def landtrend(year_start, year_end, geojson, lang, gc_client, metadata):
res = landtrend_get_data(year_start, year_end, geojson)
PLOT_LOCK.acquire()
plt = landtrend_make_plot(res, year_start, year_end)
plt.tight_layout()
f = tempfile.NamedTemporaryFile(suffix='.png').name
plt.savefig(f, bbox_inches='tight')
PLOT_LOCK.release()
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
title = metadata['landtrend_plot']['title'][lang] + ' ({} - {})'.format(year_start, year_end)
out = ImageryPNG(name='landtrend_plot',
lang=lang,
title=title,
date=[dt.date(year_start, 1, 1), dt.date(year_end, 12, 31)],
about=metadata['landtrend_plot']['about'][lang],
url=url)
schema = ImageryPNGSchema()
return {'landtrend_plot' : schema.dump(out)}
def greenness_trend(year_start, year_end, geojson, lang, gc_client, metadata):
start_date = dt.datetime(year_start, 1, 1)
end_date = dt.datetime(year_end, 12, 31)
point = ee.Geometry(geojson)
region = point.buffer(BOX_SIDE / 2).bounds()
ndvi = []
for y in range(year_start, year_end + 1):
ndvi.append(OLI_SR_COLL.filterDate('{}-01-1'.format(y), '{}-12-31'.format(y)) \
.map(maskL8sr) \
.map(calculate_ndvi) \
.mean() \
.addBands(ee.Image(y).float()) \
.rename(['ndvi','year']))
ndvi_coll = ee.ImageCollection(ndvi)
# Compute linear trend function to predict ndvi based on year (ndvi trend)
lf_trend = ndvi_coll.select(['year', 'ndvi']).reduce(ee.Reducer.linearFit())
ndvi_trnd = (lf_trend.select('scale').divide(ndvi[0].select("ndvi"))).multiply(100)
# Define visualization parameter for ndvi trend and apply them
p_ndvi_trnd = {'min': -10, 'max': 10, 'palette':['#9b2779','#ffffe0','#006500']}
map_trnd = ndvi_trnd.visualize(**p_ndvi_trnd)
map_trnd_mosaic = ee.ImageCollection.fromImages([map_trnd, ee.Image() \
.int() \
.paint(point.buffer(BOX_SIDE / 75), 1) \
.visualize(**{'palette': ['black'], 'opacity': 1})]) \
.mosaic()
# Reproject ndvi mean image so it can retrieve data from every latitude
# without deforming the aoi bounding box
map_trnd_mosaic = map_trnd_mosaic.reproject(scale=30, crs='EPSG:3857')
# create thumbnail and retrieve it by url
trnd_url = map_trnd_mosaic.getThumbUrl({'region': region.getInfo(), 'dimensions': 256})
trnd_name = 'ndvi_trnd.png'
request.urlretrieve(trnd_url, trnd_name)
# read image and pass it to a 2-d numpy array
trnd_frame = Image.open(trnd_name)
ndvi_arr_trnd = np.array(trnd_frame)
legend = Image.open(os.path.join(pathlib.Path(__file__).parent.absolute(),
'ndvi_trd_{}.png'.format(lang.lower())))
title = metadata['greenness_trend']['title'][lang] + ' ({} - {})'.format(year_start, year_end)
f = plot_image_to_file(ndvi_arr_trnd, title, legend)
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
about = metadata['greenness_trend']['about'][lang].format(YEAR_START=start_date.strftime('%Y/%m/%d'),
YEAR_END=end_date.strftime('%Y/%m/%d'))
out = ImageryPNG(name='greenness_trend',
lang=lang,
title=title,
date=[start_date, end_date],
about=about,
url=url)
schema = ImageryPNGSchema()
return {'greenness_trend' : schema.dump(out)}
def base_image(year, geojson, lang, gc_client, metadata):
start_date = dt.datetime(year, 1, 1)
end_date = dt.datetime(year, 12, 31)
point = ee.Geometry(geojson)
region = point.buffer(BOX_SIDE / 2).bounds()
# Mask out clouds and cloud-shadows in the Landsat image
range_coll = OLI_SR_COLL.filterDate(start_date.strftime('%Y-%m-%d'),
end_date.strftime('%Y-%m-%d'))
l8sr_y = (range_coll.map(maskL8sr).median().setDefaultProjection(
range_coll.first().projection()))
# Define visualization parameter for ndvi trend and apply them
p_l8sr = {
'bands': ['B4', 'B3', 'B2'],
'min': 0,
'max': 3000,
'gamma': 1.5,
}
map_l8sr = l8sr_y.visualize(**p_l8sr)
map_l8sr_mosaic = ee.ImageCollection.fromImages([map_l8sr, ee.Image().int() \
.paint(point.buffer(BOX_SIDE / 75), 1) \
.visualize(**{'palette': ['black'], 'opacity': 1})]) \
.mosaic()
# Reproject L8 image so it can retrieve data from every latitute
# without deforming the aoi bounding box
map_l8sr_mosaic = map_l8sr_mosaic.reproject(scale=30, crs='EPSG:3857')
