def search_sentinels(platform_name, df, aoi, dt=2, user=None, pwd=None,
proj_string='+init=EPSG:3995', product_type=None,
min_cloud_cover=0, max_cloud_cover=100,
swath_type=None, f_out=None):
"""
Search Sentinel-1/2 images overlapping ICESat-2 data within +- dt
Parameters:
-----------
platform_name : str ['Sentinel-1 | Sentinel-2']
name of the platform for which images will be searched
df : panda dataframe
ICESat-2 data
aoi: str, list
area of interest as WKT string or bounding box[lllon, lllat, urlon, urlat]
dt: int, float
difference in hours between CS2 and S2
user : str
username to connect to the Copernicus Scientific Hub
pwd : str
password to connect to the Copernicus Scientific Hub
proj_string: str
projection string to be used with the pyproj module
product_type : str
name of the type of product to be searched (more info at https://scihub.copernicus.eu/userguide/)
swath_type : str
name of the type of swath to be searched (Sentinel-1 only, more info at https://scihub.copernicus.eu/userguide/)
min_cloud_cover: int, float
Minimum cloud coverage in percentage (Sentinel-2 only)
max_cloud_cover: int, float
Maximum cloud coverage in percentage (Sentinel-2 only)
f_out : str
path to file where to write results
Returns: (to be finished!)
--------
"""
#==========================================================================
# Pre-processing
#==========================================================================
### Imports
from sentinelsat import SentinelAPI
# import wkt
import pyproj
import numpy as np
import shapely.geometry as sg
from shapely.wkt import dumps, loads
from astropy.time import Time, TimeDelta
from tqdm import tqdm
### Convert aoi to shapely polygon in projected CRS
# define projection
print("Creating AOI polygon...")
proj = pyproj.Proj(proj_string)
# read aoi polygon
if type(aoi) == str:
aoi_temp = loads(aoi)
elif type(aoi) in (list, tuple):
aoi_temp = sg.box(aoi[0], aoi[1], aoi[2], aoi[3])
aoi = aoi_temp.wkt
else:
print("ERROR: 'aoi' should be provided as a WKT string or bounding box (list)")
sys.exit(1)
### Check input parameters
if product_type == None:
if platform_name == 'Sentinel-1':
product_type = 'GRD'
print("product_type set to: ", product_type)
if platform_name == 'Sentinel-2':
product_type = 'S2MSI1C'
print("product_type set to: ", product_type)
if swath_type == None and platform_name == 'Sentinel-1':
swath_type = 'EW'
print("swath_type set to: ", swath_type)
# project coordinates and convert to shapely polygon
x, y = proj(aoi_temp.exterior.xy[0], aoi_temp.exterior.xy[1])
aoi_poly = sg.Polygon(list(zip(x, y)))
### Convert dt to astropy time object
dtt = TimeDelta(3600 * dt, format='sec')
#==========================================================================
# Processing
#==========================================================================
### Project IS2 data to desired CRS
print("Selecting orbit data inside AOI...")
lon, lat = np.array(df['lons']), np.array(df['lats'])
x, y = proj(lon, lat)
### Extract IS2 orbit number
is2_orbits = np.unique(df['orbit_number'])
print("N. of orbits/points inside AOI: {}/{}".format(len(is2_orbits),
len(df)))
### Extract time period from IS2 data to query the server
t_is2 = Time(df['time'], scale='utc')
t_is2_start = min(t_is2) - dtt
t_is2_stop = max(t_is2) + dtt
### Read metadata
print("Query for metadata...")
api = SentinelAPI(user, pwd,'https://scihub.copernicus.eu/dhus',
timeout=600)
if platform_name == 'Sentinel-1':
md = api.query(area=aoi, date=(t_is2_start.datetime, t_is2_stop.datetime),
platformname='Sentinel-1', area_relation='Intersects',
producttype=product_type, sensoroperationalmode=swath_type)
elif platform_name == 'Sentinel-2':
md = api.query(area=aoi, date=(t_is2_start.datetime, t_is2_stop.datetime),
platformname='Sentinel-2', area_relation='Intersects',
cloudcoverpercentage=(min_cloud_cover, max_cloud_cover),
producttype=product_type)
print("N. of total images: {}".format(len(md)))
if len(md) == 0:
return [], [], [], [], [], []
### Convert Sentinel-2 time strings to astropy time objects
t_sen = {}
print("Converting time to astropy objects...")
