def test_fail_1_to_1(self):
code = """import numpy as np
def ndvi_numpy(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
out_ar[:] = in_ar[0]"""
landsat = Landsat(self.metadata_set)
scale_params = [[0, 40000], [0, 40000], [0, 40000]]
pixel_function_details = FunctionDetails(name="ndvi_numpy",
band_definitions=[Band.RED],
code=code,
arguments={"factor": DataType.UINT16.range_max},
data_type=DataType.UINT16)
gdal.SetConfigOption('GDAL_VRT_ENABLE_PYTHON', "YES")
nda = landsat.fetch_imagery_array([pixel_function_details, Band.GREEN, Band.BLUE],
scale_params=scale_params,
polygon_boundary_wkb=self.taos_shape.wkb,
output_type=DataType.BYTE)
nda2 = landsat.fetch_imagery_array([Band.RED, Band.GREEN, Band.BLUE],
scale_params=scale_params,
polygon_boundary_wkb=self.taos_shape.wkb,
output_type=DataType.BYTE)
self.assertIsNotNone(nda)
np.testing.assert_almost_equal(nda, nda2)
np.testing.assert_equal(nda, nda2)
def test_aws_without_google(self):
metadata_service = MetadataService()
# c1/L8/083/015/LC08_L1TP_083015_20171106_20171107_01_RT/
# c1/L8/083/111/LC08_L1GT_083111_20171106_20171107_01_RT
sqs_message = {
"Type": "Notification",
"MessageId": "27f57c3d-9d2e-5fa3-8f83-2e41a3aa5634",
"TopicArn": "arn:aws:sns:us-west-2:274514004127:NewSceneHTML",
"Subject": "Amazon S3 Notification",
"Message": "{\"Records\":[{\"eventVersion\":\"2.0\",\"eventSource\":\"aws:s3\",\"awsRegion\":\"us-west-2\",\"eventTime\":\"2017-11-07T23:05:40.162Z\",\"eventName\":\"ObjectCreated:Put\",\"userIdentity\":{\"principalId\":\"AWS:AIDAILHHXPNIKSGVUGOZK\"},\"requestParameters\":{\"sourceIPAddress\":\"35.193.238.175\"},\"responseElements\":{\"x-amz-request-id\":\"F96A6CC9816FC5EF\",\"x-amz-id-2\":\"yehs3XxTY8utc9kgKfNbMe1wdtV7F0wEMUXUQtIu7zMRtGvboxahzwncrmG046yI327j5IRh8nE=\"},\"s3\":{\"s3SchemaVersion\":\"1.0\",\"configurationId\":\"C1-NewHTML\",\"bucket\":{\"name\":\"landsat-pds\",\"ownerIdentity\":{\"principalId\":\"A3LZTVCZQ87CNW\"},\"arn\":\"arn:aws:s3:::landsat-pds\"},\"object\":{\"key\":\"c1/L8/115/062/LC08_L1TP_115062_20171107_20171107_01_RT/index.html\",\"size\":5391,\"eTag\":\"0f06667fca1f707894bf579bd667e221\",\"sequencer\":\"005A023C441A8F8403\"}}}]}",
"Timestamp": "2017-11-07T23:05:40.219Z",
"SignatureVersion": "1",
"Signature": "bqrW1x6CgJntCz6f0F5uncyPZR+6ZM/tZ3OrRZDiudBv5DAtMyYR9n6KQ0aT+iYP5INfpL2GuIm8Uqco8ZHzg5AqEhHtNkpzGBQpQHvlF3t0ut9K27YNwJ6ZmnS14BgsLWyXIthVRjvHf1Hhx3ZInPMJrzTcKCOhOmBcM9zOpfWrHfnynuifpN3FaldDz6VY2d9QM0Rn8Fo8XZ4F+j01eAJVlydnRbSBbLewleuvhPQh6EG5r2EeekeniOIETrodS7o43ZClFr8OSgRE7BvpecVnnUEXBUIDDtRAPnIxo3Io0AmfPRI8xRfeKNhBIhPq3W3clm7Dxkp3N96OKoVUBw==",
"SigningCertURL": "https://sns.us-west-2.amazonaws.com/SimpleNotificationService-433026a4050d206028891664da859041.pem",
"UnsubscribeURL": "https://sns.us-west-2.amazonaws.com/?Action=Unsubscribe&SubscriptionArn=arn:aws:sns:us-west-2:274514004127:NewSceneHTML:7997d757-d1c6-4064-8935-34111968c8cc"
}
d = datetime.datetime.strptime(sqs_message['Timestamp'], "%Y-%m-%dT%H:%M:%S.%fZ")
rows = metadata_service.search_aws('/imagery', wrs_path=115, wrs_row=62, collection_date=d)
self.assertGreater(len(rows), 0)
metadata = rows[0]
self.assertIsNotNone(metadata)
landsat = Landsat(metadata)
