def eocube_input_onescene(tmpdir):
dataset = sampledata.get_dataset("lsts")
layers_paths = [Path(p) for p in dataset["raster_files"][:4]]
layers_df = pd.Series([p.stem for p in layers_paths]).str.split("_", expand=True) \
.rename({0: "sceneid", 1:"band"}, axis=1)
layers_df["date"] = pd.to_datetime(layers_df.sceneid.str[9:16],
format="%Y%j")
layers_df[
"uname"] = layers_df.sceneid.str[:3] + "_" + layers_df.date.dt.strftime(
"%Y-%m-%d") + "_" + layers_df.band.str[::]
layers_df["path"] = layers_paths
layers_df = layers_df.sort_values(["date", "band"])
layers_df = layers_df.reset_index(drop=True)
a_tmpdir = tmpdir.mkdir("temp_dst_dir-onescene")
# print(a_tmpdir)
dst_paths = [
Path(a_tmpdir) / (Path(p).stem + ".vrt") for p in layers_df["path"]
]
input_kwargs = {
"layers_df": layers_df,
"tmpdir": a_tmpdir,
"dst_paths": dst_paths
}
return input_kwargs
def test_dataset_lsts():
ds = get_dataset("lsts")
assert isinstance(ds, dict)
assert len(ds["raster_files"]) == 420
assert len(ds["raster_bands"]) == 420
assert len(ds["raster_times"]) == 420
assert "b3" in ds["raster_bands"]
assert "b4" in ds["raster_bands"]
assert "b5" in ds["raster_bands"]
assert "fmask" in ds["raster_bands"]
assert "2008118" in ds["raster_times"]
assert all([os.path.exists(path) for path in ds["raster_files"]])
def test_dataset_s2l1c():
ds = get_dataset("s2l1c")
assert isinstance(ds, dict)
assert len(ds["raster_files"]) == 13
assert len(ds["raster_bands"]) == 13
assert "B02" in ds["raster_bands"]
assert len(ds["raster_times"]) == 13
assert "20170216T102101" in ds["raster_times"]
assert isinstance(ds["vector_file"], str)
assert isinstance(ds["vector_file_osm"], str)
assert all([os.path.exists(path) for path in ds["raster_files"]])
assert os.path.exists(ds["vector_file"])
assert os.path.exists(ds["vector_file_osm"])
def extraction_input_1(tmpdir):
s2l1c = get_dataset("s2l1c")
src_raster = fnmatch.filter(s2l1c["raster_files"],
"*T33UUU_20170216T102101_B11.jp2")
src_raster += fnmatch.filter(s2l1c["raster_files"],
"*T33UUU_20170216T102101_B12.jp2")
input_kwargs = {
"s2l1c": s2l1c,
"src_vector": s2l1c["vector_file"],
"burn_attribute": "pid",
"src_raster": src_raster,
"dst_names": ["B11", "B12"],
"extracted_dir": tmpdir.mkdir("temp_dst_dir-0")
}
return input_kwargs
def testdata_1(tmpdir):
ds = get_dataset('s2l1c')
tmp_procdir = tmpdir.mkdir("temp_proc_dir")
input_kwargs = {
'src_file':
ds['vector_file'],
'template_file_raster':
ds['raster_files'][0],
'interim_file_lines':
tmp_procdir / "_interim_sample_vector_dataset_lines.shp",
'interim_file_lines_raster':
tmp_procdir / "/_interim_sample_vector_dataset_lines_raster.tif",
'dst_file_proximity':
tmp_procdir / "distance_to_polygon_border__vector_dataset.tif"
}
return input_kwargs
def eocube_input_1(tmpdir):
year = 2008
dataset = sampledata.get_dataset("lsts")
layers_paths = [Path(p) for p in dataset["raster_files"]]
layers_df = pd.Series([p.stem for p in layers_paths]).str.split("_", expand=True) \
.rename({0: "sceneid", 1:"band"}, axis=1)
layers_df["date"] = pd.to_datetime(layers_df.sceneid.str[9:16],
format="%Y%j")
layers_df["uname"] = layers_df.sceneid.str[:3] + "_" + layers_df.date.dt.strftime("%Y-%m-%d") + \
"_" + layers_df.band.str[::]
layers_df["path"] = layers_paths
layers_df = layers_df.sort_values(["date", "band"])
layers_df = layers_df.reset_index(drop=True)
layers_df_year = layers_df[(layers_df.date >= str(year))
& (layers_df.date < str(year + 1))]
layers_df_year = layers_df_year.reset_index(drop=True)
input_kwargs = {
"df_layers": layers_df_year,
"tmpdir": tmpdir.mkdir("temp_dst_dir-0")
}
return input_kwargs
import pytest
import fnmatch
import os
from pathlib import Path
import numpy as np
from eobox.sampledata import get_dataset
from eobox.raster import MultiRasterIO
dataset = get_dataset("s2l1c")
layer_files = fnmatch.filter(dataset["raster_files"], "*B04*")
layer_files += fnmatch.filter(dataset["raster_files"], "*B08*")
layer_files += fnmatch.filter(dataset["raster_files"], "*B8A*")
def process_ndvi(arr, idx_b04, idx_b08, idx_b8a):
"""A custom function for calculating an NDVI based on the two S2 NIR bands."""
import numpy as np
def normalized_difference_index(arr, idx_minuend, idx_subtrahend):
di = (arr[:, :, idx_minuend] - arr[:, :, idx_subtrahend]) / (
arr[:, :, idx_minuend] + arr[:, :, idx_subtrahend])
di[di > 1.0] = 1.0001
di[di < -1.0] = -1.0001
di = (di * 10000).astype(np.int16)
return di
ndvi_b08 = normalized_difference_index(arr, idx_b08, idx_b04)
ndvi_b8a = normalized_difference_index(arr, idx_b8a, idx_b04)
return [ndvi_b08, ndvi_b8a]