for el in md:
t_sen[el] = Time(md[el]['beginposition'], format='datetime',
scale='utc')
### Loop over orbits to find images that satisfy time costraints
TimeDict = {}
t_is2 = []
print("Looping over orbits to find intersections within {}h...".format(dt))
for c, o in tqdm(enumerate(is2_orbits)):
### select CS2 data
d_is2 = df[df['orbit_number'] == o]
### compute CS2 track central time
t_temp = Time(d_is2['time'], scale='utc')
t_start_is2 = min(t_temp)
t_stop_is2 = max(t_temp)
t_is2_o = t_start_is2 + (t_stop_is2 - t_start_is2) / 2
t_is2.append(t_is2_o)
### save dict keys of images within +-dt from CS2
i_t = np.array(
[el for el in md if np.abs((t_sen[el] - t_is2_o).sec) <= dtt.sec])
TimeDict[o] = i_t
# get unique images within +-dt from all orbit data
i_sen_t_int = set(np.concatenate(list(TimeDict.values())).ravel())
print("N. of images within {}h: {}".format(dt, len(i_sen_t_int)))
if len(i_sen_t_int) == 0:
return [], [], [], [], [], []
### Project images corner coordinates and convert to shapely polygons
print("Creating images footprint polygons...")
# loop over them, project corner coords and create polygons
SenPolygonsDict = {}
for i in i_sen_t_int:
# load S2 footprint
aoi_sen = loads(md[i]['footprint'])
# check if multipolygon has more than 1 polygon defined
if len(aoi_sen) > 1:
print("WARNING: footprint for product {}".format(i),
"is defined by more than 1 polygon!!!")
aoi_sen = aoi_sen[0]
# project corner coords
x_sen, y_sen = proj(aoi_sen.exterior.xy[0], aoi_sen.exterior.xy[1])
# add polygon to dictionary
SenPolygonsDict[i] = sg.Polygon(list(zip(x_sen, y_sen)))
### Loop over orbits to find spatial intersections
print("Looping over orbits to find intersections...")
orbit_number = []
product_name = []
browse_url = []
download_url = []
t_diff = []
md_out = {}
for c, o in tqdm(enumerate(is2_orbits)):
### select CS2 data
i = df['orbit_number'] == o
# check if track has at least 2 points
if sum(i) < 2:
continue
d_is2 = df[i]
x_is2 = x[i]
y_is2 = y[i]
### create shapely line from CS track
is2_line = sg.LineString(list(zip(x_is2, y_is2)))
### collect LS8 polygon indices
i_sen = TimeDict[o]
### Loop over S2 polygons
for i_poly in i_sen:
ls_poly = SenPolygonsDict[i_poly]
if is2_line.intersects(ls_poly):
orbit_number.append(o)
t_diff.append((t_sen[i_poly] - t_is2[c]).sec / 3600)
product_name.append(md[i_poly]['filename'])
download_url.append(md[i_poly]['link'])
browse_url.append(md[i_poly]['link_icon'])
md_out[i_poly] = md[i_poly]
print("N. of total intersections: {}".format(len(orbit_number)))
### Print to file
if f_out != None:
print("Printing results to {}...".format(f_out))
with open(f_out, 'w') as fp:
fp.write("orbit_number,t_diff_(h),product_id,dowload_url,browse_url\n")
for i in range(len(orbit_number)):
fp.write("{},{:.2f},{},{},{}\n".format(
orbit_number[i], t_diff[i], product_name[i],
download_url[i], browse_url[i]))
return orbit_number, product_name, browse_url, download_url, t_diff, md_out
