def get_pixelsize_meters(self):
'''Returns the pixelsize (deltaX, deltaY) of the domain
For projected domains, the exact result which is constant
over the domain is returned.
For geographic (lon-lat) projections, or domains with no geotransform,
the haversine formula is used to calculate the pixel size
in the center of the domain.
Returns
--------
deltaX, deltaY : float
pixel size in X and Y directions given in meters
'''
srs = osr.SpatialReference(self.vrt.dataset.GetProjection())
if srs.IsProjected:
if srs.GetAttrValue('unit') == 'metre':
geoTransform = self.vrt.dataset.GetGeoTransform()
deltaX = abs(geoTransform[1])
deltaY = abs(geoTransform[5])
return deltaX, deltaY
# Estimate pixel size in center of domain using haversine formula
centerCol = round(self.vrt.dataset.RasterXSize/2)
centerRow = round(self.vrt.dataset.RasterYSize/2)
lon00, lat00 = self.transform_points([centerCol], [centerRow])
lon01, lat01 = self.transform_points([centerCol], [centerRow + 1])
lon10, lat10 = self.transform_points([centerCol + 1], [centerRow])
deltaX = haversine(lon00, lat00, lon01, lat01)
deltaY = haversine(lon00, lat00, lon10, lat10)
return deltaX[0], deltaY[0]
def get_pixelsize_meters(self):
'''Returns the pixelsize (deltaX, deltaY) of the domain
For projected domains, the exact result which is constant
over the domain is returned.
For geographic (lon-lat) projections, or domains with no geotransform,
the haversine formula is used to calculate the pixel size
in the center of the domain.
Returns
--------
deltaX, deltaY : float
pixel size in X and Y directions given in meters
'''
srs = osr.SpatialReference(self.vrt.dataset.GetProjection())
if srs.IsProjected:
if srs.GetAttrValue('unit') == 'metre':
geoTransform = self.vrt.dataset.GetGeoTransform()
deltaX = abs(geoTransform[1])
deltaY = abs(geoTransform[5])
return deltaX, deltaY
# Estimate pixel size in center of domain using haversine formula
centerCol = round(self.vrt.dataset.RasterXSize / 2)
centerRow = round(self.vrt.dataset.RasterYSize / 2)
lon00, lat00 = self.transform_points([centerCol], [centerRow])
lon01, lat01 = self.transform_points([centerCol], [centerRow + 1])
lon10, lat10 = self.transform_points([centerCol + 1], [centerRow])
deltaX = haversine(lon00, lat00, lon01, lat01)
deltaY = haversine(lon00, lat00, lon10, lat10)
return deltaX[0], deltaY[0]
def process(self, uri, force=False, *args, **kwargs):
fn = 'WIND_'+os.path.basename(uri)
if DatasetURI.objects.filter(uri__contains=fn) and not force:
wds = Dataset.objects.filter(dataseturi__uri__contains=fn)[0]
return wds, False
try:
w = wind_from_sar_and_arome_forecast(uri)
except (TooHighResolutionError, PolarizationError, ObjectDoesNotExist) as e:
if type(e)==Dataset.DoesNotExist:
warnings.warn(uri+': '+e.args[0])
else:
# ObjectDoesNotExist could happen if there is no overlap between SAR and model
warnings.warn(e.file + ': ' + e.msg)
return None, False
metadata = w.get_metadata()
# Direct reprojection fails - gdal can't read the bands if we do w.reproject...
# Workaround: Export wind to temporary file
#tmp_filename = os.path.join(settings.PRODUCTS_ROOT,'TMP_WIND_'+os.path.basename(uri))
fd, tmp_filename = tempfile.mkstemp(suffix='.nc')
os.close(fd) # Just in case - see https://www.logilab.org/blogentry/17873
w.export(tmp_filename)
# Read temporary file
ww = Nansat(tmp_filename)
# Reproject
lon, lat = ww.get_geolocation_grids()
#lon, lat = w.get_geolocation_grids()
srs = '+proj=stere +datum=WGS84 +ellps=WGS84 +lat_0=%.2f +lon_0=%.2f +no_defs'%(np.mean(lat),np.mean(lon))
xmin, xmax, ymin, ymax = -haversine(np.mean(lon),np.mean(lat),np.min(lon),np.mean(lat)), \
haversine(np.mean(lon),np.mean(lat),np.max(lon),np.mean(lat)), \
-haversine(np.mean(lon),np.mean(lat),np.mean(lon),np.min(lat)), \
haversine(np.mean(lon),np.mean(lat),np.mean(lon),np.max(lat))
ext = '-te %d %d %d %d -tr 500 500' %(xmin-10000, ymin-10000, xmax+10000, ymax+10000)
d = Domain(srs, ext)
ww.reproject(d, tps=True)
#w.reproject(d, tps=True)
# Set global metadata
metadata['data_center'] = json.dumps(pti.get_gcmd_provider(kwargs.pop('data_center', '')))
metadata['naming_authority'] = kwargs.pop('naming_authority', '')
metadata['project'] = 'SIOS InfraNor'
metadata['entry_title'] = 'Wind field from '+os.path.basename(uri)
metadata.pop('file_creation_date')
metadata['history'] = metadata['history'] + ' ' + timezone.now().isoformat() + \
'. Calculated wind field from NRCS and Arome Arctic forecast wind directions.'
metadata.pop('institution')
metadata['keywords'] += ', ['
for key, value in pti.get_gcmd_science_keyword('U/V WIND COMPONENTS').items():
if value:
metadata['keywords'] += value + ', '
metadata['keywords'] += ']'
metadata.pop('LINE_SPACING')
metadata.pop('PIXEL_SPACING')
metadata['summary'] = 'Near surface (10m) wind from Arome Arctic forecast wind and ' + metadata['summary']
metadata['title'] = 'Near surface wind from '+metadata['title']
# Get or create data folder
start_time = parse(metadata['time_coverage_start'])
yfolder = os.path.join(settings.PRODUCTS_ROOT, '{:04d}'.format(start_time.year))
mfolder = os.path.join(yfolder, '{:02d}'.format(start_time.month))
dfolder = os.path.join(mfolder, '{:02d}'.format(start_time.day))
if not os.path.isdir(yfolder):
os.mkdir(yfolder)
if not os.path.isdir(mfolder):
os.mkdir(mfolder)
if not os.path.isdir(dfolder):
os.mkdir(dfolder)
# Export
thredds_fn = os.path.join(dfolder, fn)
wind_uri = 'file://localhost' + thredds_fn
ww.export2thredds(thredds_fn, mask_name='swathmask', metadata=metadata, no_mask_value=1)
wds, cr = super(WindManager, self).get_or_create(wind_uri)
os.unlink(tmp_filename)
return wds, cr