nda = landsat.fetch_imagery_array([4, 3, 2], [[0, 40000], [0, 40000], [0, 40000]], spatial_resolution_m=240)
self.assertIsNotNone(nda)
def test_australia(self):
# 5th Place: Lake Eyre Landsat 5 Acquired August 5, 2006
wkt = "POLYGON((136.2469482421875 -27.57843813308233,138.6639404296875 -27.57843813308233," \
"138.6639404296875 -29.82351878748485,136.2469482421875 -29.82351878748485,136." \
"2469482421875 -27.57843813308233))"
polygon = loads(wkt)
# sql_filters = ['cloud_cover=0']
d_start = date(2006, 8, 4)
d_end = date(2006, 8, 7)
bounding_box = polygon.bounds
landsat_filters = LandsatQueryFilters()
landsat_filters.wrs_row.set_value(79)
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(
SpacecraftID.LANDSAT_5,
data_filters=landsat_filters)
rows = list(rows)
self.assertEqual(len(rows), 1)
metadata = rows[0]
landsat = Landsat(metadata)
# get a numpy.ndarray from bands for specified imagery
band_numbers = [3, 2, 1]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
# nda = landsat.__get_ndarray(band_numbers, metadata, scale_params)
nda = landsat.fetch_imagery_array(band_numbers, scale_params)
self.assertEqual((3581, 4046, 3), nda.shape)
def test_polygon_wkb_metadata(self):
d_start = date(2017, 3, 12) # 2017-03-12
d_end = date(2017, 3, 19) # 2017-03-20, epl api is inclusive
self.metadata_service = MetadataService()
landsat_filters = LandsatQueryFilters()
landsat_filters.collection_number.set_value("PRE")
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_geometry(self.taos_shape.wkb)
# landsat_filters.geometry_bag.geometry_binaries.append(self.taos_shape.wkb)
metadata_rows = self.metadata_service.search(
SpacecraftID.LANDSAT_8, limit=10, data_filters=landsat_filters)
metadata_set = []
for row in metadata_rows:
metadata_set.append(row)
landsat = Landsat(metadata_set)
band_numbers = [Band.NIR, Band.SWIR1, Band.SWIR2]
scaleParams = [[0.0, 40000.0], [0.0, 40000.0], [0.0, 40000.0]]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
polygon_boundary_wkb=self.taos_shape.wkb)
self.assertIsNotNone(nda)
self.assertEqual((1804, 1295, 3), nda.shape)
def test_rastermetadata_cache(self):
# GDAL helper functions for generating VRT
landsat = Landsat(self.metadata_set)
# get a numpy.ndarray from bands for specified imagery
# 'nir', 'swir1', 'swir2'
band_numbers = [Band.NIR, Band.SWIR1, Band.SWIR2]
scaleParams = [[0.0, 40000.0], [0.0, 40000.0], [0.0, 40000.0]]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
polygon_boundary_wkb=self.taos_shape.wkb,
spatial_resolution_m=120)
self.assertIsNotNone(nda)
self.assertEqual((902, 648, 3), nda.shape)
band_numbers = [Band.RED, Band.BLUE, Band.GREEN]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
polygon_boundary_wkb=self.taos_shape.wkb,
spatial_resolution_m=120)
self.assertIsNotNone(nda)
self.assertEqual((902, 648, 3), nda.shape)
band_numbers = [Band.RED, Band.BLUE, Band.GREEN]
nda = landsat.fetch_imagery_array(band_numbers,
scaleParams,
spatial_resolution_m=120)
self.assertIsNotNone(nda)
self.assertNotEqual((902, 648, 3), nda.shape)
def test_ndvi_taos(self):
code = """import numpy as np
def ndvi_numpy(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
with np.errstate(divide = 'ignore', invalid = 'ignore'):
factor = float(kwargs['factor'])