# create thumbnail and retrieve it by url
l8sr_url = map_l8sr_mosaic.getThumbUrl({
'region': region.getInfo(),
'dimensions': 256
})
l8sr_name = 'l8sr.png'
request.urlretrieve(l8sr_url, l8sr_name)
# read image and pass it to a 2-d numpy array
l8sr_frame = Image.open(l8sr_name)
np_l8sr = np.array(l8sr_frame)
title = metadata['base_image']['title'][lang] + ' ({})'.format(year)
f = plot_image_to_file(np_l8sr, title)
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
out = ImageryPNG(name='base_image',
lang=lang,
title=title,
date=[start_date, end_date],
about=metadata['base_image']['about'][lang],
url=url)
schema = ImageryPNGSchema()
return {'base_image': schema.dump(out)}
def greenness(year, geojson, lang, gc_client, metadata):
start_date = dt.datetime(year, 1, 1)
end_date = dt.datetime(year, 12, 31)
point = ee.Geometry(geojson)
region = point.buffer(BOX_SIDE / 2).bounds()
ndvi_mean = OLI_SR_COLL.filterDate(start_date.strftime('%Y-%m-%d'), end_date.strftime('%Y-%m-%d')) \
.map(maskL8sr) \
.map(calculate_ndvi) \
.mean() \
.addBands(ee.Image(year).float()) \
.rename(['ndvi','year'])
# Define visualization parameter for ndvi trend and apply them
p_ndvi_mean = {'bands':'ndvi', 'min': 0.3, 'max': 0.9, 'palette':['#ffffcc','#006600']}
map_mean = ndvi_mean.visualize(**p_ndvi_mean)
map_mean_mosaic = ee.ImageCollection.fromImages([map_mean, ee.Image() \
.int() \
.paint(point.buffer(BOX_SIDE / 75), 1) \
.visualize(**{'palette': ['black'], 'opacity': 1})]) \
.mosaic()
# Reproject ndvi mean image so it can retrieve data from every latitute
# without deforming the aoi bounding box
map_mean_mosaic = map_mean_mosaic.reproject(scale=30, crs='EPSG:3857')
# create thumbnail and retrieve it by url
mean_url = map_mean_mosaic.getThumbUrl({'region': region.getInfo(), 'dimensions': 256})
mean_name = 'ndvi_mean.png'
request.urlretrieve(mean_url, mean_name)
# read image and pass it to a 2-d numpy array
mean_frame = Image.open(mean_name)
np_mean = np.array(mean_frame)
legend = Image.open(os.path.join(pathlib.Path(__file__).parent.absolute(),
'ndvi_avg_{}.png'.format(lang.lower())))
title = metadata['greenness']['title'][lang] + ' ({})'.format(start_date.year)
f = plot_image_to_file(np_mean, title, legend)
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
about = metadata['greenness']['about'][lang].format(YEAR_START=start_date.strftime('%Y/%m/%d'),
YEAR_END=end_date.strftime('%Y/%m/%d'))
out = ImageryPNG(name='greenness',
lang=lang,
title=title,
date=[start_date, end_date],
about=about,
url=url)
schema = ImageryPNGSchema()
return {'greenness' : schema.dump(out)}
def get_urls(self):
resp = requests.get('https://www.googleapis.com/storage/v1/b/{bucket}/o?prefix={prefix}'.format(bucket=BUCKET, prefix=self.prefix))
if not resp or resp.status_code != 200:
raise GEETaskFailure('Failed to list urls for results from {}'.format(self.task))
items = resp.json()['items']
if len(items) < 1:
raise GEETaskFailure('No urls were found for {}'.format(self.task))
else:
urls = []
for item in items:
urls.append(Url(item['mediaLink'], item['md5Hash']))
return urls
def greenness_trend(year_start, year_end, geojson, lang, gc_client):
start_date = dt.datetime(year_start, 1, 1)
end_date = dt.datetime(year_end, 12, 31)
region = ee.Geometry(geojson).buffer(BOX_SIDE / 2).bounds()
ndvi = []
for y in range(year_start, year_end + 1):
ndvi.append(OLI_SR_COLL \
.filterDate(start_date.strftime('%Y-%m-%d'), end_date.strftime('%Y-%m-%d')) \
.map(maskL8sr) \
.map(calculate_ndvi) \
.mean() \
.addBands(ee.Image(y).float()) \
.rename(['ndvi','year']))
ndvi_coll = ee.ImageCollection(ndvi)
# Compute linear trend function to predict ndvi based on year (ndvi trend)
lf_trend = ndvi_coll.select(['year',
'ndvi']).reduce(ee.Reducer.linearFit())
ndvi_trnd = (lf_trend.select('scale').divide(
ndvi[0].select("ndvi"))).multiply(100)
# Reproject ndvi_mean and ndvi_trnd images so they can be correctly passed to 'sampleRectangle()' function
ndvi_trnd_reproject = ndvi_trnd.reproject(scale=30, crs='EPSG:3857')
ndvi_trnd_band = ndvi_trnd_reproject.select('scale').sampleRectangle(
region)
