def test_min_time(self):
input = Pyramid({0: self.tiled_raster_rdd})
cube = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
env = EvalEnv()
min_time = cube.reduce_dimension(reducer=reducer('min'),
dimension='t',
env=env)
max_time = cube.reduce_dimension(reducer=reducer('max'),
dimension='t',
env=env)
stitched = min_time.pyramid.levels[0].stitch()
print(stitched)
self.assertEquals(2.0, stitched.cells[0][0][0])
for p in self.points[1:3]:
result = min_time.timeseries(p.x, p.y, srs="EPSG:3857")
print(result)
print(cube.timeseries(p.x, p.y, srs="EPSG:3857"))
max_result = max_time.timeseries(p.x, p.y, srs="EPSG:3857")
self.assertEqual(1.0, result['NoDate'])
self.assertEqual(2.0, max_result['NoDate'])
def test_apply_spatiotemporal(self):
import openeo_udf.functions
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry(), {
"bands": [{
"band_id": "2",
"name": "blue",
"wavelength_nm": 496.6,
"res_m": 10,
"scale": 0.0001,
"offset": 0,
"type": "int16",
"unit": "1"
}]
})
import os, openeo_udf
dir = os.path.dirname(openeo_udf.functions.__file__)
file_name = os.path.join(dir, "datacube_reduce_time_sum.py")
with open(file_name, "r") as f:
udf_code = f.read()
result = imagecollection.apply_tiles_spatiotemporal(udf_code)
stitched = result.pyramid.levels[0].to_spatial_layer().stitch()
print(stitched)
self.assertEqual(2, stitched.cells[0][0][0])
self.assertEqual(6, stitched.cells[0][0][5])
self.assertEqual(4, stitched.cells[0][5][6])
def test_reproject_spatial(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
ref_path = str(self.temp_folder / "reproj_ref.tiff")
imagecollection.reduce('max',
dimension="t").save_result(ref_path,
format="GTIFF")
resampled = imagecollection.resample_spatial(resolution=0,
projection="EPSG:3395",
method="max")
metadata = resampled.pyramid.levels[0].layer_metadata
print(metadata)
self.assertTrue("proj=merc" in metadata.crs)
path = str(self.temp_folder / "reprojected.tiff")
res = resampled.reduce('max', dimension="t")
res.save_result(path, format="GTIFF")
with rasterio.open(ref_path) as ref_ds:
with rasterio.open(path) as ds:
print(ds.profile)
#this reprojection does not change the shape, so we can compare
assert ds.read().shape == ref_ds.read().shape
assert (ds.crs.to_epsg() == 3395)
def test_rename_dimension(self):
imagecollection = GeopysparkDataCube(pyramid=Pyramid(
{0: self.tiled_raster_rdd}),
metadata=self.collection_metadata)
dim_renamed = imagecollection.rename_dimension('t', 'myNewTimeDim')
dim_renamed.metadata.assert_valid_dimension('myNewTimeDim')
def test_mask_raster_replacement_int(self):
def createMask(tile):
tile.cells[0][0][0] = 0.0
return tile
input = Pyramid({0: self.tiled_raster_rdd})
mask_layer = self.tiled_raster_rdd.map_tiles(createMask)
mask = Pyramid({0: mask_layer})
cube = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
mask_cube = GeopysparkDataCube(pyramid=mask)
stitched = cube.mask(mask=mask_cube, replacement=10).reduce(
'max', dimension="t").pyramid.levels[0].stitch()
print(stitched)
assert stitched.cells[0][0][0] == 2.0
assert stitched.cells[0][0][1] == 10.0
def test_aggregate_temporal(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.aggregate_temporal(
["2017-01-01", "2018-01-01"], ["2017-01-03"],
"max").pyramid.levels[0].to_spatial_layer().stitch()
print(stitched)
def test_reduce_nontemporal(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
with self.assertRaises(AttributeError) as context:
imagecollection.reduce("max",
"spectral").pyramid.levels[0].stitch()
print(context.exception)
def test_reduce_nontemporal(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
with self.assertRaises(FeatureUnsupportedException) as context:
imagecollection.reduce(
"max", dimension="gender").pyramid.levels[0].stitch()