output = np.divide((in_ar[1] - in_ar[0]), (in_ar[1] + in_ar[0]))
output[np.isnan(output)] = 0.0
# shift range from -1.0-1.0 to 0.0-2.0
output += 1.0
# scale up from 0.0-2.0 to 0 to 255 by multiplying by 255/2
# https://stackoverflow.com/a/1735122/445372
output *= factor/2.0
# https://stackoverflow.com/a/10622758/445372
# in place type conversion
out_ar[:] = output.astype(np.int16, copy=False)"""
code2 = """import numpy as np
def ndvi_numpy2(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
with np.errstate(divide = 'ignore', invalid = 'ignore'):
output = (in_ar[1] - in_ar[0]) / (in_ar[1] + in_ar[0])
output[np.isnan(output)] = 0.0
out_ar[:] = output"""
landsat = Landsat(self.metadata_set)
scale_params = [[0, DataType.UINT16.range_max, -1.0, 1.0]]
pixel_function_details = FunctionDetails(name="ndvi_numpy",
band_definitions=[Band.RED, Band.NIR],
code=code,
arguments={"factor": DataType.UINT16.range_max},
data_type=DataType.UINT16)
gdal.SetConfigOption('GDAL_VRT_ENABLE_PYTHON', "YES")
nda = landsat.fetch_imagery_array([pixel_function_details],
scale_params=scale_params,
polygon_boundary_wkb=self.taos_shape.wkb,
output_type=DataType.FLOAT32)
self.assertIsNotNone(nda)
self.assertGreaterEqual(1.0, nda.max())
self.assertLessEqual(-1.0, nda.min())
pixel_function_details = FunctionDetails(name="ndvi_numpy2",
band_definitions=[Band.RED, Band.NIR],
code=code2,
data_type=DataType.FLOAT32)
nda2 = landsat.fetch_imagery_array([pixel_function_details],
polygon_boundary_wkb=self.taos_shape.wkb,
output_type=DataType.FLOAT32)
self.assertIsNotNone(nda2)
self.assertGreaterEqual(1.0, nda2.max())
self.assertLessEqual(-1.0, nda2.min())
def test_iowa_ndarray(self):
d_start = date(2016, 4, 4)
d_end = date(2016, 8, 7)
bounding_box = self.iowa_polygon.bounds
# sql_filters = ["cloud_cover<=15"]
landsat_filters = LandsatQueryFilters()
landsat_filters.cloud_cover.set_range(end=15, end_inclusive=15)
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(
SpacecraftID.LANDSAT_8,
data_filters=landsat_filters)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
code = """import numpy as np
def ndvi_numpy(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
with np.errstate(divide = 'ignore', invalid = 'ignore'):
out_ar[:] = np.divide((in_ar[1] - in_ar[0]), (in_ar[1] + in_ar[0]))
out_ar[np.isnan(out_ar)] = 0.0
out_ar """
pixel_function_details = FunctionDetails(name="ndvi_numpy", band_definitions=[4, 5], code=code, data_type=DataType.FLOAT32)
# pixel_function_details = {
# "band_numbers": [4, 5],
# "function_code": code,
# "function_type": "ndvi_numpy",
# "data_type": DataType.FLOAT32,
# }
band_definitions = [pixel_function_details, 4, 5]
vrt = landsat.get_vrt(band_definitions)
ds = gdal.Open(vrt)
self.assertIsNotNone(ds)
arr_4 = ds.GetRasterBand(2).ReadAsArray()
arr_5 = ds.GetRasterBand(3).ReadAsArray()
arr_ndvi = ds.GetRasterBand(1).ReadAsArray()
del ds
del landsat
print(np.ndarray.max(arr_ndvi))
print(np.ndarray.min(arr_ndvi))
self.assertFalse(np.any(np.isinf(arr_ndvi)))
self.assertIsNotNone(arr_ndvi)
local_ndvi = self.ndvi_numpy(arr_5, arr_4)
del arr_4
del arr_5
self.assertFalse(np.any(np.isinf(local_ndvi)))
np.testing.assert_almost_equal(arr_ndvi, local_ndvi)
