def Sentinel2_3a_2018(return_dates=False, return_random_sample=False):
"""
Sentinel2 sample dataset on Bouconne Forest (France, near Toulouse).
Bands are ordered this way : '2','3','4','8','5','6','7','8','8A','11','12'.
Parameters
-----------
return_dates : bool, default False
If True, will return list of dates.
return_random_sample : bool, default False
If False, will return the path of the raster
If True, will return a random block of the image
Returns
-------
raster : str or array
If get_only_sample is False, returns path of raster.
If get_only_sample is True, returns array where each line is a pixel.
dates : list
List of integer
Examples
--------
>>> raster,dates = Sentinel2_3a_2018(return_dates=True)
>>> print(raster)
/mnt/bigone/lib/MuseoPheno/museopheno/datasets/2018_3A_Theia_Bouconne.tif
>>> print(dates)
[20180429, 20180513, 20180708, 20180815, 20180915, 20181015, 20181115]
>>> Sentinel2_3a_2018(get_only_sample=True)
Total number of blocks : 246
array([[ 122, 320, 109, ..., 2107, 1530, 751],
[ 140, 370, 122, ..., 2107, 1530, 751],
[ 148, 388, 117, ..., 2102, 1557, 761],
...,
[ 167, 459, 195, ..., 2251, 1482, 664],
[ 154, 470, 185, ..., 2251, 1482, 664],
[ 184, 494, 213, ..., 2429, 1507, 670]], dtype=int16)
References
-----------
This dataset is built using level 3A Sentinel-2 month syntheses :
https://labo.obs-mip.fr/multitemp/sentinel-2-level-3a-products-syntheses-of-composites/
"""
raster = os.path.join(__pathFile, '2018_3A_Theia_Bouconne.tif')
dates = [
20180429, 20180513, 20180708, 20180815, 20180915, 20181015, 20181115
]
if return_random_sample is True:
from museotoolbox.processing import RasterMath
raster = RasterMath(raster).get_random_block()
if return_dates:
return raster, dates
else:
return raster
# ------------------------------------------- from museotoolbox.processing import RasterMath from museotoolbox import datasets import numpy as np ############################################################################## # Load HistoricalMap dataset # ------------------------------------------- raster, vector = datasets.load_historical_data() ############################################################################## # Initialize rasterMath with raster # ------------------------------------ rM = RasterMath(raster) print(rM.get_random_block()) ########################## # Let's suppose you want compute the difference between blue and green band. # I suggest you to define type in numpy array to save space while creating the raster! X = rM.get_random_block() sub = lambda X: np.array((X[:, 0] - X[:, 1])).astype(np.int16) rM.add_function(sub, out_image='/tmp/sub_lambda.tif') ########################################################### # Use a python function to use arguments # ----------------------------------------
X1 = S2.generate_index(X,
S2.get_index_expression('LChloC'),
multiply_by=100,
dtype=np.int16)
X2 = S2.generate_index(X,
S2.get_index_expression('ACORVI'),
multiply_by=100,
dtype=np.int16)
return np.hstack((X, X1, X2)).astype(np.int16)
#########################################
# Use rasterMath to read and write block per block the raster according to a function
rM = RasterMath(raster)
X = rM.get_random_block()
print('Block has {} pixels and {} bands'.format(X.shape[0], X.shape[1]))
################################
# Now we can try our function
XwithIndex = createSITSplusIndex(X)
print('Raster+index produced has {} pixels and {} bands'.format(
XwithIndex.shape[0], XwithIndex.shape[1]))
##################################
# Now we add our function as the test was a success
rM.add_function(createSITSplusIndex, '/tmp/SITSwithIndex.tif')
def __init__(self, in_image, in_image_mask=False,
lag=1, weights=False):
"""
Compute Moran's I for raster.
Parameters
----------
in_image : str.
A filename or path of a raster file.
It could be any file that GDAL can open.
lag : int, optional (default=1)
The distance from the cell.
weights :False or array-like, optional (default=False).
Weights with the same shape as the square size.
"""
self.scores = dict(lag=[], I=[], band=[], EI=[])
self.lags = []
rM = RasterMath(
in_image,
in_image_mask=in_image_mask,
return_3d=True,
verbose=False)
for band, arr in enumerate(rM.read_band_per_band()):
if isinstance(lag, int):
lag = [lag]
for l in lag:
squareSize = l * 2 + 1
footprint = np.ones((squareSize, squareSize))
weights = np.zeros((footprint.shape[0], footprint.shape[1]))
weights[:, 0] = 1
weights[0, :] = 1
weights[-1, :] = 1
weights[:, -1] = 1
n = arr.count()
arr = arr.astype(np.float64)
# convert masked to nan for generic_filter
np.where(arr.mask, np.nan, arr.data)
x_ = np.nanmean(arr)
num = generic_filter(
arr,
self._compute_view_for_global_moran,
footprint=footprint,
mode='constant',
cval=np.nan,
extra_keywords=dict(
x_=x_,
footprint=footprint,
weights=weights,
transform='r'))
den = (arr - x_)**2
self.z = arr - x_
# need to mask z/den/num/neighbors
den[arr.mask] = np.nan
# local = np.nansum(num) / np.nansum(den)
l1 = np.where(arr.mask, np.nan, num)
l2 = np.where(arr.mask, np.nan, den)
local = np.nansum(l1) / np.nansum(l2)
self.I = local
self.EI = -1 / (n - 1)
self.lags.append(l)
self.scores['lag'].append(l)
self.scores['band'].append(band + 1)
self.scores['I'].append(self.I)
self.scores['EI'].append(self.EI)
##############################################################################
# Load HistoricalMap dataset
# -------------------------------------------
raster, vector = datasets.load_historical_data()