print(context.exception)
def test_aggregate_max_time(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
layer = imagecollection.reduce('max', dimension='t').pyramid.levels[0]
stitched = layer.stitch()
assert CellType.FLOAT32.value == layer.layer_metadata.cell_type
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
def _test_aggregate_temporal(self, interval_list):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
stitched = (imagecollection.aggregate_temporal(
interval_list, ["2017-01-03"], "min",
dimension="t").pyramid.levels[0].to_spatial_layer().stitch())
print(stitched)
expected_max = np.min([self.tile2.cells, self.tile.cells], axis=0)
assert_array_almost_equal(stitched.cells[0, 0:5, 0:5], expected_max)
def test_aggregate_max_time(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.reduce(
'max', 'temporal').pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
def test_mask_raster(self):
input = Pyramid({0: self.tiled_raster_rdd})
def createMask(tile):
tile.cells[0][0][0] = 0.0
return tile
mask_layer = self.tiled_raster_rdd.map_tiles(createMask)
mask = Pyramid({0: mask_layer})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.mask(
rastermask=GeotrellisTimeSeriesImageCollection(
mask, InMemoryServiceRegistry()),
replacement=10.0).reduce('max',
'temporal').pyramid.levels[0].stitch()
print(stitched)
self.assertEquals(2.0, stitched.cells[0][0][0])
self.assertEquals(10.0, stitched.cells[0][0][1])
def test_apply_kernel_int(self):
kernel = np.array([[0, 1, 0], [1, 1, 1], [0, 1, 0]])
input = Pyramid({0: self.tiled_raster_rdd})
img = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
stitched = img.apply_kernel(kernel).reduce(
'max', dimension="t").pyramid.levels[0].stitch()
assert stitched.cells[0][0][0] == 6.0
assert stitched.cells[0][0][1] == 8.0
assert stitched.cells[0][1][1] == 10.0
def test_apply_kernel(self):
kernel = np.array([[0.0, 1.0, 0.0], [1.0, 1.0, 1.0], [0.0, 1.0, 0.0]])
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.apply_kernel(kernel, 2.0).reduce(
'max', 'temporal').pyramid.levels[0].stitch()
self.assertEquals(12.0, stitched.cells[0][0][0])
self.assertEquals(16.0, stitched.cells[0][0][1])
self.assertEquals(20.0, stitched.cells[0][1][1])
def test_apply_dimension_spatiotemporal(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(
pyramid=input,
metadata=GeopysparkCubeMetadata({
"cube:dimensions": {
# TODO: also specify other dimensions?
"bands": {
"type": "bands",
"values": ["2"]
}
},
"summaries": {
"eo:bands": [{
"name": "2",
"common_name": "blue",
"wavelength_nm": 496.6,
"res_m": 10,
"scale": 0.0001,
"offset": 0,
"type": "int16",
"unit": "1"
}]
}
}))
udf_code = """
def rct_savitzky_golay(udf_data:UdfData):
from scipy.signal import savgol_filter
print(udf_data.get_datacube_list())
return udf_data
"""
result = imagecollection.apply_tiles_spatiotemporal(udf_code)
local_tiles = result.pyramid.levels[0].to_numpy_rdd().collect()
print(local_tiles)
self.assertEquals(len(TestMultipleDates.layer), len(local_tiles))
ref_dict = {
e[0]: e[1]
for e in imagecollection.pyramid.levels[0].convert_data_type(
CellType.FLOAT64).to_numpy_rdd().collect()
}
result_dict = {e[0]: e[1] for e in local_tiles}
for k, v in ref_dict.items():
tile = result_dict[k]
assert_array_almost_equal(np.squeeze(v.cells),
np.squeeze(tile.cells),
decimal=2)
def test_aggregate_temporal_median(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
stitched = (imagecollection.aggregate_temporal(
["2015-01-01", "2018-01-01"], ["2017-01-03"],
self._median_reducer(),
dimension="t").pyramid.levels[0].to_spatial_layer().stitch())
print(stitched)
expected_median = np.median(
[self.tile.cells, self.tile2.cells, self.tile.cells], axis=0)
#TODO nodata handling??