# Get individual band arrays.
ndvi_arr_trnd = ndvi_trnd_band.get('scale')
# Transfer the arrays from server to client and cast as np array.
ndvi_arr_trnd = np.array(ndvi_arr_trnd.getInfo()).astype('float64')
legend = Image.open(
os.path.join(
pathlib.Path(__file__).parent.absolute(),
'ndvi_trd_{}.png'.format(lang.lower())))
f = plot_image_to_file(ndvi_arr_trnd, 'Greenness Trend', 'PiYG', legend)
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
out = ImageryPNG(
name='greenness_trend',
lang=lang,
title='This is a title',
date=[start_date, end_date],
about=
"This is some about text, with a link in it to the <a href='http://trends.earth'>Trends.Earth</a> web page.",
url=url)
schema = ImageryPNGSchema()
return {'greenness_trend': schema.dump(out)}
def base_image(year, geojson, lang, gc_client):
start_date = dt.datetime(year, 1, 1)
end_date = dt.datetime(year, 12, 31)
region = ee.Geometry(geojson).buffer(BOX_SIDE / 2).bounds()
# Mask out clouds and cloud-shadows in the Landsat image
range_coll = OLI_SR_COLL.filterDate(start_date.strftime('%Y-%m-%d'),
end_date.strftime('%Y-%m-%d'))
l8sr_y = (range_coll.map(maskL8sr).median().setDefaultProjection(
range_coll.first().projection()))
# Reproject L8 image so it can be correctly passed to 'sampleRectangle()'
# function
l8sr_mosaic_bands = l8sr_y.reproject(scale=30, crs='EPSG:3857') \
.select(['B2', 'B3', 'B4']) \
.sampleRectangle(region)
with ThreadPoolExecutor(max_workers=4) as executor:
res = []
for b in ['B4', 'B3', 'B2']:
res.append(executor.submit(get_band, l8sr_mosaic_bands, b))
# Gather results and concatenate them
out = {}
for this_res in as_completed(res):
out.update(this_res.result())
rgb_img = np.concatenate((out['B4'], out['B3'], out['B2']), 2)
# get the image max value
img_max = np.max(rgb_img)
# Scale the data to [0, 255] to show as an RGB image.
rgb_img_plot = (255 * ((rgb_img - 100) / img_max)).astype('uint8')
f = plot_image_to_file(rgb_img_plot, "Satellite Image")
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
out = ImageryPNG(
name='base_image',
lang=lang,
title='This is a title',
date=[start_date, end_date],
about=
"This is some about text, with a link in it to the <a href='http://trends.earth'>Trends.Earth</a> web page.",
url=url)
schema = ImageryPNGSchema()
return {'base_image': schema.dump(out)}
def landtrend(year_start, year_end, geojson, lang, gc_client):
res = landtrend_get_data(year_start, year_end, geojson)
plt = landtrend_make_plot(res, year_start, year_end)
plt.tight_layout()
f = tempfile.NamedTemporaryFile(suffix='.png').name
plt.savefig(f, bbox_inches='tight')
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
#TODO: Need to return date as a period
out = ImageryPNG(
name='landtrend_plot',
lang=lang,
title='This is a title',
date=[dt.date(year_start, 1, 1),
dt.date(year_end, 12, 31)],
about=
"This is some about text, with a link in it to the <a href='http://trends.earth'>Trends.Earth</a> web page.",
url=url)
schema = ImageryPNGSchema()
return {'landtrend_plot': schema.dump(out)}
def greenness(year, geojson, lang, gc_client):
start_date = dt.datetime(year, 1, 1)
end_date = dt.datetime(year, 12, 31)
region = ee.Geometry(geojson).buffer(BOX_SIDE / 2).bounds()
ndvi_mean = OLI_SR_COLL.filterDate(start_date.strftime('%Y-%m-%d'), end_date.strftime('%Y-%m-%d')) \
.map(maskL8sr) \
.map(calculate_ndvi) \
.mean() \
.addBands(ee.Image(year).float()) \
.rename(['ndvi','year'])
# Reproject ndvi_mean and ndvi_trnd images so they can be correctly passed
# to 'sampleRectangle()' function
ndvi_mean_reproject = ndvi_mean.reproject(scale=30, crs='EPSG:3857')
ndvi_mean_band = ndvi_mean_reproject.select('ndvi').sampleRectangle(region)
ndvi_arr_mean = np.array(ndvi_mean_band.get('ndvi').getInfo())
legend = Image.open(
os.path.join(
pathlib.Path(__file__).parent.absolute(),
'ndvi_avg_{}.png'.format(lang.lower())))
f = plot_image_to_file(ndvi_arr_mean, 'Average Greenness', 'Greens',
legend)
h = get_hash(f)
url = Url(upload_to_google_cloud(gc_client, f), h)
out = ImageryPNG(
name='greenness',
lang=lang,
title='This is a title',
date=[start_date, end_date],
about=
"This is some about text, with a link in it to the <a href='http://trends.earth'>Trends.Earth</a> web page.",
url=url)
schema = ImageryPNGSchema()
return {'greenness': schema.dump(out)}