##############################################################################
# Initialize rasterMath with raster
# ------------------------------------
# Set return_3d to True to have full block size (not one pixel per row)
# Create raster mask to only keep pixel inside polygons.
image_mask_from_vector(vector, raster, '/tmp/mask.tif', invert=False)
rM = RasterMath(raster, in_image_mask='/tmp/mask.tif', return_3d=True)
#rM.addInputRaster('/tmp/mask.tif')
print(rM.get_random_block().shape)
#######################
# Plot blocks
x = rM.get_random_block()
rM.add_function(np.mean, '/tmp/mean.tif', axis=2, out_np_dt=np.int16)
rM.run()
from osgeo import gdal
dst = gdal.Open('/tmp/mean.tif')
arr = dst.GetRasterBand(1).ReadAsArray()
plt.imshow(np.ma.masked_where(arr == rM._outputs[0]['nodata'], arr))
print(S2.available_indices.keys())
###########################################################
# Write metadata in each band (date + band name)
# ------------------------------------------------------
###########################################################
# This is useful to show date in band number in Qgis
S2.set_description_metadata(raster, dates)
###########################################################
# Produce NDVI time series from and to a raster
# ----------------------------------------------
S2.generate_raster(input_raster=raster,
output_raster='/tmp/index.tif',
expression=S2.get_index_expression('NDVI'),
dtype=np.float32)
##############################
# Plot NDVI index
from museotoolbox.processing import RasterMath
from matplotlib import pyplot as plt
rM = RasterMath('/tmp/index.tif')
NDVI = rM.get_random_block() #randomly select a block
from datetime import datetime
dateToDatetime = [datetime.strptime(str(date), '%Y%m%d') for date in dates]
plt.plot_date(dateToDatetime, NDVI[:10, :].T, '-o')
plt.ylabel('NDVI')
from museotoolbox import datasets from matplotlib import pyplot as plt ############################################################################## # Load HistoricalMap dataset # ------------------------------------------- raster, vector = datasets.load_historical_data() ############################################################################## # Initialize rasterMath with raster # ------------------------------------ # Set return3d to True to have full block size (not one pixel per row) rM = RasterMath(raster, return_3d=True) print(rM.get_random_block().shape) ############################################################################## # Comparing different block size (%, fixed, full block) # ------------------------------------------------------- ####################### # You can define block by percentage of the whole width/height rM.custom_block_size(1 / 2, 1 / 2) print(rM.get_random_block().shape) ####################### # Or by fixed window
def load_pottoks(return_X_y=True, return_target=True, return_only_path=False):
"""
Load two images of pottoks.
Parameters
-----------
return_X_y : bool, optional (default=True)
If True, will return the array in 2d where there are labels (X) and the labels (y)
If False, will return only X in 3d (the array of the image)
return_target : bool, optional (default=True)
If True, will return two arrays in a list
If False, will return only the source (a brown pottok)
Examples
--------
>>> brown_pottok_arr, brown_pottok_label = load_pottoks(return_target=False)
>>> brown_pottok_arr.shape
(4610, 3)
>>> brown_pottok_label.shape
(4610,)
"""
brown_pottok_uri = os.path.join(
__pathFile,
'brownpottok')
black_pottok_uri = os.path.join(
__pathFile,
'blackpottok')
to_return = []
if return_only_path :
to_return.extend([brown_pottok_uri + '.tif', brown_pottok_uri + '.gpkg'])
if return_target :
to_return.extend([black_pottok_uri + '.tif', black_pottok_uri + '.gpkg'])
elif return_X_y:
Xs, ys = extract_ROI(brown_pottok_uri + '.tif',
brown_pottok_uri + '.gpkg', 'level')
to_return.extend([Xs, ys])
if return_target:
Xt, yt = extract_ROI(
black_pottok_uri + '.tif', black_pottok_uri + '.gpkg', 'level')
to_return.extend([Xt, yt])
elif return_X_y is False:
brown_pottok_arr = RasterMath(
brown_pottok_uri + '.tif',
return_3d=True,
verbose=False).get_image_as_array()
to_return.append(brown_pottok_arr)
if return_target:
black_pottok_arr = RasterMath(
black_pottok_uri + '.tif',
return_3d=True,
verbose=False).get_image_as_array()
to_return.append(black_pottok_arr)
return to_return