def test_vrt_extent(self):
# GDAL helper functions for generating VRT
landsat = Landsat(self.metadata_set[0])
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
vrt = landsat.get_vrt(band_numbers,
envelope_boundary=self.taos_shape.bounds)
self.assertIsNotNone(vrt)
def test_one_band(self):
# specify the bands that approximate real color
landsat = Landsat(self.metadata_set)
band_numbers = [Band.RED]
scaleParams = [[0.0, 40000.0]]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
polygon_boundary_wkb=self.taos_shape.wkb,
spatial_resolution_m=120)
self.assertIsNotNone(nda)
self.assertEqual((902, 648), nda.shape)
def test_native_vs_custom(self):
landsat = Landsat(self.metadata_set)
gdal.SetConfigOption('GDAL_VRT_ENABLE_PYTHON', "YES")
pixel_native = FunctionDetails(name="sqrt",
band_definitions=[Band.RED],
data_type=DataType.UINT16,
transfer_type=DataType.FLOAT32)
nda = landsat.fetch_imagery_array([pixel_native],
polygon_boundary_wkb=self.taos_shape.wkb,
output_type=DataType.FLOAT32)
self.assertIsNotNone(nda)
def test_cutline(self):
# GDAL helper functions for generating VRT
landsat = Landsat(self.metadata_set[0])
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
nda = landsat.fetch_imagery_array(band_numbers,
scale_params,
self.taos_shape.wkb,
spatial_resolution_m=480)
self.assertIsNotNone(nda)
def test_mosaic(self):
# GDAL helper functions for generating VRT
landsat = Landsat(self.metadata_set)
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
nda = landsat.fetch_imagery_array(
band_numbers,
scale_params,
envelope_boundary=self.taos_shape.bounds)
self.assertIsNotNone(nda)
self.assertEqual((1804, 1295, 3), nda.shape)
def test_mosaic_cutline(self):
# GDAL helper functions for generating VRT
landsat = Landsat(self.metadata_set)
# get a numpy.ndarray from bands for specified imagery
# 'nir', 'swir1', 'swir2'
band_numbers = [Band.NIR, Band.SWIR1, Band.SWIR2]
scaleParams = [[0.0, 40000.0], [0.0, 40000.0], [0.0, 40000.0]]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
polygon_boundary_wkb=self.taos_shape.wkb)
self.assertIsNotNone(nda)
self.assertEqual((1804, 1295, 3), nda.shape)
def test_pixel_function_vrt_1(self):
utah_box = (-112.66342163085938, 37.738141282210385,
-111.79824829101562, 38.44821130413263)
d_start = date(2016, 7, 20)
d_end = date(2016, 7, 28)
landsat_filter = LandsatQueryFilters()
landsat_filter.collection_number.set_value("PRE")
landsat_filter.cloud_cover.set_range(
end=5, end_inclusive=True
) #landsat_filter.cloud_cover.set_range_end(5, True)
landsat_filter.acquired.set_range(d_start, True, d_end, True)
landsat_filter.aoi.set_bounds(*utah_box)
rows = self.metadata_service.search(SpacecraftID.LANDSAT_8,
limit=10,
data_filters=landsat_filter)
rows = list(rows)
self.assertEqual(0, len(rows))
d_end = date(2016, 8, 28)
landsat_filter = LandsatQueryFilters()
landsat_filter.collection_number.set_value("PRE")
landsat_filter.cloud_cover.set_range(
end=5, end_inclusive=False
) #landsat_filter.cloud_cover.set_range_end(5, False)
landsat_filter.acquired.set_range(
end=d_end, end_inclusive=True
) #landsat_filter.acquired.set_range_end(d_end, True)
landsat_filter.acquired.sort_by(epl_imagery_pb2.DESCENDING)
landsat_filter.aoi.set_bounds(*utah_box)
