def test_kw_property_indexing(self):
cast = Cast(pressure=self.p,
temp=self.temp,
sal=self.sal,
name="Cruise station 7")
self.assertEqual(cast.p["name"], "Cruise station 7")
return
def test_concatenation(self):
p = np.arange(1, 1001, 2)
temp = 12. * np.exp(-.007 * p) - 14. * np.exp(-0.005 * (p + 100)) + 1.8
sal = -13. * np.exp(-.01 * p) + 34.5
cast2 = Cast(pres=p, temp=temp, sal=sal)
cc = self.cast + cast2
self.assertTrue(isinstance(cc, CastCollection))
self.assertEqual(len(cc), 2)
return
def setUp(self):
p = np.arange(1, 1001, 2)
temp = 10. * np.exp(-.008*p) - 15. * np.exp(-0.005*(p+100)) + 2.
sal = -14. * np.exp(-.01*p) + 34.
self.p = p
self.temp = temp
self.sal = sal
dt = datetime.datetime(1993, 8, 18, 14, 42, 36)
self.cast = Cast(pres=self.p, temp=self.temp, sal=self.sal, date=dt)
self.ctd = CTDCast(self.p, self.sal, self.temp, date=dt)
self.collection = CastCollection(self.ctd, self.ctd)
return
def setUp(self):
p = np.linspace(1, 999, 500)
casts = []
for i in range(10):
cast = Cast(pres=p,
temp=2 * np.ones_like(p),
sal=30 * np.ones_like(p),
station=i,
val=abs(i - 5),
uniq_val=-i**2)
casts.append(cast)
self.cc = CastCollection(casts)
return
def ccmeanp(cc):
if False in (np.all(cc[0]["pres"] == c["pres"]) for c in cc):
raise ValueError("casts must share pressure levels")
p = cc[0]["pres"]
# shared keys are those in all casts, minus pressure and botdepth
sharedkeys = set(cc[0].data.keys()).intersection(
*[set(c.data.keys())
for c in cc[1:]]).difference(set(("pres", "botdepth", "time")))
data = dict()
for key in sharedkeys:
arr = np.vstack([c.data[key] for c in cc])
data[key] = _nanmean(arr, axis=1)
return Cast(p, **data)
def test_defray(self):
lengths = np.arange(50, 71)
casts = []
for n in lengths:
p = np.r_[np.arange(0, 250, 5),
np.arange(250, 250 + 5 * (n - 50), 5)]
t = np.ones(n) * 10.0
s = np.ones(n) * 34.0
cast = Cast(pres=p, temp=t, sal=s)
casts.append(cast)
defrayed_casts = CastCollection(casts).defray()
for cast in defrayed_casts:
self.assertEqual(len(cast), 70)
return
def test_nearest_to(self):
casts = []
for i in range(10):
casts.append(
Cast(pressure=[1, 2, 3],
temperature=[5, 6, 7],
coordinates=(-30.0 + i, 10.0 - 2 * i)))
cc = CastCollection(casts)
pt = Point((-26.2, 2.1), crs=LonLatWGS84)
nearest, dist = cc.nearest_to_point(pt)
self.assertEqual(nearest.coordinates.vertex, (-26.0, 2.0))
self.assertAlmostEqual(dist, 24845.9422, places=4)
return
def test_eofs(self):
pres = np.arange(1, 300)
casts = [
Cast(depth=np.arange(1, 300), theta=np.sin(pres * i * np.pi / 300))
for i in range(3)
]
cc = CastCollection(casts)
structures, lamb, eofs = narwhal.analysis.eofs(cc, key="theta")
self.assertAlmostEqual(np.mean(np.abs(structures["theta_eof1"])),
0.634719360307)
self.assertAlmostEqual(np.mean(np.abs(structures["theta_eof2"])),
0.363733575635)
self.assertAlmostEqual(np.mean(np.abs(structures["theta_eof3"])),
0.350665870142)
self.assertTrue(
np.allclose(lamb, [87.27018523, 40.37800904, 2.02016724]))
return
def ccmeans(cc):
""" Calculate a mean Cast along isopycnals from a CastCollection. """
c0 = max(cc, key=lambda c: c.nvalid())
s0 = c0["sigma"]
sharedkeys = set(c0.data.keys()).intersection(
*[set(c.data.keys())
for c in cc[1:]]).difference(set(("pres", "botdepth", "time")))
nanmask = reduce(lambda a, b: a * b, [c.nanmask() for c in cc])
data = dict()
for key in sharedkeys:
arr = np.nan * np.empty((len(cc), len(c0["pres"])))
arr[0, :] = c0[key]
for j, c in enumerate(cc[1:]):
s = np.convolve(c["sigma"], np.ones(3) / 3.0, mode="same")
arr[j + 1, :] = np.interp(s0, s, c[key])
data[key] = _nanmean(arr, axis=1)
data[key][nanmask] = np.nan
return Cast(copy.copy(c0["pres"]), **data)
def test_read_property_using_alias(self):
cast = Cast(pressure=self.p, temp=self.temp, sal=self.sal, time="late")
self.assertEqual(cast.p["time"], "late")
return
def test_add_property_using_alias(self):
cast = Cast(pres=self.p, temp=self.temp, sal=self.sal)
cast.p["comment"] = "performed bottle cast #23"
self.assertEqual(cast.properties["comment"][-2:], "23")
return