no instrument fill values
# expected values
x = [
np.array([500.0]),
np.array([ifill]).astype('float'),
np.array([zfill]).astype('float'),
np.array([np.nan]),
np.array([[1.0, 2.0, 3.0, 4.0, 5.0]]),
np.array([[11, ifill, 13, zfill, np.nan]]).astype('float'),
v[6].astype(float),
np.copy(v[7].astype(float))
]
x[7][0, 0, 4] = np.nan
x[7][0, 2, 4] = np.nan
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan(None, v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x[0], x[1], x[3], x[4] and x[6] are unchanged
x[2] = np.array([np.nan])
x[5][0, 3] = np.nan # x[5] is now reset
x[7][0, 0, 3] = np.nan
x[7][0, 2, 3] = np.nan # x[7] is now reset
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan(zfill, v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; only list elements 5 and 7 will change
x[1] = np.array([np.nan])
x[5][0, 1] = np.nan # x[5] is now reset
x[7][0, 0, 1] = np.nan
x[7][0, 2, 1] = np.nan # x[7] is now reset
for ii in range(len(v)):
c = gfunc.replace_fill_with_nan([ifill, zfill], v[ii])
np.testing.assert_array_almost_equal(c, x[ii], decimal=8)
def test_replace_fill_with_nan_with_vectortime_variables(self):
"""
Description:
Tests replace_fill_with_nan for variables that are time-vectorized
with more than one time point. Functionally this means that the lead
dimension of these test variables will have a shape greater than 1.
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-11: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up 6 integer arrays to be tested, v1-v6.
# this will also designate their ordering on input and output
v1_good_1D = np.array([1, 2, 3, 4, 5])
v2_good_2D = np.tile(v1_good_1D, (3, 1))
v3_good_3D = np.tile(v2_good_2D, (3, 1, 1))
v4_fill_1D = np.array([11, ifill, 13, zfill, sfill])
v5_fill_2D = np.vstack((v4_fill_1D, v1_good_1D, v4_fill_1D))
v6_fill_3D = np.copy(v3_good_3D)
v6_fill_3D[0, 0, 0] = zfill
v6_fill_3D[1, 1, 1] = sfill
v6_fill_3D[2, 2, 2:5] = ifill # last 3 elements in row '2'
# and a float array which should pass through unchanged
v7_float_array = v3_good_3D.astype(float)
v7_float_array[0, 1, 2] = np.nan
#SET INPUTS
v1, v2, v3 = np.copy(v1_good_1D), np.copy(v2_good_2D), np.copy(
v3_good_3D)
v4, v5, v6 = np.copy(v4_fill_1D), np.copy(v5_fill_2D), np.copy(
v6_fill_3D)
v7 = np.copy(v7_float_array)
### CASE 1: no instrument fill values, only system fill values
# set expected outputs
x1 = v1_good_1D.astype(float)
x2 = v2_good_2D.astype(float)
x3 = v3_good_3D.astype(float)
x4 = np.array([11.0, float(ifill), 13.0, float(zfill), np.nan])
x5 = np.vstack((x4, x1, x4))
x6 = v6_fill_3D.astype(float)
x6[1, 1, 1] = np.nan
x7 = np.copy(v7_float_array)
# (1a): instrument_fillvalue = None
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan(None, v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (1b) instrument_fillvalue = empty list
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan([], v1, v2,
v3, v4, v5,
v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (1c) instrument_fillvalue: empty ndarray
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan(np.array([]), v1, v2, v3, v4, v5,
v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x1-x3 and x7 are unchanged
x4[3] = np.nan # x4 is now reset
x5 = np.vstack((x4, x1, x4))
x6[0, 0, 0] = np.nan # x6 is now reset
# (2a) instrument_fillvalue is a scalar
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan(zfill, v1, v2, v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (2b) instrument_fillvalue is a one-element list
(c1, c2, c3, c4, c5, c6, c7) = gfunc.replace_fill_with_nan([zfill], v1,
v2, v3, v4,
v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (2c) instrument_fillvalue is a shape (1,) ndarray
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan(np.array([zfill]), v1, v2, v3, v4,
v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; x1-x3 and x7 are unchanged
x4[1] = np.nan # x4 is now reset
x5 = np.vstack((x4, x1, x4))
x6[2, 2, 2:5] = np.nan # x6 is now reset
# (3a) instrument_fillvalues are contained in a list
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan([ifill, zfill], v1, v2, v3, v4, v5,
v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
# (3b) instrument_fillvalues are contained in an ndarray
(c1, c2, c3, c4, c5, c6,
c7) = gfunc.replace_fill_with_nan(np.array([ifill, zfill]), v1, v2,
v3, v4, v5, v6, v7)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
def test_replace_fill_with_nan_with_scalartime_variables(self):
"""
Description:
Tests replace_fill_with_nan for variables that are time-vectorized
with one time point. Functionally this means that the lead dimension
of these test variables will have shape 1.