rows = self.metadata_service.search(SpacecraftID.LANDSAT_8,
limit=10,
data_filters=landsat_filter)
rows = list(rows)
self.assertEqual(len(rows), 10)
# metadata_row = ['LC80390332016208LGN00', '', 'LANDSAT_8', 'OLI_TIRS', '2016-07-26',
# '2016-07-26T18:14:46.9465460Z', 'PRE', 'N/A', 'L1T', 39, 33, 1.69,
# 39.96962, 37.81744, -115.27267, -112.56732, 1070517542,
# 'gs://gcp-public-data-landsat/LC08/PRE/039/033/LC80390332016208LGN00']
metadata = rows[0]
# GDAL helper functions for generating VRT
landsat = Landsat([metadata])
# get a numpy.ndarray from bands for specified imagery
band_numbers = [4, 3, 2]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
nda = landsat.fetch_imagery_array(band_numbers, scale_params)
self.assertEqual(nda.shape, (3876, 3806, 3))
def test_vrt_with_alpha(self):
landsat = Landsat(self.metadata_set)
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE, Band.ALPHA]
scaleParams = [[0.0, 40000], [0.0, 40000], [0.0, 40000]]
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
envelope_boundary=self.taos_shape.bounds,
output_type=DataType.UINT16,
spatial_resolution_m=120)
self.assertIsNotNone(nda)
def test_pixel_ndvi(self):
"""
http://grindgis.com/blog/vegetation-indices-arcgis
NDVI = (NIR - RED) / (NIR + RED)
NDVI = (5 - 4) / (5 + 4)
:return:
"""
landsat = Landsat(self.m_metadata) # , gsurl[2])
code = """import numpy as np
def ndvi_numpy(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
with np.errstate(divide = 'ignore', invalid = 'ignore'):
output = np.divide((in_ar[1] - in_ar[0]), (in_ar[1] + in_ar[0]))
output[np.isnan(output)] = 0.0
# shift range from -1.0-1.0 to 0.0-2.0
output += 1.0
# scale up from 0.0-2.0 to 0 to 255 by multiplying by 255/2
# https://stackoverflow.com/a/1735122/445372
output *= 65535/2.0
# https://stackoverflow.com/a/10622758/445372
# in place type conversion
out_ar[:] = output.astype(np.int16, copy=False)"""
pixel_function_details = FunctionDetails(name="ndvi_numpy", band_definitions=[4, 5],
data_type=DataType.UINT16, code=code)
vrt = landsat.get_vrt([pixel_function_details, 3, 2])
with open('ndvi_numpy.vrt', 'r') as myfile:
data = myfile.read()
expected = etree.XML(data)
actual = etree.XML(vrt)
result, message = xml_compare(expected, actual, {"GeoTransform": 1e-10})
self.assertTrue(result, message)
gdal.SetConfigOption('GDAL_VRT_ENABLE_PYTHON', "YES")
ds = gdal.Open(vrt)
self.assertIsNotNone(ds)
arr_ndvi = ds.GetRasterBand(1).ReadAsArray()
ds = None
self.assertIsNotNone(arr_ndvi)
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
band_definitions = [pixel_function_details, 3, 2]
nda = landsat.fetch_imagery_array(band_definitions, scale_params)
self.assertIsNotNone(nda)
def test_get_file(self):
d_start = date(2015, 6, 24)
d_end = date(2016, 6, 24)
bounding_box = (-115.927734375, 34.52466147177172, -78.31054687499999, 44.84029065139799)
landsat_filters = LandsatQueryFilters()
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
metadata_service = MetadataService()
rows = metadata_service.search(SpacecraftID.LANDSAT_8,
data_filters=landsat_filters,
limit=1)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
self.assertIsNotNone(landsat)
vrt = landsat.get_vrt([4, 3, 2])
self.assertIsNotNone(vrt)