assert_array_almost_equal(stitched.cells[0, 1:2, 1:2],
expected_median[1:2, 1:2])
def test_reduce(self):
input = Pyramid({0: self.tiled_raster_rdd})
cube = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
env = EvalEnv()
stitched = cube.reduce_dimension(dimension="t",
reducer=reducer("max"),
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(2.0, stitched.cells[0][0][1])
stitched = cube.reduce_dimension(dimension="t",
reducer=reducer("min"),
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(1.0, stitched.cells[0][0][1])
stitched = cube.reduce_dimension(dimension="t",
reducer=reducer("sum"),
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(4.0, stitched.cells[0][0][1])
stitched = cube.reduce_dimension(dimension="t",
reducer=reducer("mean"),
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertAlmostEqual(1.3333333, stitched.cells[0][0][1])
stitched = cube.reduce_dimension(reducer=reducer("variance"),
dimension="t",
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(0.0, stitched.cells[0][0][0])
self.assertAlmostEqual(0.2222222, stitched.cells[0][0][1])
stitched = cube.reduce_dimension(reducer=reducer("sd"),
dimension="t",
env=env).pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(0.0, stitched.cells[0][0][0])
self.assertAlmostEqual(0.4714045, stitched.cells[0][0][1])
def test_resample_spatial(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
resampled = imagecollection.resample_spatial(resolution=0.05)
path = str(self.temp_folder / "resampled.tiff")
res = resampled.reduce('max', dimension="t")
res.save_result(path, format="GTIFF")
import rasterio
with rasterio.open(path) as ds:
print(ds.profile)
self.assertAlmostEqual(0.05, ds.res[0], 3)
def test_resample_spatial(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
resampled = imagecollection.resample_spatial(resolution=0.05)
path = str(self.temp_folder / "resampled.tiff")
resampled.reduce('max',
'temporal').download(path,
format="GTIFF",
parameters={'tiled': True})
import rasterio
with rasterio.open(path) as ds:
print(ds.profile)
self.assertAlmostEqual(0.05, ds.res[0], 3)
def test_reduce_all_data(self):
input = Pyramid({
0:
self._single_pixel_layer({
datetime.datetime.strptime("2016-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
1.0,
datetime.datetime.strptime("2017-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
5.0
})
})
cube = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
env = EvalEnv()
stitched = cube.reduce_dimension(reducer=reducer("min"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertEqual(1.0, stitched.cells[0][0][0])
stitched = cube.reduce_dimension(reducer=reducer("max"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = cube.reduce_dimension(reducer=reducer("sum"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertEqual(6.0, stitched.cells[0][0][0])
stitched = cube.reduce_dimension(reducer=reducer("mean"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertAlmostEqual(3.0, stitched.cells[0][0][0], delta=0.001)
stitched = cube.reduce_dimension(reducer=reducer("variance"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertAlmostEqual(4.0, stitched.cells[0][0][0], delta=0.001)
stitched = cube.reduce_dimension(reducer=reducer("sd"),
dimension="t",
env=env).pyramid.levels[0].stitch()
self.assertAlmostEqual(2.0, stitched.cells[0][0][0], delta=0.001)
def test_apply_dimension_spatiotemporal(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry(), {
"bands": [{
"band_id": "2",
"name": "blue",
"wavelength_nm": 496.6,
"res_m": 10,
"scale": 0.0001,
"offset": 0,
"type": "int16",
"unit": "1"
}]
})
udf_code = """
def rct_savitzky_golay(udf_data:UdfData):
from scipy.signal import savgol_filter
print(udf_data.get_datacube_list())
return udf_data
"""
result = imagecollection.apply_tiles_spatiotemporal(udf_code)
local_tiles = result.pyramid.levels[0].to_numpy_rdd().collect()
print(local_tiles)
self.assertEquals(len(TestMultipleDates.layer), len(local_tiles))
ref_dict = {
e[0]: e[1]
for e in imagecollection.pyramid.levels[0].convert_data_type(
CellType.FLOAT64).to_numpy_rdd().collect()
}
result_dict = {e[0]: e[1] for e in local_tiles}
for k, v in ref_dict.items():
tile = result_dict[k]
assert_array_almost_equal(np.squeeze(v.cells),
np.squeeze(tile.cells),
decimal=2)
def test_min_time(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
min_time = imagecollection.reduce('min', 'temporal')
max_time = imagecollection.reduce('max', 'temporal')
stitched = min_time.pyramid.levels[0].stitch()
print(stitched)
self.assertEquals(2.0, stitched.cells[0][0][0])
for p in self.points[1:3]:
result = min_time.timeseries(p.x, p.y, srs="EPSG:3857")
print(result)
print(imagecollection.timeseries(p.x, p.y, srs="EPSG:3857"))
max_result = max_time.timeseries(p.x, p.y, srs="EPSG:3857")