Cases (instances of the system fill value are always processed):
no instrument fill value
one instrument fill value of 0
two instrument fill values
Implemented by:
2013-06-12: Russell Desiderio. Initial code.
"""
# constants for convenience in filling in test variables
sfill = SYSTEM_FILLVALUE
zfill = ZERO_FILLVALUE
ifill = INST_FILLVALUE
# set up 8 integer arrays to be tested, v1-v8.
# this will also designate their ordering on input and output
v1 = np.array([500])
v2 = np.array([ifill])
v3 = np.array([zfill])
v4 = np.array([sfill])
v5 = np.array([[1, 2, 3, 4, 5]])
v6 = np.array([[11, ifill, 13, zfill, sfill]])
v7 = np.tile(v5, (3, 1))[np.newaxis, :]
v8 = np.vstack((v6, v5, v6))[np.newaxis, :]
### CASE 1: no instrument fill values, only system fill values
# set expected outputs
x1, x2, x3 = v1.astype(float), v2.astype(float), v3.astype(float)
x4 = np.array([np.nan])
x5 = v5.astype(float)
x6 = np.array([[11.0, float(ifill), 13.0, float(zfill), np.nan]])
x7 = v7.astype(float)
x8 = np.copy(v8.astype(float))
x8[0, 0, 4] = np.nan
x8[0, 2, 4] = np.nan
# (1a) instrument_fillvalue = None
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan(None, v1, v2, v3, v4, v5, v6, v7,
v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (1b) instrument_fillvalue: empty list
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan([], v1, v2, v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (1c) instrument_fillvalue: empty ndarray
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan(np.array([]), v1, v2, v3, v4, v5,
v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
### CASE (2): one instrument fill value = 0
# reset expected outputs; x1, x2, x4, x5 and x7 are unchanged
x3 = np.array([np.nan])
x6[0, 3] = np.nan # x6 is now reset
x8[0, 0, 3] = np.nan
x8[0, 2, 3] = np.nan # x8 is now reset
# (2a) instrument_fillvalue is a scalar
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan(zfill, v1, v2, v3, v4, v5, v6, v7,
v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (2b) instrument_fillvalue is a one element list
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan([zfill], v1, v2, v3, v4, v5, v6, v7,
v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (2c) instrument_fillvalue is a one element ndarray
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan(np.array([zfill]), v1, v2, v3, v4,
v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
### CASE (3): two instrument fill values
# reset expected outputs; x1,x3-x5 and x7 are unchanged
x2 = np.array([np.nan])
x6[0, 1] = np.nan # x6 is now reset
x8[0, 0, 1] = np.nan
x8[0, 2, 1] = np.nan # x8 is now reset
# (3a) instrument_fillvalues are contained in a list
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan([ifill, zfill], v1, v2, v3, v4, v5,
v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
# (3b) instrument_fillvalues are contained in an ndarray
(c1, c2, c3, c4, c5, c6, c7,
c8) = gfunc.replace_fill_with_nan(np.array([ifill, zfill]), v1, v2,
v3, v4, v5, v6, v7, v8)
np.testing.assert_array_almost_equal(c1, x1, decimal=8)
np.testing.assert_array_almost_equal(c2, x2, decimal=8)
np.testing.assert_array_almost_equal(c3, x3, decimal=8)
np.testing.assert_array_almost_equal(c4, x4, decimal=8)
np.testing.assert_array_almost_equal(c5, x5, decimal=8)
np.testing.assert_array_almost_equal(c6, x6, decimal=8)
np.testing.assert_array_almost_equal(c7, x7, decimal=8)
np.testing.assert_array_almost_equal(c8, x8, decimal=8)