dataset = landsat.get_dataset([4, 3, 2], DataType.UINT16)
self.assertIsNotNone(dataset)
def test_metadata_extent(self):
r = requests.get("https://raw.githubusercontent.com/johan/world.geo.json/master/countries/USA/NM/Taos.geo.json")
taos_geom = r.json()
print(taos_geom)
taos_shape = shapely.geometry.shape(taos_geom['features'][0]['geometry'])
metadata_service = MetadataService()
# sql_filters = ['scene_id="LC80330342017072LGN00"', 'collection_number="PRE"']
landsat_filters = LandsatQueryFilters()
landsat_filters.scene_id.set_value("LC80330342017072LGN00")
landsat_filters.collection_number.set_value("PRE")
rows = metadata_service.search(
SpacecraftID.LANDSAT_8,
data_filters=landsat_filters)
rows = list(rows)
self.assertEqual(len(rows), 1)
metadata = rows[0]
# GDAL helper functions for generating VRT
landsat = Landsat(metadata)
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE]
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
vrt = landsat.get_vrt(band_numbers, envelope_boundary=taos_shape.bounds)
with open('clipped_LC80330342017072LGN00.vrt', 'r') as myfile:
data = myfile.read()
expected = etree.XML(data)
actual = etree.XML(vrt)
result, message = xml_compare(expected, actual, {"GeoTransform": 1e-10, "xOff": 1e-10, "yOff": 1e-10})
self.assertTrue(result, message)
dataset = gdal.Open(vrt)
geo_transform = dataset.GetGeoTransform()
# self.assertEqual(geo_transform, raster_metadata.get_geotransform(taos_shape.bounds))
# self.assertNotEqual(geo_transform, raster_metadata.get_geotransform())
"""
def test_datatypes(self):
landsat = Landsat(self.metadata_set)
# get a numpy.ndarray from bands for specified imagery
band_numbers = [Band.RED, Band.GREEN, Band.BLUE]
scaleParams = [[0.0, 40000], [0.0, 40000], [0.0, 40000]]
for data_type in DataType:
if data_type is DataType.UNKNOWN_GDAL:
continue
nda = landsat.fetch_imagery_array(
band_numbers,
scaleParams,
envelope_boundary=self.taos_shape.bounds,
output_type=data_type,
spatial_resolution_m=240)
self.assertIsNotNone(nda)
self.assertGreaterEqual(data_type.range_max, nda.max())
self.assertLessEqual(data_type.range_min, nda.min())
def test_unmount_destructor_conflict(self):
wkt = "POLYGON((136.2469482421875 -27.57843813308233,138.6639404296875 -27.57843813308233," \
"138.6639404296875 -29.82351878748485,136.2469482421875 -29.82351878748485,136." \
"2469482421875 -27.57843813308233))"
polygon = loads(wkt)
# sql_filters = ['cloud_cover=0']
d_start = date(2006, 8, 4)
d_end = date(2006, 8, 7)
bounding_box = polygon.bounds
# sql_filters = ['wrs_row=79']
landsat_filters = LandsatQueryFilters()
landsat_filters.wrs_row.set_value(79)
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(
SpacecraftID.LANDSAT_5,
data_filters=landsat_filters)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
vrt = landsat.get_vrt([4])
storage = Storage("gcp-public-data-landsat")
landsat_2 = Landsat(metadata)
vrt = landsat_2.get_vrt([4])
del landsat
self.assertTrue(storage.is_mounted(metadata))
def test_vrt_not_pre(self):
d_start = date(2017, 6, 24)
d_end = date(2017, 9, 24)
bounding_box = (-115.927734375, 34.52466147177172, -78.31054687499999,
44.84029065139799)
# sql_filters = ['collection_number!="PRE"']
landsat_filter = LandsatQueryFilters()