self.assertEqual(1.0, result['NoDate'])
self.assertEqual(2.0, max_result['NoDate'])
def test_reduce_some_nodata(self):
no_data = -1.0
input = Pyramid({
0:
self._single_pixel_layer(
{
datetime.datetime.strptime("2016-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
no_data,
datetime.datetime.strptime("2017-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
5.0
}, no_data)
})
imagecollection = GeopysparkDataCube(pyramid=input,
metadata=self.collection_metadata)
stitched = imagecollection.reduce(
"min", dimension="t").pyramid.levels[0].stitch()
#print(stitched)
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"max", dimension="t").pyramid.levels[0].stitch()
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"sum", dimension="t").pyramid.levels[0].stitch()
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"mean", dimension="t").pyramid.levels[0].stitch()
self.assertAlmostEqual(5.0, stitched.cells[0][0][0], delta=0.001)
stitched = imagecollection.reduce(
"variance", dimension="t").pyramid.levels[0].stitch()
self.assertAlmostEqual(0.0, stitched.cells[0][0][0], delta=0.001)
stitched = imagecollection.reduce(
"sd", dimension="t").pyramid.levels[0].stitch()
self.assertAlmostEqual(0.0, stitched.cells[0][0][0], delta=0.001)
def test_reduce(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.reduce(
"max", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(2.0, stitched.cells[0][0][1])
stitched = imagecollection.reduce(
"min", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(1.0, stitched.cells[0][0][1])
stitched = imagecollection.reduce(
"sum", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertEqual(4.0, stitched.cells[0][0][1])
stitched = imagecollection.reduce(
"mean", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(2.0, stitched.cells[0][0][0])
self.assertAlmostEqual(1.3333333, stitched.cells[0][0][1])
stitched = imagecollection.reduce(
"variance", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(0.0, stitched.cells[0][0][0])
self.assertAlmostEqual(0.2222222, stitched.cells[0][0][1])
stitched = imagecollection.reduce(
"sd", "temporal").pyramid.levels[0].stitch()
print(stitched)
self.assertEqual(0.0, stitched.cells[0][0][0])
self.assertAlmostEqual(0.4714045, stitched.cells[0][0][1])
def test_reduce_some_nodata(self):
no_data = -1.0
input = Pyramid({
0:
self._single_pixel_layer(
{
datetime.datetime.strptime("2016-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
no_data,
datetime.datetime.strptime("2017-04-24T04:00:00Z", '%Y-%m-%dT%H:%M:%SZ'):
5.0
}, no_data)
})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
stitched = imagecollection.reduce(
"min", "temporal").pyramid.levels[0].stitch()
#print(stitched)
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"max", "temporal").pyramid.levels[0].stitch()
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"sum", "temporal").pyramid.levels[0].stitch()
self.assertEqual(5.0, stitched.cells[0][0][0])
stitched = imagecollection.reduce(
"mean", "temporal").pyramid.levels[0].stitch()
self.assertAlmostEqual(5.0, stitched.cells[0][0][0], delta=0.001)
stitched = imagecollection.reduce(
"variance", "temporal").pyramid.levels[0].stitch()
self.assertAlmostEqual(0.0, stitched.cells[0][0][0], delta=0.001)
stitched = imagecollection.reduce(
"sd", "temporal").pyramid.levels[0].stitch()
self.assertAlmostEqual(0.0, stitched.cells[0][0][0], delta=0.001)
def test_reproject_spatial(self):
input = Pyramid({0: self.tiled_raster_rdd})
imagecollection = GeotrellisTimeSeriesImageCollection(
input, InMemoryServiceRegistry())
resampled = imagecollection.resample_spatial(resolution=0,
projection="EPSG:3857",
method="max")
metadata = resampled.pyramid.levels[0].layer_metadata
print(metadata)
self.assertTrue("proj=merc" in metadata.crs)
path = str(self.temp_folder / "reprojected.tiff")
resampled.reduce('max',
'temporal').download(path,
format="GTIFF",
parameters={'tiled': True})
import rasterio
with rasterio.open(path) as ds:
print(ds.profile)
def test_apply_spatiotemporal(udf_code):
udf_code = textwrap.dedent(udf_code)
input = Pyramid({0: TestMultipleDates.tiled_raster_rdd})
imagecollection = GeopysparkDataCube(
pyramid=input,
metadata=GeopysparkCubeMetadata({
"cube:dimensions": {
# TODO: also specify other dimensions?
"bands": {
"type": "bands",
"values": ["2"]
}
},
"summaries": {
"eo:bands": [{
"name": "2",
"common_name": "blue",
"wavelength_nm": 496.6,
"res_m": 10,
"scale": 0.0001,
"offset": 0,
"type": "int16",
"unit": "1"
}]
}
}))
result = imagecollection.apply_tiles_spatiotemporal(udf_code)
stitched = result.pyramid.levels[0].to_spatial_layer().stitch()
print(stitched)
assert stitched.cells[0][0][0] == 2
assert stitched.cells[0][0][5] == 6
assert stitched.cells[0][5][6] == 4