landsat_filter.collection_number.set_exclude_value("PRE")
landsat_filter.acquired.set_range(d_start, True, d_end, True)
landsat_filter.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(SpacecraftID.LANDSAT_8,
limit=1,
data_filters=landsat_filter)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
self.assertIsNotNone(landsat)
vrt = landsat.get_vrt([4, 3, 2])
self.assertIsNotNone(vrt)
dataset = landsat.get_dataset([4, 3, 2], DataType.UINT16)
self.assertIsNotNone(dataset)
def test_band_enum(self):
self.assertTrue(True)
d_start = date(2016, 7, 20)
d_end = date(2016, 7, 28)
landsat_filter = LandsatQueryFilters()
landsat_filter.scene_id.set_value("LC80390332016208LGN00")
landsat_filter.acquired.set_range(d_start, True, d_end, True)
rows = self.metadata_service.search(SpacecraftID.LANDSAT_8,
limit=1,
data_filters=landsat_filter)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
scale_params = [[0.0, 65535], [0.0, 65535], [0.0, 65535]]
# nda = landsat.__get_ndarray(band_numbers, metadata, scale_params)
nda = landsat.fetch_imagery_array([Band.RED, Band.GREEN, Band.BLUE],
scale_params,
spatial_resolution_m=240)
self.assertIsNotNone(nda)
nda2 = landsat.fetch_imagery_array([4, 3, 2],
scale_params,
spatial_resolution_m=240)
np.testing.assert_almost_equal(nda, nda2)
def test_pixel_1(self):
metadata = self.m_metadata
landsat = Landsat(metadata) # , gsurl[2])
code = """import numpy as np
def multiply_rounded(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize,
raster_ysize, buf_radius, gt, **kwargs):
factor = float(kwargs['factor'])
out_ar[:] = np.round_(np.clip(in_ar[0] * factor,0,255))"""
function_arguments = {"factor": "1.5"}
pixel_function_details = FunctionDetails(name="multiply_rounded", band_definitions=[2],
data_type=DataType.FLOAT32, code=code,
arguments=function_arguments)
vrt = landsat.get_vrt([pixel_function_details, 3, 2])
with open('pixel_1.vrt', 'r') as myfile:
data = myfile.read()
expected = etree.XML(data)
actual = etree.XML(vrt)
result, message = xml_compare(expected, actual, {"GeoTransform": 1e-10})
self.assertTrue(result, message)
def test_landsat5_vrt(self):
# 5th Place: Lake Eyre Landsat 5 Acquired August 5, 2006
wkt = "POLYGON((136.2469482421875 -27.57843813308233,138.6639404296875 -27.57843813308233," \
"138.6639404296875 -29.82351878748485,136.2469482421875 -29.82351878748485,136." \
"2469482421875 -27.57843813308233))"
polygon = loads(wkt)
# sql_filters = ['cloud_cover=0']
d_start = date(2006, 8, 4)
d_end = date(2006, 8, 5)
bounding_box = polygon.bounds
landsat_filters = LandsatQueryFilters()
landsat_filters.wrs_row.set_value(79)
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(
SpacecraftID.LANDSAT_5,
data_filters=landsat_filters)
rows = list(rows)
self.assertEqual(len(rows), 1)
# data structure that contains all fields from Google's Landsat BigQuery Database
metadata = rows[0]
# GDAL helper functions for generating VRT
landsat = Landsat(metadata)
vrt = landsat.get_vrt([3, 2, 1])
with open('testlandsat5.vrt', 'r') as myfile:
data = myfile.read()
expected = etree.XML(data)
actual = etree.XML(vrt)
result, message = xml_compare(expected, actual, {"GeoTransform": 1e-10})
self.assertTrue(result, message)
def test_vrt(self):
d_start = date(2015, 6, 24)
d_end = date(2016, 6, 24)
bounding_box = (-115.927734375, 34.52466147177172, -78.31054687499999, 44.84029065139799)
# sql_filters = ['scene_id="LC80400312016103LGN00"']
landsat_filters = LandsatQueryFilters()
landsat_filters.scene_id.set_value("LC80400312016103LGN00")
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(SpacecraftID.LANDSAT_8,
limit=1,
data_filters=landsat_filters)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
vrt = landsat.get_vrt([4, 3, 2])
with open('test_1.vrt', 'r') as myfile:
data = myfile.read()
expected = etree.XML(data)
actual = etree.XML(vrt)
result, message = xml_compare(expected, actual)
self.assertTrue(result, message)
dataset = gdal.Open(vrt)
self.assertIsNotNone(dataset)
ds_band_1 = dataset.GetRasterBand(1)
self.assertIsNotNone(ds_band_1)
self.assertEqual(ds_band_1.XSize, 7631)
ds_band_2 = dataset.GetRasterBand(2)
self.assertIsNotNone(ds_band_2)
self.assertEqual(ds_band_2.YSize, 7771)
ds_band_3 = dataset.GetRasterBand(3)
self.assertIsNotNone(ds_band_3)
self.assertEqual(ds_band_3.YSize, 7771)
def test_iowa_scaled(self):
d_start = date(2016, 4, 4)
d_end = date(2016, 8, 7)
bounding_box = self.iowa_polygon.bounds
# sql_filters = ["cloud_cover<=15"]
landsat_filters = LandsatQueryFilters()
landsat_filters.cloud_cover.set_range(end=15, end_inclusive=True)
landsat_filters.acquired.set_range(d_start, True, d_end, True)
landsat_filters.aoi.set_bounds(*bounding_box)
rows = self.metadata_service.search(
SpacecraftID.LANDSAT_8,
data_filters=landsat_filters)
rows = list(rows)
metadata = rows[0]
landsat = Landsat(metadata)
code = """import numpy as np
def ndvi_numpy(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize, raster_ysize, buf_radius, gt, **kwargs):
with np.errstate(divide = 'ignore', invalid = 'ignore'):
factor = float(kwargs['factor'])
out_ar[:] = np.divide((in_ar[1] - in_ar[0]), (in_ar[1] + in_ar[0]))
out_ar[np.isnan(out_ar)] = 0.0
# shift range from -1.0-1.0 to 0.0-2.0
out_ar += 1.0
# scale up from 0.0-2.0 to 0 to 255 by multiplying by 255/2
out_ar *= factor/2.0"""
# pixel_function_details = {
# "function_arguments": {"factor": 255},
# "band_numbers": [4, Band.NIR],
# "function_code": code,
# "function_type": "ndvi_numpy",
# "data_type": DataType.FLOAT32,
# }
pixel_function_details = FunctionDetails(name="ndvi_numpy",
band_definitions=[4, Band.NIR],
code=code, arguments={"factor": 255},
data_type=DataType.FLOAT32)
band_definitions = [pixel_function_details, Band.RED, 5]
vrt = landsat.get_vrt(band_definitions)
ds = gdal.Open(vrt)
self.assertIsNotNone(ds)
arr_4 = ds.GetRasterBand(2).ReadAsArray()
arr_5 = ds.GetRasterBand(3).ReadAsArray()
arr_ndvi = ds.GetRasterBand(1).ReadAsArray()
del ds
del landsat
print(np.ndarray.max(arr_ndvi))
# print(np.ndarray.min(arr_ndvi))
# self.assertFalse(np.any(np.isinf(arr_ndvi)))
self.assertIsNotNone(arr_ndvi)
local_ndvi = self.ndvi_numpy(arr_5, arr_4)
del arr_4
del arr_5
local_ndvi += 1.0
local_ndvi *= float(pixel_function_details.arguments['factor']) / 2.0
self.assertFalse(np.any(np.isinf(local_ndvi)))
np.floor(arr_ndvi, out=arr_ndvi)
np.floor(local_ndvi, out=local_ndvi)
np.testing.assert_almost_equal(arr_ndvi, local_ndvi, decimal=0)